DMSense: A non-invasive Diabetes Mellitus Classification System using Photoplethysmogram signal

• Published in: 2017 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops) pp. 71 - 73
• Main Authors: Reddy, V. Ramu, Choudhury, Anirban Dutta, Deshpande, Parijat, Jayaraman, Srinivasan, Thokala, Naveen Kumar, Kaliaperumal, Venkatesh
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-03-2017
• Subjects: Classification; Diabetes; Diabetes Mellitus; Diseases; Feature extraction; Feature Selection; Heart rate variability; HRV; Mobile Computing; Monitoring; Pervasive Healthcare; Photoplethysmogram; Servers; Support vector machines
• DOI: 10.1109/PERCOMW.2017.7917526

The alarming statistics of Diabetes Mellitus (DM) Type 2 as the most common and prevalent disease in India and world over [1] has fuelled research in the direction of non-invasive and continuous monitoring of this disease. This paper describes a demonstration of an inexpensive mobile-phone based android application which can collect Photoplethysmogram (PPG) from fingertip via built-in camera and flash and transfer it to a high-end cloud server for early detection of DM. Additionally, this application allows continuous monitoring of DM patients can aid in assisting the short and long-term complication risks. The proposed application is targeted to cater to the inherent demand to for a mobile-based, pervasive system for continuous, non-invasive monitoring and detection of DM. Our application has been successfully deployed on Nexus 5 and tested on controlled and diabetic group with 80% specificity and 84% sensitivity for a 100 patient dataset and presented in this paper.