03
औषधम्·owsh · aa · dham
Your first doctor's visit. Before the doctor.
Overview
Aushadham is an AI healthcare application that replicates the diagnostic logic of a first OPD consultation. You enter what you're experiencing. The system connects symptom nodes, asks targeted questions to eliminate possibilities, and arrives at the most probable condition — then recommends home remedies or doctor consultation. It does not prescribe.
Type
Healthcare AI
Status
Concept · Built
Stack
Next.js · AI/ML
The first general OPD appointment is often a triage exercise — a doctor narrowing down a large possibility space with targeted questions. That logic is learnable, repeatable, and doesn't require a physical visit. Most people either over-Google and panic or under-investigate and delay. Aushadham is the middle path.
How the symptom graph works
A linear flow asks every question regardless of prior answers. A symptom graph only activates nodes that are connected to confirmed symptoms — exponentially reducing question count while increasing accuracy. The architecture mirrors how a clinician actually thinks.
The line between recommendation and prescription is legally and ethically significant. We drew it clearly and built the system so it cannot be crossed — outputs are either home remedies or doctor referral, with explicit confidence scores.
The system shows its uncertainty. A result with 60% confidence is displayed differently from one with 90%. Calibrated uncertainty is more useful than false confidence.
Medical edge cases broke the node graph in unexpected ways — symptoms that are common to both benign and serious conditions created false confidence. The system needed a 'escalate immediately' path that bypassed the full diagnostic flow for red-flag symptom combinations.
The symptom graph needs a separate urgency layer that runs in parallel — not sequential. Some symptom combinations should short-circuit the entire diagnostic flow and go straight to 'see a doctor now.'