**1. Introduction**

The investigation of all living organisms and complex diseases, e.g., yeast, human, cancer, and Alzheimer's has highlighted the need for a new holistic vision to shed light on the multiple interactions among several biological players, such as genes, enzymes, and small molecules. In the reductionist approach [1], a single mutation or weakness is responsible for diseases and phenotype aberrancies. In contrast, the holistic approach [2] asserts that conditions and phenotype aberrancies are due to the intricate interactions network among several biological players.

The appearance of omics sciences [3] provides the approaches to consolidate the holistic idea mandatory for studying living organisms at all structural and functional levels, including humans. Omics includes the domains ending in -omics, such as proteomics, epigenomics, metabolomics, and microbiomics. In particular, the rapid advances in High-Throughput (HT) and Molecular Biology (MB) technologies make omics sciences a central part of medical research. The continuous technological advances in HT and MB have made it possible to comprehensively analyze a simple living organism's genome, e.g., a single bacteria, and a complex organism, e.g., humans, in a few hours or a few days [4,5]. The highest quality of HT and MB produces massive data per single experiment, transforming biology and genomics into data-driven sciences [6]. Only the practical analysis of this enormous amount of data will allow us to understand the complex aberrancies starting from the genome.

The transition of life sciences toward data-driven science provides researchers with new opportunities, making it possible to yield vast amounts of omics data in a cost-

**Citation:** Agapito, G.; Cannataro, M. An Overview on the Challenges and Limitations Using Cloud Computing in Healthcare Corporations. *Big Data Cogn. Comput.* **2023**, *7*, 68. https:// doi.org/10.3390/bdcc7020068

Academic Editor: Carson K. Leung

Received: 13 March 2023 Revised: 27 March 2023 Accepted: 29 March 2023 Published: 6 April 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

and time-efficient manner. Simultaneously, acquiring, storing, distributing, analyzing, and interpreting these data is challenging [7]. The high data heterogeneity in terms of type and source requires technical improvements in many Information Technology (IT) domains, raising various privacy, security, storage, sharing, processing, and computing power issues. Hence, it is essential to develop specific algorithms and software tools for analyzing the different available types of omics data, such as protein sequences, single nucleotide polymorphisms (SNPs), and gene expressions, necessary for understanding the expression of genes and their regulation and the mutations in DNA underlying genetic diseases. A further contribution is the development of graphic interfaces that effectively display information from various data sources.

To meet these challenges, Cloud Computing and High-Performance Computing (HPC) architectures can significantly improve the speed of omics data investigation, analysis, reliability, and reproducibility.

Architectures based on multiprocessors, even multi-core, Graphics Processing Units (GPUs), and hybrids architectures, e.g., holding both GPUs and CPUs, make HPC architectures ideal for handling computations requiring significant amounts of computing power and memory. The strength of HPC systems is the extreme computational power obtained through parallel or distributed computing.

Parallel programming enables us to write code in order to take advantage of the multiple computational cores of modern CPUs. Parallel programming decomposes programs, e.g., the process, as several independent bunches, e.g., the threads allowing parallel and concurrent execution. Partitioning programs into smaller threads allows the exploitation of multiple cores within modern CPUs. Multiple cores on a single machine share memory. Hence, threads can be executed simultaneously using shared memory to synchronize and communicate with each other. A popular environment for threads is Java thread, while CUDA is a popular environment for exploiting the computational properties of Graphics Processing Units (GPU). Distributed computing uses network protocols such as TCP/IP, allowing applications to send and receive data to each other over the network by providing the services and protocols for exchanging data. Hence, a distributed application is built upon several layers. At the lower level, the network connects devices, allowing communication among them. At the higher level, services are defined on the network protocols. Finally, distributed applications run on top of these layers to perform tasks across the network. A popular library for distributed computing is Message Passing Interface (MPI) [8], which is available for many programming languages and architectures. Hence, parallel and distributed computing allows for solving complex problems in a short time by employing many computing resources simultaneously that would otherwise require a lot of time if performed sequentially.

Thus, programmers must explicitly develop parallel programs, e.g., in a global environment using a multi-threading paradigm or in a distributed environment through the Message-Passing Interface (MPI) standard [8], to exploit the computational power delivered from HPC systems. In addition, to ensure that HPC systems run at optimal performance, a suitable technical support service is required. All these constraints introduce additional expenses, e.g., purchase, maintenance, and development, making the HPC systems ideal for large IT research centers and limiting the spread in biological, medical, and genomics research centers. The limiting element for the significant employment of HPC is nowadays primarily computational. On the other hand, Cloud Computing [9,10] brings a new paradigm from the analogy with existing infrastructures, such as electricity or water. Consequently, the achievement of the results is guaranteed independently of where data or instruction sequences are stored or executed. When opening a tap or turning on a lamp, one does not wonder where the water or electricity comes from; the important thing is that these are made available. Similarly, when some commands or services need to be executed in the Cloud system, it does not matter who takes care of it; the overall system will have to deliver the correct results based on the user requests. Thus, Cloud Computing provides an on-demand system through the Internet. Therefore, it eliminates purchase,

maintenance, and development costs, making high-performance computation available even for small research centers. Cloud Computing is available in three fundamental models, such as IaaS (Infrastructure as a Service), PaaS (Platform as a Service), and SaaS (Software as a Service).

Cloud Computing could be the ideal tool for dealing with many steps of the bioinformatics analysis pipeline, from pre-processing, selection, aggregation, and analysis, including exploration and visualization.

To take full advantage of the considerable benefits of Cloud Computing, healthcare corporations must face several management, technology, security, and legal issues that affect its rapid adoption in healthcare. For example, storing confidential health information in third-party remote data storage raises serious problems related to the patient's sensitive information because patient data could be lost or misused.

Thus, the present study aims to highlight the challenges, security issues, and impediments that limit the adoption of cloud computing in healthcare corporations.

The rest of the manuscript is arranged as follows: Section 2 describes the principal service and deployment models of Cloud Computing, highlighting the main difference between them. Section 3 introduces some well-known Cloud services suitable for handling Big omics Data. Section 4 describes challenges, security issues, and impediments that are limiting the spread of Cloud Computing in healthcare corporations. Section 5 discusses some of the possible challenges and issues to address underlying the low adoption of Cloud services in healthcare. Section 6 provides some guidelines to follow, when dealing with Cloud Computing in healthcare. Finally, Section 7 concludes the manuscript.
