Smart Patient Monitoring and Recommendation (SPMR) Using Cloud Analytics and Deep Learning

• Published in: IEEE access 2024, Vol.12, p.54238-54255
• Main Authors: Sundas, Amit, Badotra, Sumit, Shahi, Gurpreet Singh, Verma, Amit, Bharany, Salil, Ibrahim, Ashraf Osman, Abulfaraj, Anas Waleed, Binzagr, Faisal
• Format: Article
• Language: English
• Subjects: Ambient assisted living; ambient assisted living (AAL); Biomedical monitoring; Blood pressure; chronic diseases; Cloud computing; Deep learning; Diseases; Health care; healthcare; Internet of Medical Things; IoT; Machine learning; Medical services; Model accuracy; Monitoring systems; Patient monitoring; predictive deep learning; Predictive models; Remote monitoring; Remote patient surveillance; Telemedicine; Temperature sensors
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2024.3383533

The escalating global prevalence of chronic and lifestyle-related illnesses presents substantial societal and economic challenges. This work delves into an extensive review of healthcare monitoring systems tailored for chronic and lifestyle disorders. Subsequently, we propose a pioneering Smart Patient Monitoring and Recommendation (SPMR) framework, leveraging Deep Learning (DL) and cloud-based analytics. SPMR ensures continuous monitoring and predictive insights into a patient's authentic health status using data from vital signs and contextual activities collected via Ambient Assisted Living devices. Within the predictive DL component of the LIP module, we employ Categorical Cross Entropy (CCE) Optimization to forecast real-world health conditions using unbalanced datasets derived from Chronic Blood Pressure Disorder case studies. Significantly, SPMR's capability to deliver real-time preventive measures and treatments persists even without Internet or cloud connectivity. This circumvents the need to replicate Machine Learning (ML) models and associated procedures in local setups, thus streamlining operations. Comparative analysis against analogous models showcases the considerable effectiveness of our proposed model, notably enhancing accuracy by up to 8 to 18 percent. Moreover, both the overall F-score and the emergency class F-score exhibit marked improvements of 17% and 36%, respectively. These outcomes underscore SPMR's pivotal role, especially during crises, emphasizing its significance in healthcare monitoring systems.