From Data Silos to Health Records Without Borders: A Systematic Survey on Patient-Centered Data Interoperability

Abstract

The widespread use of electronic health records (EHRs) and healthcare information systems (HISs) has led to isolated data silos across healthcare providers, and current interoperability standards like FHIR cannot address some scenarios. For instance, it cannot retrieve patients’ health records if they are stored by multiple healthcare providers with diverse interoperability standards or the same standard but different implementation guides. FHIR and similar standards prioritize institutional interoperability rather than patient-centered interoperability. We explored the challenges in transforming fragmented data silos into patient-centered data interoperability. This research comprehensively reviewed 56 notable studies to analyze the challenges and approaches in patient-centered interoperability through qualitative and quantitative analyses. We classified the challenges into four domains and categorized common features of the propositions to patient-centered interoperability into six categories: EMR integration, EHR usage, FHIR adaptation, blockchain application, semantic interoperability, and personal data retrieval. Our results indicated that “using blockchain” (48%) and “personal data retrieval” (41%) emerged as the most cited features. The Jaccard similarity analysis revealed a strong synergy between blockchain and personal data retrieval (0.47) and recommends their integration as a robust approach to achieving patient-centered interoperability. Conversely, gaps exist between semantic interoperability and personal data retrieval (0.06) and between FHIR adaptation and personal data retrieval (0.08), depicting research opportunities to develop unique contributions for both combinations. Our data-driven insights provide a roadmap for future research and innovation.

1. Introduction

1.1. Motivation

Current healthcare information systems (HISs) face significant challenges with fragmented health data silos across multiple healthcare providers and countries. As presented in Table 1, The increasing adoption of EHR in healthcare information systems has created isolated data silos across healthcare providers and countries since 1990s. Current interoperability standards cannot address some scenarios despite advancements in data interoperability standards such as FHIR. These standards are not able to support data interoperability if patients’ medical records are distributed across multiple institutions with diverse interoperability standards or implementation guides. These standards often focus on institutional data exchange rather than patient-centered interoperability [1]. Patients cannot access their own medical records in a timely manner when they are stored in various healthcare systems and countries. The lack of a unified system for timely access to complete health records is a critical gap in current healthcare systems. Timely access to complete medical histories can mean the difference between life and death in some cases, especially in emergencies. It supports healthcare professionals in making more effective and accurate decisions [2]. Healthcare professionals need access to complete medical records to take more precise and urgent action regardless of business, bureaucratic, or technological barriers [1]. Regulatory organizations like HIPAA supervise patient privacy concerns, systematically creating interoperability limits between healthcare information systems through solid data governance. The lack of interoperability in healthcare data costs the US health system more than USD 30 billion annually [3]. This research aims to analyze the current approach to data interoperability by considering unsupported interoperability scenarios and summarizing the patient-centered data interoperability challenges.

1.2. Problem

Healthcare professionals need access to patient health records to make effective treatment decisions regardless of regulations or technological barriers. Medical records need to be provided to healthcare professionals in an understandable format within the needed timescale to make more precise decisions to save lives. Interoperability standards like FHIR have been developed to solve the interoperability problem between healthcare providers’ health information systems (HISs), but current interoperability standards require addressing scenarios where patients’ medical records are distributed across multiple institutions with diverse interoperability standards or different implementation guides [1]. Organizations commonly customize these standards based on their own requirements and use a variety of software, which is the main technical obstacle to connecting different EHRs [4]. Therefore, healthcare providers’ HISs create fragmented data silos for health records that can exchange data only with limited healthcare providers [5]. The lack of a unified access point slows down treatment decisions. It compromises safety, especially in emergencies for patients seeking care across multiple providers, particularly if they have records in multiple countries [6]. Interoperability standards like FHIR prioritize requirements of institutional data interoperability needs over patient-centered data interoperability. Regulation organizations and technological barriers exacerbate these challenges as well. It is a crucial need to comprehend the challenges of transforming fragmented health data silos into health records without borders. In particular, the research question is “What are the challenges for achieving patient-centered interoperability and the limitations of existing solutions for transforming data silos from institutional interoperability to patient-centered interoperability?”.

1.3. Objective

Our research aims to build an understanding of the current limitations and obstacles of patient-centered data interoperability in healthcare information systems; as Socrates said, “Understanding a question is half an answer.” We summarize all the propositions for this matter into the state of the art to build knowledge on the challenges of transforming health data silos into patient-centered data interoperability. We have conducted a literature review of notable related papers and analyzed 56 papers to identify and categorize the common features of their proposition as our second objective. It helps to understand the characteristics of a desired solution for patient-centered data interoperability in HISs. We designed the categories of features and mapped them to the reviewed propositions in Table 2. As the third objective, we aimed to provide data-driven insight through a correlation analysis of the features. We discuss the frequency of shared feature implications in the reviewed papers in Table 3. We synthesized the outcomes of the correlation analysis to create a comprehensive understanding of diverse aspects of the problem. This multifaceted analysis can be essential in developing an effective solution for patient-centered data interoperability.

2. Methodology

We used the following method to ensure that we developed a comprehensive literature review of patient-centered interoperability and data-driven insight from the extraction of selected papers.

1.

We defined the following keywords to search and select relevant research. The search keywords were formulated based on the research problem and objectives to identify the most relevant notable papers.

• “Healthcare interoperability standards”;
• “Healthcare data interoperability”;
• “Patient-centered interoperability”;
• “Health record interoperability”;
• “Health data sharing”;
• “Healthcare data exchange”;
• “Data interoperability”.

Also, we combined the above phrases with “FHIR”, “EMR”, “EHR”, and “PHR” to find more papers as well.

2.

We selected notable relevant papers from Google Scholar, IEEE, Elsevier, Springer, and MDPI with the above keywords and the following criteria by focusing on recent publications (since 2016):

a. 

Inclusion:

a.

Papers that point out health data interoperability challenges, frameworks, and solutions were included.

b.

Papers with significant citations or related contributions were prioritized in our literature review.

b. 

Exclusion:

a.

Papers unrelated to healthcare information systems were excluded.

b.

Papers with insufficient detail or unclear methodologies were also excluded.

3.

We conducted a review of the selected studies to identify notable points of each study to develop the state of the art. After the first round of review, we classified the extracted contents as follows:

a.

EHR/EMR/PHR systems;

b.

Blockchain-based systems;

c.

FHIR systems.

4.

We identified the challenges and classified them into four groups based on the summary of the literature review by characterizing patient-centered interoperability approaches to the solution.

5.

We analyzed the propositions and contributions of the reviewed papers to identify the most frequent features (common features). Then, we developed a map between the features of the indicated studies in Table 2.

6.

We calculated correlation indices such as Jaccard index and symmetric differences for common features based on the number of cross-referenced papers in Table 3.

7.

We analyzed the correlation indices to synthesize a qualitative approach and validated a qualitative analysis in the Results section.

8.

Based on the outcomes of our analyses, we developed data-driven insights and determined a domain for future research and innovation using a similarity analysis of the Jaccard indices.

9.

We sorted the feature combinations according to the similarity analysis to determine which combinations are stabilized enough for system development and which are great for developing research with a unique contribution.

3. State of the Art

This section reviews interoperability in existing solutions and frameworks in recent research to identify their limitations toward achieving patient-centered data interoperability in healthcare information systems. Some studies and frameworks propose interoperable health information systems using FHIR or blockchain models in their approaches. Although their approaches presented some notable improvements, they could only partially support patient-centered interoperability across diverse platforms. The interoperability standards facilitate structured data exchange between healthcare entities; they often need to be improved in terms of enabling patients to access their health records, especially when data are distributed across diverse systems with different standards and implementation guides. Furthermore, related research highlights challenges such as inconsistent standard implementations, data security and privacy constraints, and patient-centered interoperability. We reviewed recent, commonly used technologies and approaches in the following to provide a foundation for understanding the advantages of currently proposed interoperability solutions and to highlight the need for enhanced models that prioritize patient-centered interoperability.

3.1. Related Applied Technologies

The recently proposed blockchain-based healthcare system proposes an interesting vision for the level of interoperability and security. A distributed healthcare system for managing electronic medical records has various significant advantages in comparison to centralized healthcare systems. In a distributed system without a central authority, many threats such as data leakage by human mistakes or a single point of failure are no longer feasible. IPFS and blockchain are distributed systems that have a transparent process and clear logic via smart contracts. In recent years, some proposed healthcare systems have used these technologies [7].

Blockchain technology is a series of blocks arranged chronologically in a linked list data structure. The distributed ledger is immutable, which is significant in blockchain technology and makes it famous in distributed secure software architecture. Once an added block is committed, it cannot be altered to ensure data integrity within the ledger. The integrity of the blockchain is maintained because each block contains the previous block’s hash value. The replicated ledger across multiple nodes in the blockchain network makes it secure. The blockchain’s security is bolstered by various cryptographic techniques, such as Public Key Infrastructure (PKI) protocols, digital signatures, and hashing algorithms [8,9,10].

Blockchain provides a zero-trust environment to ensure that all parties involved in a transaction know the terms and outcomes, fostering collaboration between different blockchain systems via smart contracts. Smart contracts enhance interoperability and enable the automated execution of agreements between different permissioned blockchain managers to facilitate seamless transactions and interactions across various blockchain networks. Smart contracts help to ensure that all participating blockchains apply standardized protocols with clear rules and conditions. This standardization and these protocols allow various systems to work together effectively [10].

