Key result
A Random-Forest machine learning algorithm trained on the MIMIC-III dataset for cardiac health monitoring achieved the highest classification accuracy of 95% among 7 tested algorithms.
Why the study?
Cardiovascular diseases are the leading cause of death in Pakistan, creating a need for sustainable remote real-time continuous health monitoring to detect anomalous physiological behavior.
A Random-Forest machine learning algorithm trained on the MIMIC-III dataset achieved 95% accuracy in detecting abnormal human vitals, supporting the development of IoMT-based remote cardiac monitoring.
May support IoMT cardiac monitoring development; leaves open prospective validation beyond retrospective data.
The Internet of Medical Things (IoMT) has revolutionized the healthcare domain, with the introduction of remote real-time monitoring. This emerging technology has not only relieved the burden of hospital resources, but also paved the way for efficient monitoring and management of patients. According to World Health Organization (WHO), in Pakistan, cardiovascular diseases (CVD) are leading cause of deaths that amount to nearly 200,000 deaths annually. This results in high mortality rates all over Pakistan. To uplift the current architecture of healthcare in Pakistan, it is vital to develop a sustainable solution for continuous health monitoring and arrest anomalous physiological behavior before they become life-threatening. In the same pretext, this study proposes a working paper which aims to disseminate interim results of smart cardiac health monitoring using an amalgam of IoMT and Machine Learning (ML) techniques. The primary objective of the proposed research is to integrate state-of-the-art ML classification algorithms to detect, in near real-time, abnormal human vitals like electrocardiogram(ECG), heart-rate(HR), blood pressure(BP), etc. As network latency is critical to this application, therefore, to improve overall Quality of Service (QoS) of the system, we propose to fuse Edge Intelligence interfaced with multiple IoMT enabled bio-sensors. These sensors will form a body area sensor network(BSN) that records cardiac-related human vitals. In our preliminary research, that is presented in this paper, we trained several machine learning algorithms on the MIMIC-III clinical data-set and reviewed their performance. Among the 7 tested supervised classification algorithms, Random-Forest achieved the highest accuracy of 95% on the test set. Finally, to offer a remote patient management and monitoring panel, we developed an authenticated web-portal to inculcate data privacy and security in the proposed system.
No takes yet. Share an insight, caveat, or question.
Talha et al. (2022) studied Cardiovascular diseases. Random-Forest machine learning algorithm vs. Other supervised classification algorithms was evaluated on Classification accuracy on the test set. A Random-Forest machine learning algorithm trained on the MIMIC-III dataset for cardiac health monitoring achieved the highest classification accuracy of 95% among 7 tested algorithms.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: