Intel-AI MediLocker

Architecture of the whole concept behind IntelAI MediLocker

Inspiration

India’s healthcare system faces a dual challenge — patients often carry handwritten records that are prone to loss and damage, and doctors are forced to rely on incomplete histories, especially in emergencies. Moreover, follow-up care like diet guidance, dosage tracking, or mental health support is often missing once a patient leaves the hospital.

We wanted to build a solution that works like a digital companion for both patients and doctors — storing medical records, enhancing care with AI, and simplifying appointment workflows. With chronic illnesses like diabetes, hypertension, and thyroid on the rise, a long-term digital health assistant that travels with the patient, across doctors and cities, is the need of the hour.

Intel-AI MediLocker is our response — a smart, secure, and personalized mobile health platform tailored to Indian needs and scalable for global use....Our team envisioned an all-in-one AI-powered healthcare assistant that offers digital medical history storage, smarter doctor-patient interaction, and enhanced decision-making through intelligent features like side-effect prediction and AI-generated diet plans based on region and personal preferences.

What it does

Intel-AI MediLocker is a comprehensive, AI-integrated mobile platform divided into three portals: Patient, Doctor, and Admin.

Patient Features:

Medical History Timeline: Stores encrypted patient records (diagnosis, prescriptions, medicines, visit date) with a downloadable PDF prescription.

Appointment Booking: Patients can browse doctors filtered by specialization and availability and book appointments. Dashboard clearly shows tracking appoinments.

AI-Powered Diet Recommendation: Suggests a personalized diet plan based on: Region (Tamil Nadu, Punjab, etc.) Diet type (veg/non-veg) Current critical condition (e.g., diabetes, BP) E.g., A diabetic vegetarian from Tamil Nadu gets Ragi Dosa, Bitter Gourd Juice, and low-glycemic fruits.

SymptomDoc: Users can type symptoms (e.g., chest pain, dizziness) and get matched to the appropriate specialist using NLP and keyword mapping.

Telemedicine: Secure video/audio consultations with doctors are enabled inside the app. After consultation, prescriptions are updated in real time.

Dosage Tracker: Converts prescribed dosage into reminders. E.g., “Take 1 tablet of Metformin at 8 AM.” Users get in-app and system notifications.

Mental Health Chatbot: AI-powered chatbot for emotional support. Messages are encrypted and stored via hashing. If risky or concerning behavior is detected, the chatbot escalates the case to a real doctor for follow-up.

Lab Reports Upload: Patients can upload blood reports, scans, X-rays in PDF/image format. Doctors can preview and download. Reports are tagged for easy filtering (e.g., CBC, ECG, Thyroid).

Doctor Features:

Assigned Patients via OTP Verification: Doctors access patient history only after OTP verification sent to the patient’s registered number. Ensures privacy and consent.

View Medical History Summary: AI-generated summaries of major conditions (e.g., “Chronic diabetes, two infections in last year”) help doctors save time

Update Medication: Doctors enter: Diagnosis Medicines Prescription Notes Dosage timings System generates digital PDF and updates patient timeline.

Walk-in Patients: For physical visits without prior booking, doctor can manually enter patient’s Unique ID → OTP sent → access granted → update prescription.

Side Effect Prediction: --Before prescribing new medicine, doctor can check: --If the medicine is safe for the patient based on history --Potential allergic reactions, drug interactions --AI-suggested safer alternatives

Appointment Tracker: View upcoming bookings with patient details.

Patient Analytics: Shows: Total number of patients seen , Most common illnesses, Returning vs new patients, Graphs for monthly trends

Admin Features:

Doctor Verification: When a doctor registers, their details (license number, name, contact info) are submitted. The admin checks against national registry websites (like NMC India) and either accepts or rejects. Only verified doctors are added to the platform, ensuring patient safety and credibility.

How we built it

Flutter (Frontend): Cross-platform UI using Dart. Modular components for doctor/patient/admin dashboards. Navigation routes manage screen transitions.

Supabase (Backend): Authentication: Email/password login for all users Database: PostgreSQL hosted by Supabase for storing encrypted records API: Supabase REST endpoints for querying, inserting, and updating data Row-Level Security policies ensure patient privacy

Google Gemini API (AI): -> Mental health chatbot: Responds to queries, detects emotional distress by sending emergency help to specified doctor for a patient if any threat is detected. -> SymptomDoc: Matches input symptoms to medical specialization

RAG model : Diet AI: Recommends meals by mapping region-food-illness and based on personalized medical history with personalized preferences .

PDF Generation: ReportLab generates professional digital prescriptions with images and dosage tables.

Encryption: All sensitive fields (diagnosis, medicine names, messages) are encrypted with Fernet encryption.

OTP System: Custom module using Supabase and third-party SMS API (in production version) to verify patient identity before doctor can access history.

Challenges we ran into

AI Interpretation: Balancing overly generic responses and overly specific recommendations from the AI was difficult. Prescription Formatting: Getting dynamic medicines with varying dosage timings into a neat PDF format required complex table management. Multi-role Handling: Designing separate flows for patient, doctor, and admin — while reusing components — took planning and conditional logic. Security: Managing end-to-end encryption without degrading performance and ensuring only authorized access to data. Feature Prioritization: With 10+ features in the pipeline, scoping for the demo videos was tricky.

Accomplishments that we're proud of

->Built a fully functional app with secure medical data storage and AI integration -> Successfully predicted side effects and safer alternatives based on personalized history -> Delivered OTP-based secure access for privacy and consent -> AI diet plan recommendation based on personalized medical history with specified regional and personalized preferences -> Chatbot, and dosage reminders — all implemented with contextual logic -> Designed and implemented digital prescriptions with encrypted PDF generation

What we learned

-> How to structure large-scale mobile apps with multiple user types (patient, doctor, admin). -> Handling sensitive data with encryption and privacy-first design. -> Building with Supabase and using it beyond traditional Firebase-based apps. -> Implementing generative AI in real-world, domain-specific applications. -> Managing state, navigation, and modularization in Flutter apps -> Balancing innovation with regulatory realities (e.g., data privacy, medical verification)

What's next for Intel-AI MediLocker

-> Integration with Government Health APIs for real-time license verification. -> QR-based login and prescription validation for pharmacies. -> AI Summarizer for entire medical history in plain language (for non-technical users). -> Gamified medication reminders to improve adherence. -> Community support forums & multilingual AI assistants. -> Launching iOS version & expanding to wearable integrations (e.g., Fitbit sync).