Advancements in IoT-enabled medical services present themselves as increasing trends in academic research projects and modern lifestyle management; however, it has caused new demands for creating connected environments and developing integrated patient health records for effective treatment in healthcare services [11]. These medical services aim for a variety of health services and provide stream data in large volumes and with good veracity. In this context, interoperability takes into account the increasing effectiveness of these services.

Privacy protection and the secure storage of medical data are crucial issues during medical services. Secure storage and making full use of personal medical records have always been a concern for the general population. Blockchain technology brings a new solution to this problem. As a hash chain with the characteristics of decentralization, verifiability, and immutability, blockchain technology can be used to securely store personal medical data [12]. There are several applications and different designs for interoperability to share patient data. Many blockchain-based healthcare systems have been presented in recent years and show that this technology provides a clear solution for security and privacy in health data storage; however, its very first steps are not entirely clear for every user and organization. Nevertheless, it has some indisputable potential to be part of a solution for the next generation of healthcare systems. Kotsiuba et al. [13] presented three categories for the medical data issues in the healthcare sector:

• Medical data are very extensive and cumbersome. In other words, health data have volume and variety.
• The quality level of medical data is a problem which complicates the analysis, diagnosis, and prognosis. Therefore, it requires veracity.
• Confidentiality is another major public health problem. In this regard, there is a need for additional protection of information in this area, especially in connection with the growing number of cybercrimes.

The most widespread approach among the existing platforms is to use a single core information model that all platforms must comply with. This means that a platform can only view the data from this core information model as custom extensions are not permitted. If a platform needs to reveal data from the core information model, the platform cannot expose these data and cannot inter-operate with other platforms [14].

Table 1. Timeline of related technologies [15,16,17,18,19,20,21,22].

Technology	Timeline	Comment
EMR	1970s	The digitization of medical records started in the 1960s and 1970s, but the term “EMR” only became commonly used in the 1980s.
HL7	1987	HL7 is a semantic data model designed for managing for all kinds of health records data in healthcare systems.
EHR	1990s	EHR helps to store all types of health records in a structured format within various devices and systems.
PHR	2000s	PHRs come with new emerging technologies such as the wearable sensors.
OPENEHR	2003	OPENEHR is an interoperability standard for patients’ data structures.
BLOCKCHAIN	2008	Blockchain is a system for immutable, distributed, and encrypted record transactions.
FHIR	2013	FHIR is an interoperability standard for the data exchange of patients’ data between health information systems by HL7.

Table 1. Timeline of related technologies [15,16,17,18,19,20,21,22].

Technology	Timeline	Comment
EMR	1970s	The digitization of medical records started in the 1960s and 1970s, but the term “EMR” only became commonly used in the 1980s.
HL7	1987	HL7 is a semantic data model designed for managing for all kinds of health records data in healthcare systems.
EHR	1990s	EHR helps to store all types of health records in a structured format within various devices and systems.
PHR	2000s	PHRs come with new emerging technologies such as the wearable sensors.
OPENEHR	2003	OPENEHR is an interoperability standard for patients’ data structures.
BLOCKCHAIN	2008	Blockchain is a system for immutable, distributed, and encrypted record transactions.
FHIR	2013	FHIR is an interoperability standard for the data exchange of patients’ data between health information systems by HL7.

Protected Health Information (PHI) is data that are sensitive to both healthcare providers and patients. PHI data can be maintained by multiple healthcare providers, thus resulting in separate data. Moreover, PHI data are stored in the provider’s database; hence, the patients have no authority to manage their own information [23].

3.2. Existing Interoperability Approaches

The previous section explained that technologies refer to the tools and systems used to exchange and make patient data available to healthcare professionals in healthcare information systems. Therefore, these technologies are commonly used by health information system designers to design health information systems. This section discusses the approaches for using these technologies to solve the problem of how individuals or organizations use technologies to accomplish their objectives. The long-standing focus of developing EHRs faces a critical need for innovation, like personalization and data science, prompting patients to engage in discussions about the details of their healthcare and restore agency over their medical data [24]. Although there has been a noticeable growth in the digitization of medical records, sharing electronic health data between hospitals and healthcare providers has several issues. The adoption of EHRs encounters problems for various reasons, including technical, operational, and privacy-related concerns. In the healthcare industry, interoperability is frequently centered on data transmission across commercial companies, such as numerous hospital systems, via a state-wide Health Information Exchange (HIE). A significant trend is the move toward patient-centered interoperability. However, new problems and requirements are raised around security and privacy, technology, incentives, and governance and must be addressed to achieve this form of data sharing successfully at various scales. Many of these challenges are still present for traditional interoperability. Therefore, searching for unique or unconventional solutions that might be useful in aiding the transition to patient-centered interoperability makes sense. Such interventions could reduce the tension between the numerous interoperability obstacles that characterize the environment of health data sharing and the clinical and academical research, and operational advantages [1].

Semantic data interoperability has attracted the attention of health system providers, in particular. Various studies have investigated methods for resolving semantic interoperability issues. However, adopting health standards and tools for acceptable data representation (ontologies, databases, and clinical models) still faces challenges in enabling healthcare workers to access health records efficiently. In this context, they are presenting how data integration and interchange across organizational boundaries can enhance the quality of care and work processes effectively. The administration of large-scale data ontology is one of the key challenges. Although many approaches have been proposed in relational database management systems (RDBMSs) and object-oriented database management systems (OODBMSs) to develop large-scale ontology management systems, they are limited by the fact that ontology data structures are fundamentally distinct from traditional data structures in RDBMSs and OODBMSs [25]. In addition, users have trouble utilizing ontology data because a huge number of terminologies (ontology nodes) in large-scale ontology data match a particular keyword string.

Due to the existence of hackers and malevolent users, big medical data in contemporary healthcare systems face several security challenges in several aspects, including decentralization, secrecy, security, privacy, etc. Blockchain technology greatly assists in healthcare information system problems such as single-point failure, centralized management of data resources, and privacy leaks that plague healthcare’s traditional cloud and client–server-based information storage architecture. In order to address these issues, Sharma et al. [26] investigated how blockchain technology might be used to assist the healthcare system. They demonstrated the interconnections between healthcare and other fields, such as telemedicine, big data, and business intelligence. They mentioned that blockchain expands capabilities in the elimination of redundancy, the reduction in errors, and the promotion of tailored patient care. Blockchain facilitates Mello controlling and monitoring of diseases, reduces costs, and boosts efficiency in public health programs.

Mello et al., after exploring the technological issues of healthcare data integration and data interoperability, outlined the benefits of implementing integrated data repositories as a clinical data warehouse. They mentioned that it facilitates clinical research, specialized analysis, and sophisticated data processing and proposed an international health record integration model. Other papers concentrated on a systematic review of implementing HL7 FHIR, which is a widely adopted health standard, and highlighted the robust trend of standard adoption. They mentioned that some authors were interested in ontology-based literature analyses, specifically fuzzy ontology, and they provided a detailed context for the top health standards and their various frameworks. They referred to the established standards as “e-Health Standards”, while we use the more general term health standard. The authors of a paper divided the trends of semantic interoperability into four sections that contribute to identifying the challenges and research opportunities [27]:

• Frameworks to solve semantic interoperability problems.
• Using ontologies to resolve interoperability issues.
• Standards in interoperable EHR.
• Barriers to and the heterogeneous problem of EHR semantic interoperability.

In addition, that study explained the healthcare industry’s interest in utilizing common interoperability standards for electronic medical records to overcome technological barriers across healthcare providers. They performed their study on challenges in adopting standards over the past few years and the technologies that will eventually comprise the ecosystem for a semantically interoperable EHR.

Patients’ interactions with their records reflect the nature of how these records should be managed. MedRec is one of the early proposed models for a distributed ledger-based medical record management system designed for patients based on their interactions. It employs the Ethereum blockchain and smart contracts to store EHRs. The health record is not saved on the blockchain but, rather, on the database of healthcare providers, which uses third party administers. Therefore, these documents remain vulnerable to assault or abuse [24].

Society has become more health-conscious with the increase in chronic diseases, and well-being has become a higher priority than healthcare. Patients have become “health consumers” looking for better health management. The early detection of diseases requires a health delivery system that can monitor health status. The early detection of physical and mental changes requires sensitive and frequent measurement of physiological and behavioral data [28]. Consequently, the healthcare domain produces large quantities of data from many different sources such as relational databases, standards, XML, ADL files, images, scans, tabular records, or any other source. These data have heterogeneous structures and semantics and could be more specific and precise in most cases [24].

3.2.1. Blockchain-Based Healthcare Systems

The recently proposed blockchain-based healthcare system proposes an interesting vision for the level of data integrity, interoperability, and security [6]. Blockchain technology is a fast-evolving sector that has become more popular with researchers, and innovative projects are presented every day. The number of research projects in the healthcare system domain that is conducted with blockchain technology has exhibited disruptive technology that has a high potential to be used in healthcare systems. The various types of innovations propose new advantages to other sectors which are similar but centralized. Dubovitskaya et al. used this technology to improve data accessibility between multiple healthcare providers and hospitals. Shared and immutable data storing in the blockchain technology make it a popular option for removing intermediaries, and it is a possible option to omit the centralized dependency [29].

