MedLens AI

Because living alone shouldn't mean dying alone

Comment

Inspiration

There is a silent tragedy unfolding around us: 80 million people live alone globally, facing daily, life-threatening medication risks. Through personal observation and research, we realized that isolated individuals—especially the elderly—face three critical dangers:

  1. Medication Errors: Complex prescriptions leading to fatal drug interactions.
  2. The Isolation Crisis: Health emergencies occurring with zero human oversight.
  3. The Accessibility Barrier: Current medical apps are too complex for elderly patients to navigate.

We realized that current solutions are just "passive reminders." We wanted to build an active, life-saving intervention. This led to the birth of MedLens AI, driven by a single core belief: Because living alone shouldn't mean dying alone.

What it does

MedLens AI is an offline-first, smart care companion acting as a vigilant pharmacist and emergency responder. Our 4 core features prevent deaths before they happen:

  • Smart Scanning: AI-powered instant cross-referencing detects fatal drug interactions and allergy conflicts in real-time.
  • Elderly Mode: A dedicated UI that auto-enlarges, increases contrast, and uses Text-to-Speech (TTS) Voice Guidance in 5 languages.
  • Auto-Emergency SOS: A proactive system that triggers automated emergency contact alerts if a user misses critical doses and fails their daily safety check-in.
  • Native Sync: Generates .ics calendar files to rely on the device's native alarm system, ensuring reminders work even if the app is closed.

How we built it

We engineered a Zero-Latency, Enterprise-Grade Architecture focused completely on privacy and offline capability.

  • Frontend: We used React.js and Tailwind CSS to build a highly responsive, component-driven UI that seamlessly transitions between standard and "Elderly Mode".
  • Accessibility Engine: We integrated the native Web Speech API to enable dynamic, multilingual synthesis without needing external cloud calls.
  • Data Fortress (Storage): We utilized HTML5 LocalStorage to keep sensitive medical profiles (allergies, schedules) strictly on-device.
  • Client-Side AI & Logic: The continuous risk assessment model runs locally. We engineered a real-time risk evaluation algorithm that calculates the danger level of a scanned medication based on historical data. Conceptually, the risk factor is calculated as: $$Risk_Level = \sum (Interaction_Severity \times Patient_Vulnerability) + Time_Since_Last_Dose$$ If $Risk_Level$ exceeds the safe threshold, the UI immediately flags the scan with a high-priority red alert.

Challenges we ran into

  • The "Offline-First" Mandate: Relying entirely on client-side processing without calling external databases for drug interactions was difficult. We had to structure our local data effectively to ensure instant cross-referencing without lagging the browser.
  • Designing for the Elderly: It's surprisingly challenging to build a UI that is large and accessible without making it look "childish" or poorly designed. Balancing Tailwind utility classes to dynamically scale everything (padding, fonts, icons) via a single elderlyMode state required meticulous styling.
  • Simulating Emergencies Safely: Testing an SOS system without actually spamming emergency contacts was tricky. We built a custom "Judges Demo Panel" to safely simulate missed check-ins and emergency countdowns during the pitch.

Accomplishments that we're proud of

  • Successfully creating a 100% offline Data Fortress that respects patient privacy while still providing AI-level insights.
  • Implementing a flawless Elderly Mode that completely transforms the UX with a single tap, accompanied by reliable voice guidance.
  • Building a system that doesn't just wait for the user to act, but actively takes action (Auto-SOS) if the user is incapacitated.

What we learned

  • Accessibility (a11y) is not an afterthought: Building for visual and motor impairments fundamentally changed how we approach UI/UX design. We learned the deep value of high-contrast ratios and large tap targets.
  • The Power of Local Computing: We discovered how capable modern browsers are at running complex logic and text-to-speech engines entirely on the client side, eliminating server latency.

What's next for MedLens AI

Our Hackathon MVP is just Phase 1. Our roadmap to becoming a global healthcare standard includes:

  1. Vision Intelligence: Integrating Google Cloud Vision API for real-time OCR scanning of physical pill bottles and difficult-to-read prescription labels.
  2. Family Ecosystem: Deploying a Firebase cloud database to enable a secure "Family Dashboard," where doctors and relatives can remotely monitor medication adherence.
  3. Wearable Integration: Connecting with smartwatches to monitor real-time vital signs, creating a predictive early-warning system before a crisis even hits. Inspiration

Built With

Share this project:

Updates