Name: Devender Kumar
Project Title: Investigating the role of context-awareness in improving ambulatory arrhythmia diagnosis
Group: Personalized Health Technology
Supervisors: Jakob Eyvind Bardram, Sadasivan Puthusserypady
Project Description:
Heart arrhythmias are the number one cause of strokes and cardiovascular-related deaths. The ECG analysis is the cheapest way to detect arrhythmias. However, in the early stage, it is difficult to catch them in a routine in-hospital ECG due to their paroxysmal nature. Thus it requires long-term ECG monitoring in the patient's natural setting. Due to many confounding noises and lack of understanding of the patient's context, the arrhythmias detection algorithms result in a large number of false-positives when applied under free-living conditions. These false positives, even if small, could lead to overdiagnosis and patient anxiety in longitudinal screening.
My thesis aims to investigate the role of context-awareness for improving ambulatory arrhythmia diagnosis algorithms under free-living conditions.
Perspective:
We investigate if specific user-contexts are (e.g., activities, body positions) triggering more false-positives under free-living ambulatory conditions; if so, then how to fine-tune the arrhythmia detection algorithms dynamically for those user contexts and reduce false positives rate. The final out will be an improved Deep Leaning-based solution that can be used for large-scale pre-emptive arrhythmia screening (atrial fibrillation in-particular) under free-living conditions.