Latha Palaniappan

One of the costliest complications of diabetes is the development of neuropathic foot ulcers which, if infected, can lead to amputation and death. Ulcers are preventable however, current methods for assessing ulcer risk are time-consuming, costly and require expertise to interpret the results. Thus, screening for foot ulcers is not performed optimally in low resource clinics. However, new literature indicates that ulcers can be detected up to 6 weeks before they form by identifying inflamed, pre-ulceration through temperature monitoring. Cost effective and scalable technology for early detection of pre-ulcers is urgently needed, as timely interventions can reduce ulcer incidence 3-fold.

Our team has developed a deep-learning algorithm to classify pre-ulcers from the foot thermal images alone ($300 camera) and have found promising results with the same algorithm trained on normal phone camera images (negligible cost). To confirm this novel finding and to lengthen the current prediction window to detect pre-ulcers (>6 weeks), we are proposing an exploratory pilot to prepare for a 2-year longitudinal study in India. In this exploratory phase, we would conduct user research to understand if any current assumptions regarding implementation of the technology would compromise pilot study success and future implementation.