Deep reinforcement learning for multi-class imbalanced training: applications in healthcare.

With the rapid growth of memory and computing power, datasets are becoming increasingly complex and imbalanced. This is especially severe in the context of clinical data, where there may be one rare event for many cases in the majority class. We introduce an imbalanced classification framework, based on reinforcement learning, for training extremely imbalanced data sets, and extend it for use in multi-class settings. We combine dueling and double deep Q-learning architectures, and formulate a custom reward function and episode-training procedure, specifically with the capability of handling multi-class imbalanced training. Using real-world clinical case studies, we demonstrate that our proposed framework outperforms current state-of-the-art imbalanced learning methods, achieving more fair and balanced classification, while also significantly improving the prediction of minority classes. Supplementary Information: The online version contains supplementary material available at 10.1007/s10994-023-06481-z.

Continuous patient state attention model for addressing irregularity in electronic health records.

BACKGROUND: Irregular time series (ITS) are common in healthcare as patient data is recorded in an electronic health record (EHR) system as per clinical guidelines/requirements but not for research and depends on a patient's health status. Due to irregularity, it is challenging to develop machine learning techniques to uncover vast intelligence hidden in EHR big data, without losing performance on downstream patient outcome prediction tasks. METHODS: In this paper, we propose Perceiver, a cross-attention-based transformer variant that is computationally efficient and can handle long sequences of time series in healthcare. We further develop continuous patient state attention models, using Perceiver and transformer to deal with ITS in EHR. The continuous patient state models utilise neural ordinary differential equations to learn patient health dynamics, i.e., patient health trajectory from observed irregular time steps, which enables them to sample patient state at any time. RESULTS: The proposed models' performance on in-hospital mortality prediction task on PhysioNet-2012 challenge and MIMIC-III datasets is examined. Perceiver model either outperforms or performs at par with baselines, and reduces computations by about nine times when compared to the transformer model, with no significant loss of performance. Experiments to examine irregularity in healthcare reveal that continuous patient state models outperform baselines. Moreover, the predictive uncertainty of the model is used to refer extremely uncertain cases to clinicians, which enhances the model's performance. Code is publicly available and verified at https://codeocean.com/capsule/4587224 . CONCLUSIONS: Perceiver presents a computationally efficient potential alternative for processing long sequences of time series in healthcare, and the continuous patient state attention models outperform the traditional and advanced techniques to handle irregularity in the time series. Moreover, the predictive uncertainty of the model helps in the development of transparent and trustworthy systems, which can be utilised as per the availability of clinicians.

Blockchain vehicles for efficient Medical Record management.

The lack of interoperability in Britain's medical records systems precludes the realisation of benefits generated by increased spending elsewhere in healthcare. Growing concerns regarding the security of online medical data following breaches, and regarding regulations governing data ownership, mandate strict parameters in the development of efficient methods to administrate medical records. Furthermore, consideration must be placed on the rise of connected devices, which vastly increase the amount of data that can be collected in order to improve a patient's long-term health outcomes. Increasing numbers of healthcare systems are developing Blockchain-based systems to manage medical data. A Blockchain is a decentralised, continuously growing online ledger of records, validated by members of the network. Traditionally used to manage cryptocurrency records, distributed ledger technology can be applied to various aspects of healthcare. In this manuscript, we focus on how Electronic Medical Records in particular can be managed by Blockchain, and how the introduction of this novel technology can create a more efficient and interoperable infrastructure to manage records that leads to improved healthcare outcomes, while maintaining patient data ownership and without compromising privacy or security of sensitive data.

Design Choices and Trade-Offs in Health Care Blockchain Implementations: Systematic Review.

BACKGROUND: A blockchain is a list of records that uses cryptography to make stored data immutable; their use has recently been proposed for electronic medical record (EMR) systems. This paper details a systematic review of trade-offs in blockchain technologies that are relevant to EMRs. Trade-offs are defined as "a compromise between two desirable but incompatible features." OBJECTIVE: This review's primary research question was: "What are the trade-offs involved in different blockchain designs that are relevant to the creation of blockchain-based electronic medical records systems?" METHODS: Seven databases were systematically searched for relevant articles using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Papers published from January 1, 2017 to June 15, 2018 were selected. Quality assessments of papers were performed using the Risk Of Bias In Non-randomized Studies-of Interventions (ROBINS-I) tool and the Critical Assessment Skills Programme (CASP) tool. Database searches identified 2885 articles, of which 15 were ultimately included for analysis. RESULTS: A total of 17 trade-offs were identified impacting the design, development, and implementation of blockchain systems; these trade-offs are organized into themes, including business, application, data, and technology architecture. CONCLUSIONS: The key findings concluded the following: (1) multiple trade-offs can be managed adaptively to improve EMR utility; (2) multiple trade-offs involve improving the security of blockchain systems at the cost of other features, meaning EMR efficacy highly depends on data protection standards; and (3) multiple trade-offs result in improved blockchain scalability. Consideration of these trade-offs will be important to the specific environment in which electronic medical records are being developed. This review also uses its findings to suggest useful design choices for a hypothetical National Health Service blockchain. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.2196/10994.

Implementing Blockchains for Efficient Health Care: Systematic Review.

BACKGROUND: The decentralized nature of sensitive health information can bring about situations where timely information is unavailable, worsening health outcomes. Furthermore, as patient involvement in health care increases, there is a growing need for patients to access and control their data. Blockchain is a secure, decentralized online ledger that could be used to manage electronic health records (EHRs) efficiently, therefore with the potential to improve health outcomes by creating a conduit for interoperability. OBJECTIVE: This study aimed to perform a systematic review to assess the feasibility of blockchain as a method of managing health care records efficiently. METHODS: Reviewers identified studies via systematic searches of databases including PubMed, MEDLINE, Scopus, EMBASE, ProQuest, and Cochrane Library. Suitability for inclusion of each was assessed independently. RESULTS: Of the 71 included studies, the majority discuss potential benefits and limitations without evaluation of their effectiveness, although some systems were tested on live data. CONCLUSIONS: Blockchain could create a mechanism to manage access to EHRs stored on the cloud. Using a blockchain can increase interoperability while maintaining privacy and security of data. It contains inherent integrity and conforms to strict legal regulations. Increased interoperability would be beneficial for health outcomes. Although this technology is currently unfamiliar to most, investments into creating a sufficiently user-friendly interface and educating users on how best to take advantage of it would lead to improved health outcomes. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): RR2-10.2196/10994.