Voice-first health interfaces for low-literacy users
Open research: speech-to-intent for Indian-language health queries, accent robustness, and where voice can lower the threshold for elderly and low-literacy patients.
Typing is a barrier for elderly users, low-literacy users, and anyone speaking a non-English language as their first.
The opportunity
Voice-first interfaces can reach populations that text-based AI never will. Indian-language ASR has improved dramatically with multilingual models, though accent robustness for code-mixed speech (Hinglish, Tamlish, etc.) remains imperfect.
Specific challenges for health
Medical terminology is mixed (English drug names, Hindi symptom descriptions). Numbers (HbA1c values, dosages) need precise transcription. Misunderstanding a single digit can be dangerous.
Where DiaCare is heading
Voice input on mobile is on the roadmap. We will roll it out conservatively, with strict confirmation of any numerical input before it is used to generate advice.