Ghadi et al. [30] reviewed blockchain technologies’ role in enhancing data interoperability within healthcare systems through several core mechanisms. Blockchain’s decentralized framework facilitates decentralization and trustless communication among hospitals, healthcare providers, and patients by fostering transparency across the healthcare ecosystem. Blockchain technology supports interoperability by effectively establishing standardized data-sharing protocols for collaboration between diverse healthcare providers and EHR systems. Smart contracts automate and enforce data-sharing rules and authorized access by simplifying complex data exchange processes. All these advantages of blockchain technology create a robust platform for efficient and secure data interoperability. Moreover, blockchain can address long-standing challenges in healthcare data interoperability and significantly enhance collaborative patient care.

The complexity and cost of a modern healthcare system are high in many cases, and data flows are different based on each design, but applying blockchain technology could be more effective in improving health record management and costs [31]. Blockchain emerged to provide a distributed financial record exchange and storing infrastructure that has enabled data accessibility in a securely distributed manner. It can be a disruptive technology that increases the interoperability of healthcare providers in access to patient health records and increase the data integrity level [29].

Smart contracts can improve data interoperability in heterogeneous systems by automating processes, standardizing interactions, and fostering trust via diverse systems. The ability to operate independently is a key factor in the success of data interoperability. The peer-to-peer distributed ledger technology in blockchain provides an immutable and transparent understanding of all the transactions. Data are stored in the form of transactions in the chain ledger, and it is signed digitally into blocks in chronological order. Shared and immutable data storing in blockchain technology makes it a popular option for removing intermediaries, and it is a possible option to omit centralized dependency. The complexity and cost of a modern healthcare system are high in many cases, and the data flows are different based on each design, but applying blockchain technology could be more effective in improving health record management and costs [29].

As discussed in the above research, blockchain has a high potential to integrate health records from different sources as presented in Figure 1. Additionally, the above researchers have specified high security and privacy as features of their proposed systems. Therefore, blockchain has a high potential to apply security, integrity, and privacy to healthcare systems. Blockchain characteristics such as immutability, transparency, and decentralized distributed data storage present a range of applications in healthcare systems. There are some challenges in accessing the patient’s medical records. Healthcare providers and regulators have different processes to grant access to a single medical record. Most of the reviewed blockchain-based healthcare systems in this proposal are on a distributed ledger database on a peer-to-peer (P2P) network comprising chronologically ordered blocks. These are decentralized distributed systems without any dependency on third parties for regulation but themselves. Because of decentralization, a blockchain-based system has no limits, such as a single point of failure, and data are stored in a ledger that is accessible by all the nodes in the blockchain [32].

The high redundancy rate has made blockchain expensive for storing data in high volumes. There is a range of solutions for data storing in high volumes, but distributed systems are more popular for their excellent performance with high traffic in addition to high volumes. Instability, lack of auditing, and incentive mechanisms in Peer-to-Peer (P2P) distributed file systems raise a need for alternative technology. In this context, distributed file systems (DFSs) are welcomed as new technologies, such as Inter-Planetary File System (IPFS) and Swarm. Blockchain-based DFSs successfully provide a solution to cover the disadvantages of blockchain in data warehousing and using blockchain’s strengths such as scalability and privacy [33].

Hussein et al. [34] discussed an existing gap between PHRs and blockchain, and they stated it is feasible to resolve this gap by encrypting medical data and outsourcing it to InterPlanetary File System (IPFS) storage. They called it “smart contract-based attribute-based searchable encryption (SC-ABSE)”, and it has been made by combining ciphertext-policy attribute-based encryption (CP-ABE), searchable symmetric encryption (SSE), smart contracts, and IPFS storage. Pournaghi et al. raised keeping medical records in different data sources as a problem. Medical entities have added a controlling layer on medical records, which is an obstacle to accessing medical records fast enough, even for patients [35].

Currently, blockchain and the Internet of Things (IoT) are two rising fields within the Information Technology (IT) industry. These two growing areas are utilized in numerous industries, including supply chain, logistics, and the automotive sector. Due to the limited processing power and storage space of IoT devices, users’ medical information is typically stored with a centralized third party, such as a clinical repository or on the cloud. Consequently, users frequently lose control of their medical information, which can lead to security leakage and a single-point failure. Therefore, a solution is necessary to simplify the data-sharing process for enhancing security. The combination of blockchain technology and IoT can substantially impact the healthcare industry by improving its efficacy, security, and transparency [10]. The effective sharing of EHR can enhance the treatment process, accuracy of diagnosis, security, and confidentiality using the Identity-Based Encryption (IBE) algorithm. Sharma et al. developed a blockchain-based Internet of Things (IoT) architecture to improve healthcare data security. In this instance, the smart contract describes all the fundamental functions of the healthcare system, which is advantageous for all parties involved. Numerous experiments were conducted to determine the efficacy of the suggested method. The results indicate that the proposed strategy is superior to other well-known schemes [11].

Traditional healthcare systems store and process patient records using a centralized client–server architecture, and healthcare organizations hold data in silos due to technical and architectural limitations that cannot be easily shared with other institutions. The lack of an efficient and secure data-sharing system in hospitals results in monetary and resource losses when a patient visits multiple hospitals. Blockchain can provide a safe and trustworthy decentralized foundation to overcome difficulties in traditional healthcare architecture for securing EHR storage, sharing, and retrieval [36]. Jayabalan and Jeyanthi proposed a blockchain-based system connected with IPFS for EHR in healthcare administration. The proposed system will decentralize the maintenance of failsafe and tamper-proof healthcare ledgers by healthcare facilities. Hospitals and physicians serve as lightweight nodes, whereas patient nodes may be either full or lightweight. The patient-centric access paradigm enables patients to serve as digital stewards of their health data, granting on-demand access to doctors and hospitals and withdrawing it after a predetermined period. Data are encrypted using symmetric key encryption (AES-128) before being stored in IPFS. Asymmetric encryption (RSA4096) is used to generate digital envelopes for symmetric key transmission to authorized parties. Digital signatures (RSA-1024) ensure that transactions are legitimate and originated from authorized nodes. The encrypted data are hashed using the SHA-256 technique. Multiple levels of security incorporated in this approach ensure that adversaries cannot obtain IPFS data; even if they do, the data will be unintelligible due to encryption. The proposed system for off-chain storage of health data using IPFS prevents scaling concerns in blockchain architecture. In addition, blockchain integration with IPFS contributes to preserving privacy inside the healthcare system, thereby making it highly safe, scalable, and robust [36].

Multi-organization data sharing is becoming progressively widespread due to the interoperability of systems and the necessity for collaboration across enterprises. However, data security problems are due to a lack of trust between enterprises that may be located in jurisdictions with varied security and privacy legislation (zero-trust environment). One needs to implement a more robust yet efficient access control mechanism to support cross-organizational data access and exchange requests in such a zero-trust environment. Due to more extraordinary security expenses, contemporary access control solutions often focus on defending a single aim rather than many parties [37]. Gai et al., in their article, presented a blockchain-based access control system intended to transfer lightweight data between companies. In the present cyber threat scenario, it is realistic to presume that any device or system can be compromised, including those in highly protected and secret environments. One must, therefore, presume that any exchange of data or resources occurs in a zero-trust setting. Gai et al. aimed to design a blockchain-based RBAC access control mechanism to provide collaborative access control as a primary objective of their research. Their solution was supported by a multi-signature protocol and smart contract, allowing them to facilitate the joint management of co-governance resources based on a set of responsibilities from multiple organizations [37].

The Internet of Things (IoT) industry is undergoing a transformation as a result of the introduction and development of rapidly advancing technologies. Wearable sensors create stream data, and the IoT data streams are seen as valuable assets that are processed, controlled, and even traded on marketplaces due to their high value and massive amounts of data. Complex activities for IoT devices include querying and filtering IoT streams and granting access permissions that need to be automated. Massive amounts of collected data provide a number of issues in management and storage. Finding a secure and reliable method for data streams is a further factor that requires consideration [38]. Hasan et al. proposed IoT communication utilizing side chains and a consortium blockchain. The research refers to a side chain as a private blockchain for a collection of IoT devices. The smart contract must contain the addresses of all Internet of Things (IoT) devices in order to verify their legitimacy and fulfil their requests. A requester must additionally join the consortium’s blockchain in order to obtain the IPFS hash of the stored data (InterPlanetary File System). Their architecture was tested on Ethereum and Monax. They presented a blockchain-based approach to enable decentralized, dependable, transparent, traceable, auditable, secure, and trustworthy IoT streaming data management and access control. To point out the issue of large-scale data storage, they combined the Ethereum blockchain with off-chain storage, such as IPFS. Using this approach, users can access data stored off-chain and verify its integrity using immutable logs and on-chain provenance information. They secured security and privacy by employing a proxy re-encryption network to provide users access to encrypted IoT data chunks, which enables users to have numerous accesses to the data chunk files within a predetermined time frame using the same access request. Additionally, all encrypted data chunk files are written to the off-chain storage only once but can be viewed several times [38].

