MedClear — India's Dual-Track Healthcare AI Platform

MedClear homepage — dual-track selector for Patient and Clinic Staff, with India healthcare crisis stats and multilingual toggle.
Urgency meter hits ORANGE — "See Doctor Urgently" — with SEVERE diagnosis cards for Savitha's poorly controlled diabetes report.
Auto-matched PM-JAY ₹5 lakh coverage + Aarogyasri scheme in Telugu — surfaced instantly from the patient's diagnosis.
Insurance policy analyzer fires RED — Clause 3.1 excludes hemoglobin deficiency — IRDAI challenge letter generated instantly.
Submission-ready PA letter — Yashoda Hospitals header, HbA1c 11.2% cited, AP-MCI-2018-33421 in signature. Zero placeholders.
PA rejection appeal citing IRDAI regulations — Insurance Ombudsman escalation if insurer doesn't respond within 15 days.

Inspiration

My grandmother received a cancer diagnosis report last year. She held it for three days without understanding a single word. She didn't know what "Stage IIIB adenocarcinoma" meant. She didn't know her insurance could cover ₹5 lakhs under PM-JAY. She didn't know she had the right to appeal a rejected claim under IRDAI regulations.

She is not alone.

$$\text{500 Million Indians} \times \text{zero health literacy support} = \text{a crisis no one is solving}$$

Indian clinics waste $1\text{–}2$ hours per patient writing Prior Authorization letters — paperwork rejected $\approx 30\%$ of the time due to missing clinical detail. MedClear was built to fix this.

What it does

MedClear is India's first dual-persona healthcare AI platform — one app, two users, one complete patient journey.

🤒 Patient Track

Multi-turn symptom chat — AI maintains context across $n \geq 3$ exchanges to build a complete clinical picture
Plain-language diagnosis in English | हिंदी | తెలుగు — generated, not translated
Medical Urgency Meter — animated gauge from $[1, 10]$ with color zones:
- $[1,3]$ → 🟢 Monitor
- $[4,6]$ → 🟡 See Doctor Soon
- $[7,8]$ → 🟠 Urgent
- $[9,10]$ → 🔴 Emergency
Government Scheme Matcher — PM-JAY ($\leq$ ₹5L/family), CGHS, ESI, Aarogyasri
Insurance Policy Clause Analyzer — IRDAI-backed coverage verdict + challenge letters
Treatment Cost Estimator — realistic INR ranges with generic medicine savings (up to $70\text{–}90\%$ cheaper alternatives)
WhatsApp share — because $500M+$ Indians communicate on WhatsApp

🏥 Clinic Staff Track

Patient report → submission-ready PA letter with ICD-10 codes in $\leq 30$ seconds
Doctor MCI registration, hospital letterhead, clinical justification — dynamically assembled
PA Rejection Appeal Generator — cites IRDAI Circular No. IRDA/HLTH/CIR/GLD/013/02/2020, demands response within $t \leq 15$ days
Print / PDF / WhatsApp to patient directly ## How we built it Built entirely through MeDo's Deep Build — zero manual code.

MeDo Capability	How MedClear Uses It
Multi-turn conversation	Symptom chat maintains clinical context across 3+ exchanges
Full-stack generation	Entire app — frontend, AI inference, document assembly
Dynamic document generation	PA letters combine Stage 1 AI analysis + Stage 2 credentials
Multilingual AI	Telugu/Hindi generation with culturally specific content
Plugin integration	Government scheme data + PDF generation

Insurance Logic Engine:

$$\text{Policy Clause} + \text{Diagnosis} \xrightarrow{\text{MeDo AI}} \text{IRDAI-compliant Challenge Letter}$$

PA Letter Assembly:

$$\underbrace{\text{Stage 1 Analysis}}{\text{ICD-10, severity, lab values}} + \underbrace{\text{Stage 2 Credentials}}{\text{MCI no., hospital, insurer}} \rightarrow \text{Submission-Ready PA Letter}$$

Challenges we ran into

Challenge 1 — Modeling the real PA workflow correctly.
Prior Authorization must come from clinic staff, not patients. Early versions gave PA letters to patients — architecturally wrong. We built the dual-track system to solve this:

$$\text{Patient} \xrightarrow{\text{Track 1}} \text{Understand + Coverage Check}$$ $$\text{Clinic Staff} \xrightarrow{\text{Track 2}} \text{Generate + Submit PA Letter}$$

Challenge 2 — Multilingual medical accuracy.
Telugu explanations needed culturally specific dietary advice (పాలకూర, పప్పు, అల్లం) and correct state scheme names — not just translated English.

Accomplishments that we're proud of

Zero placeholder text in any generated PA letter
Cancer PA letter quality matching hospital billing departments — in $< 30$ seconds
IRDAI regulatory accuracy with actual circular citations
Complete production-quality product built in $\leq 48$ hours using MeDo

What we learned

MeDo's Deep Build is a serious full-stack generation engine. Treating prompts as product specifications — not commands — produces architecturally sound applications.

Real-world impact requires honesty in design. An app that misrepresents the PA workflow is worse than no app at all.