Efficient Algorithms For Diabetic Prediction

Abstract

Many people around the world have diabetes, a serious health condition affecting metabolism. Each year, the number of cases is increasing rapidly. Diabetes can harm vital organs, leading to dangerous complications if not treated promptly. Detecting diabetes early is crucial to prevent serious problems. Heart rate variability (HRV) data, obtained from electrocardiogram (ECG) readings, can help diagnose diabetes without invasive procedures. This study shows how deep learning methods can distinguish between normal and diabetic HRV data. We use advanced techniques like convolutional neural networks (CNNs), long short-term memories (LSTMs), and combinations of both to analyze complex patterns in the HRV data over time. We then use a support vector machine (SVM) to classify the data based on these patterns. Comparing our current work to previous methods that didn't use SVM, we found that using CNN improved performance by 0.03%, while CNN-LSTM combination improved it by 0.06%. With our suggested approach, doctors could diagnose diabetes using ECG data with a success rate of 95.7%.