Medical diagnosis decision-making framework on the internet of medical things platform using hybrid learning

There are no obvious clinical signs of a silent disease until it causes irreversible damage. As computerized technologies become more common for early detection of ailments, severity can be minimized. In order to minimize the severity of ailments, computerized technologies are being used more and more commonly for early detection through internet of things (IoT) platforms for e-Health, sometimes called the "internet of medical things (IoMT)”. Nevertheless, unpredictability, inapplicability, and instability were problems with previous automated decision-making models. In this study, we present a hybrid model that combines optimum learning and iterative neighborhood component analysis (iNCA) by relying on neighborhood component analysis and feature aggregation. The support vector machine (SVM) algorithm has shown promising results when classifying numerous diseases based on water cycle algorithms (WCA). By using the WCA method, it is possible to find effective parameters at the local and global levels. In addition, the features with the lowest error level are selected from the pool of features. Hence, we developed a procedure to improve diagnostic accuracy and avoid overfitting. Using the IoMT platform, we validated the method on diabetes, hepatitis, breast cancer, and dermatology data from the UCI database. Moreover, we compared the proposed strategy with the state-of-the-art and thus, our decision-making system performed better than similar methods in identifying silent diseases. The proposed approach combines a model-oriented hybrid design with IoMT platform to assign proper treatments to patients and be both clinically applicable and appropriate for computer-aided design.