AI Health Diagnosis Avatar for Kiosk/ Box/ Api
Budget: ₹150,000 – ₹250,000 INR
I’m creating an interactive AI avatar that can talk users through their symptoms and suggest the most likely illnesses behind them. The goal is not to replace a physician, but to give people an informed, friendly first step before deciding whether to seek professional care.
What I need from you
• A conversational engine that accepts free-text and Voice, asks follow-up questions, then maps answers against a clinically vetted symptom database ( Health LLMs ) to produce a probable diagnosis for conditions such as colds, flu, allergies, minor infections, stomach upsets, etc. and for severe disease refers to other online doctors.
• A 2D avatar front-end that lip-syncs and delivers the advice naturally. WebGL, Unity, or similar frameworks are fine as long as performance is smooth on desktop and Health Kiosks.
• A concise, plain-language summary for the user that explains the possible condition, typical next steps, and red-flag signs that should trigger a real doctor visit.
• Need complete software for Health Kiosk/ Box ( Windows based) and box and api to access this from Android
Acceptance criteria
1. User talks to Avatar and tells symptoms in everyday language ( Hindi/ English); the avatar responds in under one second.
2. Suggested conditions are limited to common illnesses and must include confidence scores or similar transparency.
3. All medical text is explainable at about an eighth-grade reading level and carries a “not a substitute for professional advice” disclaimer.
4. Code passes my security review (no hard-coded keys, OWASP top-10 safe) and installs locally with a one-command deploy script.
5. Diagnosis's is reviewed by remote doctor at background who can edit LLM results so AI can learn
Workflow :-
User arrives at kiosk
↓
AI chatbot asks symptoms/ if user just wants some tests ( voice )
↓
AI recommends tests based on symptoms
↓
User measures BP, SpO2, temp etc ( all devices will we provided to users via kisosk e.g Cloudclinic)
↓
AI analyzes results
↓
AI gives health report and suggestive diagnostics + medicines advice ( which is reviewed by remote doctor)
↓
Optional doctor consultation( video consultation)
↓
Users exits and test report, medical history, diagnostics etc are send to user phone
If you’ve worked with symptom-checker datasets, clinical NLP models, or avatar rigs before, tell me how you’d approach training, bias control, and ongoing updates. Looking forward to seeing your ideas.
What I need from you
• A conversational engine that accepts free-text and Voice, asks follow-up questions, then maps answers against a clinically vetted symptom database ( Health LLMs ) to produce a probable diagnosis for conditions such as colds, flu, allergies, minor infections, stomach upsets, etc. and for severe disease refers to other online doctors.
• A 2D avatar front-end that lip-syncs and delivers the advice naturally. WebGL, Unity, or similar frameworks are fine as long as performance is smooth on desktop and Health Kiosks.
• A concise, plain-language summary for the user that explains the possible condition, typical next steps, and red-flag signs that should trigger a real doctor visit.
• Need complete software for Health Kiosk/ Box ( Windows based) and box and api to access this from Android
Acceptance criteria
1. User talks to Avatar and tells symptoms in everyday language ( Hindi/ English); the avatar responds in under one second.
2. Suggested conditions are limited to common illnesses and must include confidence scores or similar transparency.
3. All medical text is explainable at about an eighth-grade reading level and carries a “not a substitute for professional advice” disclaimer.
4. Code passes my security review (no hard-coded keys, OWASP top-10 safe) and installs locally with a one-command deploy script.
5. Diagnosis's is reviewed by remote doctor at background who can edit LLM results so AI can learn
Workflow :-
User arrives at kiosk
↓
AI chatbot asks symptoms/ if user just wants some tests ( voice )
↓
AI recommends tests based on symptoms
↓
User measures BP, SpO2, temp etc ( all devices will we provided to users via kisosk e.g Cloudclinic)
↓
AI analyzes results
↓
AI gives health report and suggestive diagnostics + medicines advice ( which is reviewed by remote doctor)
↓
Optional doctor consultation( video consultation)
↓
Users exits and test report, medical history, diagnostics etc are send to user phone
If you’ve worked with symptom-checker datasets, clinical NLP models, or avatar rigs before, tell me how you’d approach training, bias control, and ongoing updates. Looking forward to seeing your ideas.