Healthcare professionals encounter the task of securely accessing, maintaining, integrating, and exchanging health records. The existing technologies employed by the healthcare sector fail to sufficiently meet these needs due to constraints associated with privacy, security, and comprehensive ecosystem compatibility. There is a necessity for more openness in health data and the implementation of an improved healthcare data management system capable of forecasting, preventing, and managing new infectious illnesses. Blockchain technology addresses healthcare data management difficulties and enables healthcare providers to automate medical record extraction and to facilitate data exchange and enhanced diagnostic accuracy. Attaran mentioned in their research the difficulties, possibilities, advantages, and disadvantages related to the implementation of blockchain technology in healthcare record administration [39].

Blockchain may have the potential to be the solution to these issues. This technology provides a secure, tamper-resistant platform for storing medical records and other healthcare-related data [40]. The opportunity of using blockchain on EHR systems may increase security and privacy for data interoperability and data integration, accountability, and accessibility, which have been discussed in this section.

3.2.2. Electronic Health Records

In the days before modern technology, the healthcare industry kept medical records on handwritten papers. The paper-based medical records had a data redundancy problem because there were many copies of the patient’s medical records at each institution the patient attended. Standard EMRs emerged as a part of HL7 after the implementation of the Clinical Document Architecture (CDA). However, the complexity of installing the CDA prevented many hospitals from implementing standard EMRs. However, in some countries, it is implemented countrywide. Taiwan’s countrywide exchange infrastructure for EMRs has been in operation for many years [41].

The adoption of the EHR system is changing the healthcare industry. The purpose of EHR systems was to address the issues with paper-based medical records and offer a productive system that would change how the healthcare industry operates. Hospitals worldwide have adopted EHR systems because of their advantages, particularly the increased security and cost-effectiveness. Since they give the healthcare industry much functionality, they are seen as an essential component. These features include the electronic storage of medical information, scheduling patient appointments, handling billing and accounts, and ordering lab tests. They are accessible in many EHR systems used in the healthcare industry. The main goal is defined as the accessibility of medical records across various platforms while maintaining security and calm. The idea behind using EHR systems in hospitals or other healthcare facilities was to increase the quality of care, but these systems had several issues [40].

EHRs were never designed to manage multi-institutional or lifetime medical records. Patients leave data scattered across various organizations as life events and take them away from one provider’s data silo and into another. Through the HIPAA Privacy Rule, providers can take up to 60 days to respond (not necessarily to comply) to a request for updating or removing a record. Beyond the time delay, record maintenance can prove quite challenging to initiate as patients are rarely encouraged and seldom enabled to review their full records [24].

The healthcare industry is interested in blockchain technology due to its security, privacy, secrecy, and decentralization. However, EHR systems are troubled by data security, data integrity, and data management issues. Improvements in security, user experience, and other aspects of the healthcare industry are the primary benefits of technological progress. EHR and EMR systems are needed to embrace new technologies and face challenges. However, they continue to face challenges regarding the security of medical records, user ownership of data, and data integrity, among others [42].

Aggregating electronic health records from various healthcare providers offers significant benefits for both healthcare professionals and patients during treatment. Historical health records help physicians make more informed decisions on treatments and medications. Data integration on various EHR systems helps to identify better treatment and ensure patient information is accurate and up to date for making more accurate decisions. This integration promotes interoperability and the breakdown of data silos and allows healthcare providers to access patient records regardless of where care is provided. Aggregated EHR data also help with research advancements and regulatory compliance, and upholding safety standards. Moreover, near real-time patient monitoring enables timely interventions and improves recovery outcomes. When patients have access to complete health information, they are empowered to participate actively in their healthcare decisions [43].

Clinical information that is not utilized has adverse effects on healthcare quality. During ordinary clinical procedures, physicians prioritize clinical information pertinent to their specialties but may be indifferent or less concerned with information, indicating disease risks beyond their specialties, resulting in delayed and missed diagnoses or incorrect management. Shang et al. presented an EHR-centric knowledge graph system to efficiently utilize information buried in EHRs that need to be utilized. Under the ontological structure of a knowledge graph, EHR data were converted into a semantic information model. The knowledge graph then generates an EHR data trajectory and identifies significant clinical findings within EHR data using reasoning based on semantic criteria. A graphical reasoning trail depicts the reasoning footage and explains the clinical meaning so that physicians can comprehend the overlooked information with more ease. The application revealed that the recommended approach to clinical information consumption is possible and successful [44]. In addition, it seems that it is beneficial as an explainable artificial intelligence that provides interpretable advice for expert physicians to comprehend the significance of underutilized data and make exhaustive decisions.

In the concept of health or medical records, there is a slight difference between EHR, EMR, and PHR which could be confusing. As presented in Figure 2, EMR is all the records a single healthcare provider provides, and PHR is the records patients provide. Therefore, if a patient has a personal copy of their records from a healthcare provider, the records would be PHR and EMR, respectively. On the other hand, all electronic health records are called EHR, a more general term.

EHR systems can store both structured and unstructured text data in a variety of formats. An extensive portion of medical histories are covered by EHR, which also contains more comprehensive patient information and potential risk factors. It also facilitates the delivery of routine treatment in hospitals and primary care clinics and maintains patient health information. Additionally, it permits the reuse of patient data for a variety of tasks, such as treating individual patients, conducting research on medical and healthcare services, and managing healthcare facilities. To manage healthcare and facilitate data transfer between healthcare companies, an EHR must be adopted. Despite having various systems, an EHR enables communication between doctors, nurses, laboratories, and hospitals. To achieve semantic interoperability, data sharing between health organizations and health agents must encourage proper interpretation with the same precision and meaning adopted by the sender. Thus, semantic interoperability can communicate data between systems while ensuring conceptual understanding at the domain level [27].

In the current context, patient health records are stored in various data sources, such as different hospitals and clinics, which are not connected nor available in a timely manner, even in the case of an emergency. In emergency care, access to medical records is an indisputable need to make efficient decisions as fast as possible. Privacy is an important matter in the storage and transmission of medical and health records. Records need to be protected from unauthorized access, but they should be available in a timely manner during emergencies so that healthcare specialists can make vital decisions on patients’ treatment quickly and effectively [45]. This matter is critical in saving patient lives by delivering EMRs and EHRs in a timely manner, which is mentioned as our main motivation for this research besides respecting patients’ privacy rights.

Dubovitskaya et al. have focused on high levels of patient mobility by making EMR access manageable through blockchain. They proposed a prototype to share EHR access through blockchain and encrypted EMRs for storage in the public cloud so patients can manage their health records. In their prototype, they implemented an independent pluggable module regarding the FHIR standard to facilitate the adoption of their system. They mentioned that they use HL7 in their prototype for better interoperability, too. Blockchain can improve the verification and integrity of health data and impact cost and data quality in healthcare systems. It eliminates the “middleman” for healthcare systems and removes the required multiple levels of authentication [29].

Blockchain technology has created an opportunity to transform the design of current EHR management systems by proposing new architecture, frameworks, and models based on blockchain. They demonstrated EHR management system transformation in the aspect of some quality features [6]. Even though the idea behind using EHR systems in hospitals or other healthcare environments was to improve the quality of care, these systems had several issues and fell below expectations. An investigation of the experiences of nursing staff with the EHR was conducted in Finland, and it was found that these systems had issues with reliability and user-friendliness, interoperability, information asymmetry, and data breaches [40].

Data interoperability is the method by which various information systems communicate with one another. The data must be transferable and useable for other systems’ applications. Health practices and patient care have been enhanced by the rapid expansion and acceptance of EHRs and standards to exchange EHRs. However, only the organization’s physicians and specialists have access to the data stored in an EHR, not the patient [46]. Health Information Exchange (HIE) is a crucial data-sharing feature of EHR systems. However, there is no widely accepted standard for EHR systems because they are deployed in various institutions with different terminologies and technical and functional capabilities. These differences create a considerable obstacle to health data interoperability [40]. The focus of interoperability in the healthcare industry has historically been on data exchange between healthcare providers, such as various hospital systems, and most of the current standards followed this policy in their development. Recently, there has been a push for patient-driven interoperability, which involves patient-mediated and patient-driven health data exchange. However, additional issues and requirements related to security and privacy, technology, incentives, and governance must be addressed to succeed at various scales for this kind of data sharing [1].

3.2.3. Fast Health Interoperability Resources (FHIRs)

HL7 FHIR has progressed through four releases since its original presentation in May 2012. It has expanded from a standard with 49 resources to its present 145 resources and is still growing [16]. FHIR and equivalent standards are domain-dependent by nature. There was a need to consider being domain-independent to be compatible with a variety of devices such as smartphones, fitness trackers, smartwatches, and any other future advances. New standards are needed due to the rise of ubiquitous computers and devices [47].

Grahame Grieve led a group of healthcare information systems developers in 2012 to create FHIR as a modern approach to exchanging health data. The exponential growth of health data and the development of smartphone apps were both taken into consideration. The team developed a draft standard that combined API and widely used World Wide Web technologies, including JSON, XML, HTTP, and OAuth, with HL7 v2 messages.

FHIR was developed with the intention of developing a set of default resources that satisfy many use cases. Single or a combination of resources are brought together in an Implementation Guide to handle a specific use case, such as a provider directory or patient-reported outcomes. This structure is ideally suited for development beyond FHIR’s essential features. Many different terminologies and codes that develop over time are used to represent healthcare records. As a result, it is critical that both the sender and the receiver can interpret the shared data with the same meaning, which is known as “semantic interoperability” [16].

