Clinical judgments are typically reliant on the practitioners' experiences, with only a very limited amount of support from data-centric analytic methods from medical databases. This frequently results in unfavorable biases, human mistakes, and excessive healthcare expenses that influence the standard of attention given to patients.The application of intelligent technology in clinical decision making in the telehealth environment has recently started to play a vital role in enhancing the quality of patients' lives and lowering the expenses and workload associated with their ongoing medical treatment. Predicting chronic diseases is a crucial task in clinical decisions. Adopting EHRs to forecast the beginning of diabetes, one of the chronic illnesses could improve the efficiency and standard of medical care.Existing research use machine learning algorithms based on patient attributes to meet this need; however, they are biased and have high dimensional data issues.In this work, two novel techniques, the novel Convolution Neural Network based on SMOTE prediction and the Genetic Particle Swarm Cuckoo Optimization Model (GPSCOM), are proposed to address these issues.Several performance measures are utilized to analyse the prediction performance of this technique. The suggested deep learning-based approach outperforms conventional machine learning models when it comes to risk prediction tasks. The investigational finding demonstrates that the suggested approach offers a greater prediction effect compared to other conventional machineearning models.

Smarthealth;EHRs;Heart Disease, RPA;GPSCOM;and SMOTECNN