PhysioDimClassifier—imbalance data classifier model for IoMT-based remote patient monitoring systems

Remote patient monitoring systems (RPMS) using the Internet of Medical Things (IoMT) continuously collect and exchange periodic sensor-observations through communication modules. However, these data streams often contain relevant and irrelevant series, leading to imbalance issues in physiological disease assessment. This research introduces a PhysioDimClassifier (PDC), a novel model to detect and mitigate imbalanced data in physiological disease diagnosis. The proposed model identifies the likenesses and permanence within observation sequences, classifying them as normal or imbalanced based on monitoring duration and sensor communication time. A rotational tree classifier trackspermanence sequences, ensuring accurate classification of imbalanced data. By analyzingsequence interruptions, the model improves the retention of imbalanced data patterns, reducing misclassification. Experimental validation demonstrates that PDCM enhances data accuracy by up to 12.61 %, improves imbalance data detection by 13.23 %, increases classification rate by 10.98 %, lowers data imbalance by 11.22 %, and decreases assessment time by 10.5 %. These improvements contribute to timely and accurate physiological disease diagnosis in IoMT-based RPMS, optimizing clinical decision-making and patient outcomes. The proposed approach providesa robust, scalable, and efficient solution for handling imbalanced physiological data in real-time healthcare applications.

• •
  Introduces PhysioDimClassifier (PDC), a novel model to detect and mitigate imbalanced physiological data.
• •
  Employing a rotational tree classifier for sequence performance tracking and imbalance classification.
• •
  Enhances classification accuracy and reduces imbalance effects, ensuring improved disease diagnosis in IoMT-based RPMS.