Healthcare data in FHIR is classified into numerous categories, such as patients, test results, and insurance claims. Each category is represented by an exchangeable record structure called FHIR’s resource. It consists of the component data elements, data constraints, and data relationships that are defined in an FHIR resource for each of these categories.

FHIR was created with the goal of enabling interoperability through well-structured data representations and straightforward, effective exchange mechanisms. FHIR adopted the following principles [16]:

• Usability: Both technical professionals and non-technical people can understand FHIR resources. Non-technical persons can view these in a browser or text reader and understand their contents even if the specifics of the XML/JSON format are not understood.
• Reusability: In order to prevent complex and redundant resource collection, FHIR resources are created to address the general demands of the healthcare industry. Resources can be modified for certain use cases by way of extensions and other adaptations (the profiling process). FHIR resources also link to other resources, creating complicated structures.
• Performance: FHIR resources are built simpler in terms of construction than prior standards, which makes it better suited for data exchange and simpler for developers to understand and use.
• Fidelity: Values cannot be mixed with other values of different data types in FHIR resources. Along with predefined sets of business rules, they can also be validated by their syntax.
• Implement ability: FHIR’s main objective was to build a standard that would be widely adopted by various developer communities. FHIR resources are made to be simply understood and exchanged using accepted programming languages, industry standards, and data exchange technologies.

Interoperability is one of the main issues that IT professionals are worried about as healthcare becomes more digital. FHIR, particularly its resources, are designed to exchange healthcare data and enhance interoperability across various medical devices and applications. FHIR resources are commonly used to transfer healthcare data between various apps and healthcare information systems. Patient vital data, such as heart rate, blood pressure, pulse, BMI, etc., are typically collected periodically or constantly in most research scenarios. Healthcare researchers and medical professionals are eager to take advantage of smartphones’ rising popularity. Studies in 2015 and 2016 showed that 72% of the population in the US and 54% in emerging countries use smartphones. Smartphones have the capability to close the serious data exchange and communication gap between patients, practitioners, and healthcare information systems [47]. This gap between incompatible entities could be filled by the FHIR standard [47]. According to Pfiffner et al. in 2016, mobile applications have the ability to quickly gather, aggregate, and report individual’s (patient or research subject) health data using common APIs like Research Kit or ResearchStack15. However, these APIs typically share the data with proprietary apps rather than an organization’s healthcare application, creating isolated data silos that are a significant barrier for healthcare big data applications [48].

4. Blind Spots of the Health Data Interoperability Standards

4.1. A Real-World Failed Scenario

For instance, people’s blood sugar data can be stored both on their own devices or on the online systems of their device manufacturers, as well as in several laboratories, clinics, and hospitals that the person may request. In this approach, the information of a person, especially the amount of blood sugar, can be stored on different dates, using different measurement standards, and in different places. Therefore, in such an environment, commonly used standards such as FHIR can only facilitate data interoperability between systems that follow FHIR with the same implementation guide that is online and connected depending on local privacy and security policies. Otherwise, the interoperability process is either complicated and time-consuming or impossible. Therefore, none of the presented approaches can support patients in similar scenarios.

For example, Cyrus is a 40-year-old Canadian patient from Iran who was raised in Sweden, studied in Germany for 6 years, and then migrated to Canada 10 years ago. Cyrus has had asthma, a chronic disease, since childhood. He received several healthcare services throughout his life, including physician consultations, clinical examinations, and medicines from different countries. He suffered injuries in a car accident while he lived in Germany. This caused a surgeon to implant a plastic prosthesis in his chest and a platinum prosthesis in his leg. Now, he lies unconscious in an emergency room. Healthcare professionals know his name, but there is no way to have all his medical records in time to make an effective decision. In the current situation, there is no way to be aware of all his existing records.

FHIR is designed to provide all the records accessible through an interoperable environment if all the systems are implemented with the same set of guidelines. However, most countries have developed implementation guidelines using FHIR to allow all healthcare providers’ information systems to exchange medical records. Despite all these efforts, there is no way to exchange records between hospitals in different countries. Additionally, a patient cannot access their own records through a single point of contact such as an interface. This is even if all of them are stored in a single country, and the FHIR standard has been implemented in the same guidelines. If Cyrus wants to provide all his records to his physician, he needs to contact all providers and receive services from them. This means that, in some cases, it is only possible to gather some medical records quickly to make effective clinical decisions.

Patient-centered interoperability enables healthcare practitioners to access a patient’s whole medical history as well as information from other providers, allowing them to make more informed decisions about patient treatment. Patients who have access to their health information are more engaged in their own care and can make better health decisions, which can lead to improved healthcare outcomes [49]. Patient satisfaction and more effective treatment plan adherence could be other outcomes of this approach. Furthermore, when healthcare practitioners can access patient records quickly, care coordination improves and medical errors are reduced. This can result in better patient outcomes and care quality.

4.2. Challenges of Data Interoperability in Current Healthcare Information Systems

The current existing interoperability approaches in healthcare information systems have been reviewed in the State of the Art section. The operating healthcare information systems can be classified into two categories: digital and paper-based according to Figure 3. In the digital category, systems can be subclassified into two aspects: using common interoperability standards and being online. Please note that the graph below can be used to classify existing operational healthcare information systems. This research focused on the interoperability approaches in online healthcare information systems that use common standards that are proposed in recent academic research in the field of healthcare information systems.

In the above classification, digital online systems that follow common standards can exchange information with each other with less effort if they are not connected. These standards have been developed to facilitate interoperability and information exchange at the institutional level. According to the announced standards for exchanging information between these systems, a team of specialists must establish the connection. Generally, this type of information exchange falls under institutional interoperability.

According to Gregory Zacharewicz’s model-based approaches for interoperability, interoperability barriers can be categorized in two ways. Horizontal barriers arise between different organizations or systems and are typically due to variations in processes, structures, and technologies. In contrast, vertical barriers occur within an organization and result from differing perspectives among stakeholders. This research focuses more on the horizontal than vertical barriers [50].

In the mentioned approaches in these systems, it is impossible to collect a specific patient’s health data online if it is stored in discrete systems with different interoperability standards or even the same standards but different implementation guides. Discrete systems are systems that are not able to exchange data with each other online, even if they are all online and accessible via the Internet. Currently, the most practical solution for implementing interoperability is to connect these systems online to each other by using common interoperability standards such as FHIR and OpenEHR.

Since the start, the EHR has presented various obstacles preventing consistent deployment. In a research study, Rincon identified a need for more security, audibility, and interoperability among these obstacles in Columbia. In addition, there is no comprehensive view of a patient’s medical history throughout their lifetime since different systems keep information separately. This lack of a unified history exposes patients to several hazards, including the exposure of private information, because each system has various safeguards and protections for the information. In some circumstances, these safeguards do not exist. Numerous scholars have attempted to develop many information systems to address this issue. However, these systems need a formal and rigorous architectural design for analyzing and obtaining health needs using architectural models to build robust systems to address these issues [51].

Based on the current situation for creating interoperability even between two EHR repositories with different data models, we need a tailored data pipeline which is designed especially for mapping those entities to integrate the EHR for a single patient. Obviously, it is a time-consuming and pricy process. Therefore, it means that it is not affordable nor scalable for a huge number of EHR. The inception of EHR has shown a lot of potential and virtually eliminated the drawbacks of paper-based medical notes. However, the transition has not been seamless due to various technical and political drawbacks [47].

Richardson evaluates her proposed framework for digital health equity in four levels:

• Individual Level: personal factors that affect a person’s ability to use digital health tools, including their technical skills, confidence, access to devices, and attitudes toward technology.
• Interpersonal Level: how relationships and interactions between people (including patients and healthcare providers and family members) influence the use of digital health tools.
• Community Level: the local infrastructure, resources, and norms that either enable or limit access to and use of digital health technologies in specific communities.
• Social Level: broader systemic factors like policies and standards on how digital health technologies are developed, distributed, and used across the entire population.

These levels show us that there is a hierarchy with different points of view, and they need to be considered when developing digital health solutions [52]. These levels exist in other countries as well, and if a system aims to make a digital transformation to make medical information globally accessible, it needs to interact with the legacy routines of each level.

Mandle et al. [7] mentioned the key challenges and limitations of accessing and leveraging data from EHRs for data-driven healthcare and research in four categories. Firstly, there is a lack of expertise in efficiently using EHR data for analytics, as these systems are not designed to be nimble analytic or data-sharing platforms. The process of extracting, transforming, and loading (ETL) EHR data into an analytic platform often requires a team of IT professionals, which is both expensive and time-consuming. Additionally, there is a lack of uniform data standardization; even when data are extracted, they may not be in a standardized format, making large-scale aggregation and analysis difficult. Accessing patient-level data across a population is also challenging, as EHRs traditionally do not provide an easy programmatic interface (API) to access all data elements without special effort. Finally, there are significant challenges with data privacy and security, as data extracted from EHRs must be protected under HIPAA requirements, adding an additional burden on organizations.

Based on the literature review, there are two types of interoperability: organizational and patient-centered. Current common standards such as FHIR are designed for the purpose of exchanging information at an institutional level. These standards do not have the capability nor were they designed to exchange information for a specific person from discrete online systems that do not have inter-organizational connections. They are designed to exchange data between organizations that are connected using the same implementation guide.

