Dosing Walking Like a Drug: An AI Chatbot for Diabetic Foot Remission Following Limb Reconstruction
Summary: This protocol paper outlines a conversational AI chatbot designed to guide diabetic foot remission after wound healing or limb reconstruction by treating walking and therapeutic footwear as titratable “drugs.” The system integrates daily data (step counts via phone/wearable, paired foot skin temperature differences, shoe wear time, and symptom checks) to assign risk tiers (green/amber/red) and deliver personalized next-day guidance. Activity advances gradually (~500 steps/week) while gated by thermometry (hold/reduce if ≥2.2°C side-to-side difference). Footwear progression starts conservatively and escalates only with stable skin checks. Built on prior SmartBoot sensor work, the single-arm feasibility pilot (n=30) will assess engagement, safety, and implementation before larger trials. The approach aims to bridge the high-relapse gap (up to 40% within a year) through structured, data-driven home monitoring.
Key Highlights:
- AI chatbot titrates walking (~500 steps weekly increments) and footwear based on real-time thermometry and sensors
- 2.2°C temperature difference threshold triggers de-escalation and clinical escalation if persistent
- Addresses 40% recurrence risk with precise, responsive remission guidance
- Authors/key contributors: Lucian Feraru, David Klonoff, Bijan Najafi et al.
Keywords: AI chatbot diabetic foot, diabetic foot remission, walking dosing, Bijan Najafi