whatswrongwithyourmitochondria/smol_med_helper

0

stars

88

commits

Python

primary language

Jun 15, 2026

updated

README


title: Voice Health Companion sdk: gradio sdk_version: 6.17.3 python_version: "3.12" app_file: app.py license: apache-2.0 tags:

  • build-small-hackathon
  • health
  • voice
  • accessibility
  • openbmb tracks:
  • backyard badges:
  • offbrand
  • tiny
  • demo

Voice-First Health Companion

A voice-first health companion for a stroke survivor with low vision and one impaired hand. All models are ≤4B parameters and run entirely on-device — no cloud LLM calls.

What it does:

  • Daily spoken check-ins (no typing required)
  • Camera → OCR: reads medicine boxes, letters, device readings aloud
  • Generates a change-focused doctor brief before appointments
  • Reads the brief back by voice for confirmation

Models (all ≤4B — qualifies for Tiny Titan):

RoleModelParams
Vision / OCRMiniCPM-V 4.6 (OpenBMB)~1.3B
Text / briefMiniCPM3-4B (OpenBMB)4B
Speech-inWhisper small~244M
Speech-outKokoro-82M82M

Safety: The model never diagnoses and never advises on doses or interactions. All symptom patterns surface as questions for the doctor.

Built for the HuggingFace "Build Small" hackathon — Backyard AI track · OpenBMB prize track.

Contributors

whatswrongwithyourmitochondria/smol_med_helper

0

stars

88

commits

Python

primary language

Jun 15, 2026

updated

README


title: Voice Health Companion sdk: gradio sdk_version: 6.17.3 python_version: "3.12" app_file: app.py license: apache-2.0 tags:

  • build-small-hackathon
  • health
  • voice
  • accessibility
  • openbmb tracks:
  • backyard badges:
  • offbrand
  • tiny
  • demo

Voice-First Health Companion

A voice-first health companion for a stroke survivor with low vision and one impaired hand. All models are ≤4B parameters and run entirely on-device — no cloud LLM calls.

What it does:

  • Daily spoken check-ins (no typing required)
  • Camera → OCR: reads medicine boxes, letters, device readings aloud
  • Generates a change-focused doctor brief before appointments
  • Reads the brief back by voice for confirmation

Models (all ≤4B — qualifies for Tiny Titan):

RoleModelParams
Vision / OCRMiniCPM-V 4.6 (OpenBMB)~1.3B
Text / briefMiniCPM3-4B (OpenBMB)4B
Speech-inWhisper small~244M
Speech-outKokoro-82M82M

Safety: The model never diagnoses and never advises on doses or interactions. All symptom patterns surface as questions for the doctor.

Built for the HuggingFace "Build Small" hackathon — Backyard AI track · OpenBMB prize track.

Contributors

Languages

Python

100.0%