PharmGuard Ai

AI clinical safety layer analyzing drug packaging photos and patient context to flag contraindications and dosage risks.

Comment

Inspiration

Medication errors, unintended drug interactions, and improper dosing are major contributors to preventable health complications. Patients often struggle to read complex prescription labels or cross-reference active ingredients against their specific conditions. PharmaGuard AI was created to provide an instant, accessible clinical safety layer that empowers individuals to check their medications before consumption.

What it does

PharmaGuard AI analyzes images of medication packaging or prescriptions alongside patient-specific health context (age, pregnancy status, symptoms, and known allergies). Using multimodal vision models, it extracts active ingredients, flags critical contraindications, and outputs a clear safety badge (SAFE, CAUTION, or DANGER) with actionable clinical guidance.

How we built it

  • UI/UX Design: Built using Lovable with a clean clinical design system.
  • AI Core: Integrated Google Gemini 3.6 Flash for multimodal vision OCR and clinical reasoning.
  • Structured Data Processing: Prompt-engineered the AI backend to parse health inputs and return structured JSON schemas for real-time UI rendering.

Challenges we ran into

Balancing ultra-low latency with reliable extraction of dense packaging text from low-light mobile photos, while ensuring the output remains clinically structured and easy for everyday patients to understand.

Accomplishments that we're proud of

Successfully executing end-to-end multimodal OCR and context-aware safety evaluation that returns structured status badges and clear contraindications in under two seconds.

What we learned

Optimizing vision prompts to output strict JSON schemas vastly improves UI stability and eliminates raw text parsing errors in frontend components.

What's next for PharmaGuard AI

  • Integrating Paystack/Stripe micro-payments for per-audit transactions.
  • Expanding multi-drug interaction checks for patients taking multiple daily medications simultaneously.

Built With

Share this project:

Updates