A review of machine learning and the internet of things for remote patient monitoring covering advances challenges and opportunities

As Machine Learning (ML) and the Internet of Things (IoT) have matured, their convergence has placed Remote Patient Monitoring (RPM) at the centre of personalized, cost-effective care. The present work reviews progress in ML and IoT based RPM, covering how ML is applied within RPM, the principal ML models employed, the IoT infrastructures that underpin it, the use of RPM across different patient populations, the challenges that arise, and the directions the field may take. After a PRISMA-guided screening of work indexed in Springer, IEEE, MDPI, ACM, and Elsevier, 101 peer-reviewed studies published from 2020 onward were retained for a qualitative (narrative) synthesis structured around five research questions. The review is narrative in form yet is carried out under a structured and transparent protocol. Within the studies examined, RPM was reported to enable real-time monitoring, cut the number of in-person hospital visits, and reduce the cost of patient-centred care, while IoT- and ML-based designs that built in privacy-preserving techniques strengthened data security; across these studies, deep learning and hybrid ML models often reported higher predictive accuracy than conventional approaches, though typically on particular datasets and at greater computational cost. Across the ten systems compared here, the eight that reported a single determinate figure gave accuracies with a median of 94.5% (range 92% to 98%) and latencies with a median of 265 ms (range 120 ms to 350 ms), and not one of them was evaluated at a site other than the one where it was developed, so these numbers describe individual study conditions rather than performance at population level. Among the obstacles that surface repeatedly are patient adherence, interoperability, scalability, and the protection of data. On balance, the body of evidence gathered here suggests that RPM built on ML and IoT can make monitoring safer, more flexible, and more anticipatory; reaching that level across whole populations, however, will call for sturdier systems, broader patient participation, and thorough clinical validation.