Kouremenou et al. [53] proposed an interoperability solution for data integration between diverse healthcare providers. However, they mentioned their systems limitations as follows. Firstly, the model’s effectiveness heavily relies on the uniform adoption of standards like HL7 FHIR. This adoption is not available across healthcare institutions, and they follow different customized versions, which is leading to gaps in interoperability between healthcare providers with various customization. Secondly, small healthcare providers often lack the resources and technical expertise to fully implement and maintain such models, which creates disparities in healthcare data-sharing capabilities. The terminologies and data structures have a high level of variety across FHIRs to support semantic and syntactic alignment, which means reconciling systems is complex and requires careful implementation. Finally, robust security and privacy within an interoperable framework is challenging. Different security protocols across systems can expose sensitive health information to risks. These limitations indicate the requirements of an ideal solution to deliver strong value to healthcare systems.

The University of Utah Health encountered inconsistencies in FHIR across diverse EHR systems to pursue interoperability EHR innovations. This variability often resulted in unreliable data access and made a real struggle in data integration. Although FHIR aims to standardize exchanging health data, its implementation varied across vendors, creating even more inconsistency, and also, privacy concerns increased, particularly with FHIR’s potential to expose sensitive patient information, such as HIV status, when sharing data with third-party apps. Resolving these issues requires intensive terminology mapping to align local EHR data with universal coding standards. The EHR initiative was reimagined and developed into FHIR Wrapper to provide an interface for applications to interact regardless of different FHIR implementations by EHR vendors. While FHIR Wrapper improved data access, it came with its own challenges. FHIR Wrapper required significant expertise and higher costs. The implementation of FHIR Wrapper can be costly and time-intensive, as it often involves building custom APIs to access data to solve inconsistencies in data access and formatting [54].

Gohar et al. designed a combination of blockchain and the cloud on his proposed Architecture for Better Semantic Interoperability in five layers. He categorized the necessity for using blockchain in healthcare information systems [55] as follows:

• Single point of failure;
• Privacy issues;
• Data breaches;
• Heterogeneity of healthcare data integration;
• Lack of interoperability.

The research mentioned that blockchain technology provides secure and tamper-resistant data storage to improve data management and security of EHRs. Blockchain provides secure data sharing and interoperability between healthcare providers, patients, and other stakeholders as a patient-centric healthcare infrastructure. Blockchain technology can help with data fragmentation and lack of transparency in the current healthcare system. Overall, blockchain has the potential to transform healthcare systems by providing secure, decentralized, and patient-centric data management solutions for integration with emerging technologies like AI, IoT, and big data more efficiently.

Gohar remarks on some attributes of their proposed architecture when they compare their proposition to the others in terms of the gap in needed functionality. The mentioned gap is patient-centric data management, including secure data sharing, access control, and privacy preservation [55]. They mentioned that another issue is integrating blockchain with existing healthcare systems and data standards like HL7 FHIR.

The critical issue facing PHR systems today is the absence of a secure, efficient, and universally accessible platform that supports patient co-ownership.

Lidstromer et al. conducted a systematic review to assess the status of PHR in current systems. They identified the key challenges and grouped them into twelve parameters, including security vulnerabilities, inefficiencies, high operational costs, and interoperability barriers. These factors limit the delivery of consistent global healthcare access and advances in AI-driven health innovations. The most notable challenges in creating a Global Patient Co-Owned Cloud (GPOC) are categorized as follows [56]:

• Security and Privacy: Protecting sensitive patient data on a global scale requires advanced encryption methods and secure access protocols to prevent unauthorized access while allowing controlled data sharing.
• Interoperability: Achieving compatibility across different healthcare systems worldwide is challenging but critical for a global PHR system. Interoperability would enable data exchange and integration with existing healthcare information systems infrastructures, and it requires adherence to standards like HL7 FHIR and close alignment with international healthcare practices.
• Data Ownership and Control: Granting patients co-ownership and control over their health records while supporting the healthcare providers and researchers needs is complex [56].

Addressing these challenges is essential to establishing a GPOC that is not only secure and efficient but also empowers patients in their healthcare journey and fosters global health innovation.

Integrating AI into healthcare through a shared digital data system is complicated, especially with issues in interoperability and establishing common standards. A big part of the problem is that data often remain locked in silos, different systems unable to exchange with each other, and the slow adoption of uniform standards does not help. On top of that, the long regulatory approval process can delay the implementation of new electronic health record (EHR) standards, which holds back efforts to improve how data flows across platforms. Building this kind of shared infrastructure also requires huge investment in technology, which can be a major hurdle for smaller healthcare providers. This financial burden often limits their ability to participate in broader data-sharing initiatives, which, in turn, narrows the pool of diverse contributors [57].

Another issue is a conflict of interest between public and private healthcare sectors. Some companies with proprietary EHR models prefer to keep their systems closed, rather than adopting public FHIR APIs, which makes it harder for systems to integrate. Private-sector incentives to profit from data access can restrict the availability of EHR data for outside developers and researchers who could benefit from it. Additionally, regulatory organizations add more complexity to the described situation, with diverse regulations and restrictions about patient rights, data privacy, and AI transparency. Furthermore, large closed-off ecosystems in the private sector can limit independent developers’ access to the comprehensive healthcare data needed to create solutions that truly work across different platforms [57].

One study aimed to investigate how the Patient-Centered Data Home (PCDH) can facilitate national interoperability via interconnected Health Information Exchange (HIE) networks to improve the continuity and quality of patient care [49]. Williams and Grannis mentioned that some of their study’s notable data interoperability challenges were data fragmentation, various technical standards, data governance misalignment, and patient matching, among other privacy and security concerns. Some of these are repeated in different studies and are obvious. Data fragmentation and various technical standards are only some of the obstacles to preventing broad access to healthcare data. There are conflicting priorities among healthcare information systems, state governments, and private sector entities that often delay data interoperability progress. Williams and Grannis mentioned that such misalignments make inconsistent policies, data governance structures, and operational data frameworks. Therefore, it limits the scalability and cohesion of data interoperability efforts at a national level [49].

We focused on various data interoperability approaches to identify the challenges of data interoperability in achieving patient-centered interoperability within healthcare information systems. We discussed the critical issues of patient-centered interoperability in existing standards such as FHIR and OpenEHR to empower patients for personal data retrieval across diverse systems. We categorized the challenges to simplify the complexity of healthcare interoperability issues, focus analysis, and align with research trends. These categories were developed through a detailed literature review, clustering challenges with shared attributes in a qualitative way. Each category captures a critical aspect of the problem:

4.2.1. Various Interoperability Standards

As mentioned earlier, lack of standardization for exchanging data between patients and healthcare service providers such as hospitals (patient-centered interoperability) and the variety of interoperability standards between healthcare information systems (exchanging data between hospitals) can lead to issues with interoperability. There are some interoperability standards, but they are designed for the implementation of interoperability between institutions such as from hospital to hospital, and they require the use of the same standard structure to be interoperable.

4.2.2. Privacy and Security

While health management information systems protect health records and are known for their security features, the decentralized nature of the interoperability approaches can also create privacy challenges in the handling of sensitive healthcare data. There are several organizations that ensure the secure exchange of patient records, and they apply many different regulation processes on the medical records to protect patients’ privacy based on privacy laws. We did not find any governmental organization that forces zero-trust environment concerns onto the healthcare information systems.

4.2.3. Governance

The regulatory organization in healthcare is still evolving its rules, which can create challenges for widespread compatibility. As mentioned earlier, there is no way to gather all the records of specific patients from a single point of contact (an interface) while those records are stored under certain standards or supervised by different organizations, such as gathering records of a patient from two different hospitals in two different countries or even same country but different standards.

4.2.4. Semantic Interoperability

A variety of measuring standards in medical records and different formats for presenting the same concept in medical records is solved partially by EMR/EHR, but it is still limited. For instance, the interpretation of the text in patients’ records is not supported yet. There are different approaches to solving this issue by researchers; using ontology or AI is the most common approach in the proposed solutions in the literature.

We extracted the patient-centered interoperability challenges in healthcare information systems through a comprehensive review of existing research on interoperability. We simplified the complexity of these challenges by categorizing the challenges into four main areas to focus our analysis. We reviewed the potential solution and categorized their approaches in the following section.

5. Results

5.1. Characteristics and Features of Patient-Centered Interoperability

Patient-centered interoperability allows patients to share and integrate their health information with multiple healthcare providers. It empowers patients to participate in their care processes with diverse healthcare providers worldwide as they wish. Patient-centered interoperability shifts the focus from healthcare provider-centered (institutional interoperability) to a more inclusive approach (patient-centered interoperability) in which patients have permission to access and manage their health data and have the flexibility to communicate with various healthcare services [1].

