Predicting Chronic Kidney Disease and Coronary Heart Disease Risks in NIDDM Patients Using CNN-Adam: A Deep Learning Approach

Non-insulin-dependent diabetic mellitus (NIDDM) is a complicated metabolic disease that can lead to numerous consequences, such as kidney failure (CKD), heart attack and stroke (CHD), nerve damage, retinal detachment, and others. In this study, the adaptive momentum optimizer (CNN-Adam) improves a convolutional deep learning model to find out how likely it is that people with non-insulin-dependent diabetes mellitus (NIDDM) will get chronic kidney disease (CKD) and coronary heart disease (CHD). The study utilizes electronic medical records (EMRs) to extract pertinent demographic, clinical, and diagnostic properties, which CNN-Adam then analyses to improve predictive accuracy. Experimental validation on benchmark medical datasets demonstrates the model's effectiveness, evaluated through key performance metrics such as accuracy, precision, recall, and Fl-score. The findings underscore the potential of CNN-Adam in improving risk prediction and clinical decision-making for CKD and CHD in NIDDM patients. We utilized benchmark clinical data for the experimental verification, and validated the results across multiple parameters. The suggested system attained a peak precision of 89%, a recall of 88%, a specificity of 86%, an accuracy of 88%, and an F-score of 88%. The experimental results demonstrated that the CNN-Adam model outperforms other techniques for prediction capabilities.