*2.1. CGM Applications*

During the last decade different authors studied the use of CGMs in order to apply them in different scenarios. For instance, in [12] the possibility of adding intelligence to CGM sensors is analyzed in order to issue alerts when glucose levels are not within the appropriate range (e.g., in case of a hypoglycaemia/hyperglycaemia). The same authors focused later on reviewing the state of the art on contributions related to the development of hardware and smart algorithms for CGMs [13]. Similar reviews on the evolution of CGM technology can be found in [14–16].

Other CGM features have been studied by other authors. For instance, some authors focused on analyzing off-the-shelf CGM calibration algorithms and on how the amount of glucose in blood plasma is obtained [17]. Other researchers studied how CGMs detect losses in the performance on insulin infusion set actuation [18], which may derive into prolonged hyperglycemia in Type-1 DM patients. A CGM is also able to decipher the influence of daily habits on the health of DM patients. For example, physical exercise clearly alters glucose concentration levels on Type-1 DM patients: glucose regulation problems may arise during or after exercising and even when performing certain daily activities [19].

Finally, it is worth mentioning that a CGM can collaborate with an artificial pancreas in order to provide the optimal insulin dose through an insulin pump. However, there are different parameters that influence the infusion of insulin, with physical activity being one of the most challenging as it may result in hypoglycemia in Type-1 DM patients [20].

### *2.2. Fog Computing, Blockchain and Crowdsourcing for Healthcare Applications*

In the last years, cloud computing has achieved remarkable success thanks to its ability to offload computational-intensive tasks from clients [21]. Nonetheless, in applications where low latency responses are needed, other paradigms like fog computing have proven to be valuable [6]. Fog computing is usually considered a paradigm that is an extension of cloud computing where part of the computational and communication capabilities of the cloud are moved close to the sensor nodes [7], which derives into several advantages [22]:

• Novel IoT healthcare real-time applications can be provided thanks to decreasing latency.

The fog allows for sharing computational and storage resources, which can be harnessed by distributed wireless sensor networks that can be deployed in hospitals or in other healthcare facilities.


Despite the mentioned benefits and its multiple applications in diverse fields [23], there are not many practical examples of the use of fog computing in healthcare applications. A good compilation of fog computing based healthcare applications can be found in [24], which describes potential applications for monitoring Chronic Obstructive Pulmonary Disease (COPD) patients [25], tools for in-home Parkinson disease treatments [26] or hospital platforms that make use of e-textiles [27] and wireless sensor networks [28]. Similarly, other researchers proposed applications for healthcare monitoring in smart homes [29] or for diagnosing and preventing outbreaks of certain viruses [30].

Another challenge faced by today's healthcare researchers and practitioners is the lack of interoperability between different technological platforms. Incompatible, unscalable and independent systems hinder the development of novel end-to-end patient-centered research and healthcare solutions. Unfortunately, it is difficult to exchange information among healthcare providers, patients and third-parties (e.g., insurance companies, governments, app developers). Records are often incomplete, fragmented or unavailable at the point of care and it is difficult to access patient's health information [31,32]. The integration complexity lies mainly in the lack of access outside a specific healthcare environment and in the use of incompatible and proprietary software and hardware. Although there are some standards (e.g., Health Level Seven (HL7) Fast Healthcare Interoperability Resources (FHIR), HL7 Clinical Document Architecture (CDA), ISO13606, openEHR, CDISC Operational Data Model (ODM)) for exchanging data between trusted parties, their implementation requires data mapping and additional interface adaptations [32].

A promising alternative to solve the aforementioned problems consists in using Distributed Ledger Technologies (DLTs). Examples of DLT platforms currently in use are Ethereum [33], Hyperledger Fabric [34] or IOTA [35]. DLTs, specifically blockchain, are predicted to be key technologies within the Industry 4.0 era, since they guarantee the exchange of information between different stakeholders and interested parties that do not necessarily trust each other [36]. Moreover, blockchain holds the promise of enhanced data transparency, trustworthiness, immutability, privacy and security. Furthermore, it enables Peer-to-Peer (P2P) transactions, decentralized Apps (DApps), operational efficiency and a high degree of automation thanks to the use of smart contracts that execute code autonomously [37].

There are recent reviews on the usage of blockchain to enhance the healthcare sector [31,32,38–42]. Such works essentially analyze promising technologies, potential applications and discuss potential challenges for their further adoption. Currently, in the literature there are only a couple of recent preliminary works that use a blockchain for diabetes research and care. An example is [43], where the authors give an overview of a blockchain-based architecture that implements data and access management. An implementation of an Ethereum IoT platform architecture to take care and monitor DM patients can be found in [44].

Finally, it is worth noting that, with the rise of IoT [45,46], crowdsourcing is gaining momentum in a wide range of sectors and tasks [47–51]. In healthcare, crowdsourcing has been mainly employed to accomplish problem solving, data processing, surveillance/monitoring, and surveying [52], but there are not many mobile crowdsourcing applications [11,53] and even less decentralized while focused on healthcare [48]. In health crowdsourcing, engagement is essential since it can transform users from mere passive recipients of information to active participants in a collaborative community, raising awareness for diseases like diabetes and helping to improve their own health as well as the health of those around them [54]. In addition, there are incentive mechanisms that enable community participation [48]. For example, the authors of [55] proposed an incentive mechanism to encourage hospitals to share high-quality data, which can then be aggregated to generate prediction models with higher accuracy rates.