We reviewed the propositions and their diverse approaches to the above-mentioned issues. We applied a qualitative analysis to develop a summary of the reviewed paper to characterize the features of a patient-centered interoperability solution. We the studied research propositions to understand their features and then developed the following categorization based on the frequent features. We identified the common features in the propositions and developed the following classification. If they mentioned any of the following features in their proposition, we categorized that paper in the corresponding group. Here are the common features of the propositions:

• EMR Integration: It refers to the ability of the health information system to integrate with electronic medical records that are common in old systems. Often, obsolete file formats and data structures need to be integrated with current systems and files in an organization.
• Using EHR: This system can communicate patient electronic health records in the form of EHRs. It is like EMRs but in a larger scope across some connected healthcare entities such as clinics, labs, etc.
• Adapting FHIR: This approach involves the compatibility of systems with the FHIR standard, which allows for data exchange with other systems. Supporting common interoperability standards reduces integration complexity in a heterogeneous ecosystem. It facilitates efficient data interoperability between the data silos of the healthcare information systems.
• Using Blockchain: Blockchain technology has multiple features to level up HIS in several aspects like security and transparency. These advantages are the cause of the emerging blockchain-based healthcare systems. One of these features is DAO. A decentralized autonomous organization (DAO) is a fascinating concept that can be implemented by blockchain technology. It is a self-governing organization that is constructed by smart contracts. DAO can provide automated decision-making and resource distribution within decentralized frameworks with no supervision. It relies on blockchain’s transparency and security capabilities to ensure trust and accountability.
• Semantic Interoperability: Semantic interoperability focuses on accurately interpreting and transmitting the meaning of data across healthcare systems. It provides accurate and contextually appropriate information transmission between various healthcare information systems. Supporting semantic interoperability means standardized vocabulary, ontologies, and data models are used to improve the shared understanding of health information, reduce ambiguity, and improve communication between disparate systems. Systems with semantic interoperability features offer more meaningful information exchange, facilitating improved clinical decision-making and overall healthcare delivery quality.
• Personal Data Retrieval: It emphasizes empowering individuals to manage and control their health information. It allows secure access to health records across many healthcare platforms.

These categories help us evaluate the similarity and differentiation of solutions and framework approaches. They were developed through a detailed literature review and qualitative classification of shared challenges. Each category captures an aspect of problem-solving to evaluate a proposed solution or framework. This categorization is valuable for analyzing the similarity of solutions and frameworks. It helps us compare which combination of features is more common and which is not. Table 2 presents the stats of the features in the reviewed papers.

Table 2. Feature stats of the reviewed interoperability propositions.

Feature	Papers	n 1	n/N 2	~% 3
EMR Integration	[2,5,7,12,13,21,24,27,28,29,31,33,35,39,40,42,51,54,58]	19	19/56	34%
Using EHRs	[1,6,13,21,27,28,29,31,33,39,40,44,46,54,55,58,59,60]	18	18/56	32%
Adapting FHIR	[4,6,13,14,21,27,28,41,46,47,48,53,54,57,61,62]	16	16/56	29%
Using Blockchain	[1,6,7,9,10,11,12,13,23,24,26,29,30,31,32,33,34,35,36,37,38,39,40,55,60,61,63]	27	27/56	48%
Semantic Interoperability	[13,14,21,25,27,28,31,44,46,55,64,65]	12	12/56	21%
Personal Data Retrieval	[1,6,7,12,13,20,23,24,29,32,33,34,35,36,40,43,45,48,49,52,55,56,60]	23	23/56	41%

¹ Number of related papers. ² Total unique papers in this table. ³ Approximated percentage.

Table 2. Feature stats of the reviewed interoperability propositions.

Feature	Papers	n 1	n/N 2	~% 3
EMR Integration	[2,5,7,12,13,21,24,27,28,29,31,33,35,39,40,42,51,54,58]	19	19/56	34%
Using EHRs	[1,6,13,21,27,28,29,31,33,39,40,44,46,54,55,58,59,60]	18	18/56	32%
Adapting FHIR	[4,6,13,14,21,27,28,41,46,47,48,53,54,57,61,62]	16	16/56	29%
Using Blockchain	[1,6,7,9,10,11,12,13,23,24,26,29,30,31,32,33,34,35,36,37,38,39,40,55,60,61,63]	27	27/56	48%
Semantic Interoperability	[13,14,21,25,27,28,31,44,46,55,64,65]	12	12/56	21%
Personal Data Retrieval	[1,6,7,12,13,20,23,24,29,32,33,34,35,36,40,43,45,48,49,52,55,56,60]	23	23/56	41%

¹ Number of related papers. ² Total unique papers in this table. ³ Approximated percentage.

5.2. Similarity Analysis by Jaccard Coefficient

The analysis (Table 2) reveals that using blockchain and personal data retrieval were the most frequently mentioned features, each appearing in 48% and 41% of the reviewed papers. The results indicate that researchers prioritize two key functionalities as the most desired features for advancing interoperability in healthcare systems, which means they are not commonly utilized across these domains operationally, and these efforts tried to address these gaps to improve the interoperability of healthcare data.

We have conducted a similarity analysis using the Jaccard coefficient to evaluate the relationships between each pair of features in Table 2. This correlation analysis is a data-driven approach for understanding the interdependencies and gaps between the extracted features from reviewed papers. It helps to understand the reviewed papers on data interoperability in healthcare information systems. It provides us a big picture of the complex interdependencies within interoperability research and gives us insights for understanding the trends of studies in patient-centered data interoperability in healthcare information systems. We analyzed these interdependencies as a steppingstone to address data interoperability challenges. The outcome is assessed by how advancements in each area influence other areas.

Table 3. A summary of the calculated metrics for all possible combinations of features.

Feature A	Feature B	n(A ∩ B)	n(A U B)	n(A ∆ B)	J(A, B)
EMR Integration	Using EHRs	11	26	15	0.42
EMR Integration	Adapting FHIR	5	30	25	0.17
EMR Integration	Using Blockchain	10	36	26	0.28
EMR Integration	Semantic Interoperability	5	26	21	0.19
EMR Integration	Personal Data Retrieval	8	34	26	0.24
Using EHRs	Adapting FHIR	7	27	20	0.26
Using EHRs	Using Blockchain	10	35	25	0.29
Using EHRs	Semantic Interoperability	8	22	14	0.36
Using EHRs	Personal Data Retrieval	8	33	25	0.24
Adapting FHIR	Using Blockchain	3	40	37	0.08
Adapting FHIR	Semantic Interoperability	6	22	16	0.27
Adapting FHIR	Personal Data Retrieval	3	36	33	0.08
Using Blockchain	Semantic Interoperability	3	36	33	0.08
Using Blockchain	Personal Data Retrieval	16	34	18	0.47
Semantic Interoperability	Personal Data Retrieval	2	33	31	0.06

All possible pairs of features are taken into account.

Table 3. A summary of the calculated metrics for all possible combinations of features.

Feature A	Feature B	n(A ∩ B)	n(A U B)	n(A ∆ B)	J(A, B)
EMR Integration	Using EHRs	11	26	15	0.42
EMR Integration	Adapting FHIR	5	30	25	0.17
EMR Integration	Using Blockchain	10	36	26	0.28
EMR Integration	Semantic Interoperability	5	26	21	0.19
EMR Integration	Personal Data Retrieval	8	34	26	0.24
Using EHRs	Adapting FHIR	7	27	20	0.26
Using EHRs	Using Blockchain	10	35	25	0.29
Using EHRs	Semantic Interoperability	8	22	14	0.36
Using EHRs	Personal Data Retrieval	8	33	25	0.24
Adapting FHIR	Using Blockchain	3	40	37	0.08
Adapting FHIR	Semantic Interoperability	6	22	16	0.27
Adapting FHIR	Personal Data Retrieval	3	36	33	0.08
Using Blockchain	Semantic Interoperability	3	36	33	0.08
Using Blockchain	Personal Data Retrieval	16	34	18	0.47
Semantic Interoperability	Personal Data Retrieval	2	33	31	0.06

All possible pairs of features are taken into account.

We marked the strengths and gaps of feature combinations in a quantitative data-driven approach. This analysis guides research priorities, alignments, and future development in data interoperability of HIS through numeric relationships. We aimed to provide actionable insights for pointing out gaps and synergies in the feature combinations to drive future innovations and research.

5.2.1. Description of Metrics

Table 3 calculates the metrics based on the indication of each paper on each pair of approaches for all possible pairs. The following summarizes the metrics formula and their implications:

1. 

(n(A ∩ B));

a.

Number of papers in the intersection between feature A and feature B;

b.

The intersection refers to the number of distinct papers that are present in both feature A and feature B;

c.

For instance, the value of (n(A ∩ B)) between “using blockchain” and “personal data retrieval” at the intersection is 16, indicating strong alignment;

2. 

(n(A ∪ B));

a.

Number of papers in the union of feature A and feature B;

b.

The union refers to the total number of distinct papers that are present in either feature A or feature B;

c.

For instance, the value of (n(A ∪ B)) between “adapting FHIR” and “using blockchain” at the union is 40;

3. 

(n(A ∆ B));

a.

Number of papers in the symmetric difference between feature A and feature B;

b.

The symmetric difference (delta) refers to the total number of distinct papers that are present in either feature A or feature B but not both;

c.

For instance, the value of (n(A ∆ B)) between “adapting FHIR” and “using blockchain” at the union is 37;

d.

(n(A ∆ B)) = (n(A ∪ B)) − (n(A ∩ B));

4. 

J(A, B);

a.

The Jaccard index is used to calculate the similarity between two sets (feature A and feature B). It is defined as follows:

i.

