Enhanced K-Nearest Neighbors for Smart Cardiovascular Disease Prediction in IoT System

Cardiovascular disease (CVD) is becoming more prevalent as a health issue and ranks as a top cause of mortality globally.Effectively identifying CVD is frequently a challenging process, given that minor errors can result in significant consequences.To address this challenge, healthcare organizations have recently embraced Internet of Things (IoT) to collect patients' vital signs using wearable sensors, these data are then stored and transmitted to machine learning (ML) based prediction systems.Among the various ML algorithms, the K-Nearest Neighbors (K-NN) algorithm stands out for its simplicity and effectiveness in CVD prediction.However, its reliance on majority voting can lead to classification errors, especially when test vectors are closer to minority class neighbors.To address this limitation, we propose the Enhanced K-Nearest Neighbors (E-KNN) algorithm, specifically designed to refine classification accuracy by incorporating a weighted distance measure that considers both neighbor proximity and class distributions.The E-KNN model has undergone comprehensive testing in comparison to standard ML methods.The experimental findings demonstrate that the introduced model surpasses current methodologies based on performance assessment indicators, recording a 91.43% notable accuracy level.To leverage the E-KNN algorithm, we have developed an IoT platform that gathers crucial patient data and transmits it to the E-KNN based model.