Physical Activity Detection for Diabetes Mellitus Patients Using Recurrent Neural Networks.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Metabolic Disorders Research

Background:

• Diabetes Mellitus (DM) management requires understanding physical activity's impact on blood glucose.
• Accurate detection and classification of physical activity are crucial for effective DM therapy.
• Continuous Glucose Monitoring Systems (CGMS) and Heart Rate (HR) signals offer potential for activity detection.

Purpose of the Study:

• To develop an Artificial Intelligence (AI) algorithm for detecting physical activity in patients with DM.
• To utilize combined CGMS and HR signals for identifying physical activity patterns.
• To enhance personalized management strategies for diabetes through activity recognition.

Main Methods:

• Development of an AI algorithm integrating CGMS and HR data.
• Application of multiple recurrent models, including Recurrent Neural Networks (RNNs).
• Evaluation of classifier performance using the area under the receiver operating characteristic curve (AUC).

Main Results:

• The AI algorithm demonstrated high efficacy in detecting physical activity.
• Recurrent Neural Networks achieved a best-ever performance of 0.99 AUC.
• The developed approach accurately analyzes physical activity in DM patients.

Conclusions:

• Recurrent Neural Networks provide a powerful and efficient solution for physical activity detection in DM.
• This AI-driven approach can significantly improve understanding of individual activity patterns.
• The research contributes to more personalized and effective diabetes mellitus management.