J(A, B) = ${\displaystyle \frac{\mathrm{n}(\mathrm{A}\cap \mathrm{B})}{\mathrm{n}(\mathrm{A}\cup \mathrm{B})}}$;

b.

The resulting Jaccard index ranges from 0 to 1;

i.

J(A, B) = 1: features are identical, with complete overlap;

ii.

J(A, B) = 0: features have no similarity or overlap.

5.2.2. Analysis of Metrics and Insights

• Highest Jaccard index:

  a.

  The highest Jaccard index is 0.47, between “using blockchain” and “personal data retrieval”. This indicates significant alignment, with 16 shared papers out of 34.

  b.

  The second highest Jaccard index is 0.42, between “EMR integration” and “using EHRs”. This indicates significant alignment, with 11 shared papers out of 26.

  c.

  Both represent strong synergies, where advancements in one feature may influence or support the other. This combination was seen in various propositions, and both features are compatible for simultaneous use.

  d.

  Both combinations are consistent and found in various studies, which means it can be used as an effective practice in system development.

• Lowest Jaccard index:

  a.

  The lowest Jaccard index is 0.06, between “semantic interoperability” and “personal data retrieval”. Only 2 shared papers exist out of 33.

  b.

  The second lowest Jaccard index is 0.08, between three pairs of features:

  i.

  “Semantic interoperability” and “personal data retrieval”;

  ii.

  “Adapting FHIR” and “personal data retrieval”;

  iii.

  “Adapting FHIR” and “semantic interoperability”.

  c.

  These combinations with a few shared papers show minimal alignment and synergies. It indicates distinct approaches, which suggests that researchers are not exploring these combinations of features in their proposals.

  d.

  These combinations can be used for future research to cover the gaps in the research, allowing researchers to develop unique contributions.

• Sorting features combinations by Jaccard index:

A higher Jaccard index represents a greater balance between the intersection of elements (number of papers) and their symmetric difference over their union. Such paired features represent an effective feature combination for system development and practical innovations due to their strong research foundation.

A lower Jaccard index indicates a weak balance between the intersection of elements (i.e., the number of papers) and their symmetric difference relative to their union. This combination of paired features needs to be sufficiently studied and emphasizes the need for further research to determine their effectiveness for system development and practical innovations. This combination has significant potential for unique contributions due to the need for a solid research foundation in this area.

Table 3 narrates the relationships and gaps in the common features of propositions in healthcare data interoperability. The Jaccard indices of these feature combinations can be used as a knowledge foundation to facilitate starting research on patient-centered interoperability. The analysis above also helps easily identify new domains for innovations in healthcare information systems. This analysis can also be used as a roadmap for future research to emphasize the strengths and limitations of current approaches.

6. Contributions

Our contributions to healthcare data interoperability are focused on the critical gap in patient-centered data interoperability and providing a roadmap for approaching solutions to bridging this gap, as follows:

• Comprehensive patient-centered data interoperability review: We reviewed the challenges of patient-centered interoperability and categorized them into four groups to simplify the problem. We categorized common features of the propositions to simplify understanding of the characteristics of a solution. We also challenged and developed the first pillar of the multifaceted analysis of the problem.
• Scenario-driven analysis through a failed scenario: We contextualized the limitations of current data interoperability in healthcare information systems through real-world scenarios. Our scenario-driven approach defined a realistic understanding of the patient-centered interoperability challenge outcomes.
• Systematic evaluation of existing solutions: We characterized the patient-centered interoperability approaches via a literature review. Our qualitative approach extracted common features of the propositions from 56 notable research papers to simplify solution evaluation. We evaluated the approaches of the proposed solution extracted from each study to point out patient-centered interoperability.
• Correlation analysis for developing data-driven results and insights: Our quantitative correlation analysis using the Jaccard index validated the qualitative findings from the literature by mapping the approaches to related studies. We calculated the Jaccard coefficient on the number of papers that show co-occurrence of each pair of features and their symmetric differences over their union. The calculation is fully explained in the Section 5. We analyzed the outcomes of the calculations to develop data-driven insights and bridged the gap between qualitative analysis and quantitative analysis.
• Roadmap for future research: Insights were derived from synthesizing the qualitative and quantitative analyses. We synthesized the challenges and the characterization of patient-centered approaches to the solutions to develop a roadmap for future research. This approach developed a framework for understanding the challenges and solutions in patient-centered interoperability.

This research not only provides an understanding of patient-centered interoperability challenges but also provides data-driven insight into the foundation of future research and innovations. It could be used as a roadmap for developing patient-centered solutions or research to prioritize patient needs for data interoperability.

7. Discussion

We discussed various data interoperability challenges and approaches to achieving patient-centered interoperability within healthcare information systems. We reviewed the limits of FHIR to achieve patient-centered interoperability in diverse systems, such as personal data retrieval. Using blockchain (48%) and personal data retrieval (41%) emerged as the most frequently cited features, which indicates their potential for addressing healthcare data interoperability. The state of the art revealed that shifting the focus from institutional interoperability to a patient-centered model can empower patients to manage and share their health records without borders. This change can help align healthcare delivery with the contemporary needs of data accessibility and patient autonomy. In this environment, patients are able to retrieve health records from various healthcare systems via a secure single point of connection (such as an interface). Further research can work on the diverse methods of data retrieval in supporting patient-centered interoperability through a single connection from diverse healthcare providers with various implementation guides and standards.

The Jaccard coefficient (similarity analysis) provided data-driven insights into the relationships between features. The high Jaccard index between blockchain technology and personal data retrieval (0.47) represents the alignment and synergy of these two features and reveals where advancements in one feature can complement the other effectively. These findings suggest that integrating these features into solutions can improve patient-centered data interoperability. Conversely, the low Jaccard indices between semantic interoperability and personal data retrieval (0.06) and between adapting FHIR and personal data retrieval (0.08) reveal significant gaps in the research. These low-alignment combinations represent opportunities for future exploration to develop unique contributions.

We comprehensively explored the challenges in achieving patient-centered interoperability in healthcare information systems. However, this research is not conducted on practical approaches like semantic interoperability tools or governance regulation rules. We have performed a qualitative analysis on the reviewed papers to categorize them by common features of the propositions. We have also performed a quantitative analysis on the feature correlations using Jaccard indices and symmetric differences to create insights and suggest future directions in a data-driven manner. Feature combinations with low Jaccard indices are good subjects for future research to deliver a unique contribution, and all combinations with high Jaccard indices are good practices to embed into solutions for future innovation. We were limited to the number of reviewed papers, and the order of all the proposed feature combinations located at the bottom section of Figure 4 can be changed with new research. The main reason for this is that the Jaccard index for these combinations was created based on the number of indicated papers, and new research can change their order from that in Figure 4. In our research, we do not assess the risk of bias in the reviewed papers; therefore, the outcome likewise lacks such an assessment.

8. Conclusions

The widespread use of EHRs and healthcare information systems (HISs) has led to isolated data silos across healthcare providers and countries [66]. It is a crucial need to comprehend the challenges of data interoperability to transform fragmented health data silos into health records without borders. We have conducted a literature review on various data interoperability studies to understand the diverse aspects of the problems and review the propositions to identify challenges. This survey reviewed the challenges and approaches to achieving patient-centered interoperability within healthcare information systems.

We synthesized the challenges and approaches to patient-centered data interoperability solutions to develop a set of common features as a result of the synthesis. We extracted these common features via a literature review based on frequent use in the reviewed propositions. We categorized the propositions of the reviewed papers by their common features in Table 2 and analyzed the categorization results by calculating their intersections, unions, symmetric differences, and Jaccard indices to provide a data-driven foundation for our analysis in Table 3. The similarity analysis in Table 3 presents the quantitative results and data-driven insights from this review. The following is the developed categorization classifying the common features of the reviewed papers’ propositions as the outcome of the qualitative analysis.

• EMR integration;
• Using EHRs;
• Adapting FHIR;
• Using blockchain;
• Semantic interoperability;
• Personal data retrieval.

The analysis (Table 2) reveals that using blockchain and personal data retrieval were the most frequently mentioned features, appearing in 48% and 41% of the reviewed papers, respectively. The results indicate that researchers prioritize these two key functionalities as the most desired features for advancing interoperability, and interoperability in healthcare systems is not commonly utilized across these domains.

We performed a correlation analysis in Table 2 to evaluate the relationships between the features for creating data-driven insights. All the formulas and calculations of the Section 5 created the outcome that is presented in Table 3. The calculation of Jaccard indices, symmetric differences, intersections, and unions was based on the co-occurrence of the paired features mentioned in the propositions of the reviewed papers and extracted from Table 2.

The Jaccard index between “using blockchain” and “personal data retrieval” is 0.47, which indicates strong alignment and synergies between these two features. Personal data retrieval is an essential part of patient-centered interoperability. Therefore, this alignment also shows blockchain as being an essential part of a desired patient-centered data interoperability solution. We analyzed all the possible combinations of the common features. We sorted them in Figure 4 from a good combination for developing systems (higher Jaccard index) to great candidates for further research to create unique contributions (lower Jaccard index).

In conclusion, this research reviewed and analyzed 56 notable papers on data interoperability in healthcare information systems to develop data-driven insight. We applied qualitative and quantitative analyses to highlight the challenges’ ambiguity and extract data-driven insight from their propositions.