Non-invasive detection system for hypoglycemia and silent heart attack in diabetes patients

Abstract

Hypoglycemia is a serious complication in diabetes patients, especially for those on insulin treatment. It is typically characterized by abnormally low sugar levels in the body. The common symptoms of hypoglycemia are found to be heart palpitations, anxiety, excessive sweating, shakiness, fatigue, and tremors. In some serious cases, it might lead to seizures or death. Sometimes, intense insulin treatment can lead to hypoglycemia unawareness. It is frequent in people having diabetes for a long duration or those having frequent hypoglycemic episodes. This complication results in the failure of detecting the early symptoms/warnings of low blood sugar levels. Feature extraction is performed on the ECG signal to extract the mean averaged ST level, heart rate and QRS complex. The frequency of the accelerometer signals is extracted. Around 195 ECG recordings from standard data sets, of multiple categories, were tested for validating the system developed. It is observed that only 1.03 percentage of false positives recorded by the system. The design of the proposed hypoglycemia detection system is especially for people of high age group who are highly prone to hypoglycemia and helpless at the same time. The proposed system aims to detect such episodes for them and alert the caregiver.