Sonarive

Relieve, receive, believe—Sonarive

Comment

Sonarive - AI-Powered Health Assistant

🌟 Inspiration

In a world of fragmented healthcare systems and growing mental health concerns, I was inspired to build Sonarive—a unified AI-driven platform that bridges diagnostic intelligence with personalized care.

My goal was to combine the power of Perplexity Sonar with real-world medical challenges to make early diagnosis, emotional well-being assessment, and treatment guidance more accessible—especially in underserved or remote areas. I envisioned a tool that empowers users and caregivers with insights that are often scattered across multiple appointments and opinions.

🚀 What it does

Sonarive is an AI-powered healthcare assistant that offers:

🏥 Smart Hospital Recommendation

  • Recommends nearby hospitals using user demographics (age, gender, location) and medical needs.

💊 Drug Research Assistant

  • Helps users understand drug purposes, side effects, alternatives, and dosages using live research via Perplexity Sonar.

🧾 Treatment Planner

  • Based on user-inputted symptoms and demographics, Sonarive provides potential diagnoses.
  • Leverages Perplexity Sonar for contextual reasoning to deliver early medical insights.

✅ Second Opinion on Treatments

  • Evaluates prescribed treatments and offers a second opinion based on trusted sources and research literature.

🧠 Mental Health Analysis (Key Feature)

  • Includes PHQ-9 and GAD-7 forms for emotional screening.
  • Combines questionnaire results with demographic and textual inputs to generate well-being insights and mental health suggestions.

🖼️ Scan Analysis (Key Feature)

  • Allows users to upload CT, X-ray, or MRI scans.
  • Uses image preprocessing models to detect anomalies and suggest possible diagnoses or next steps.

🛠️ How I built it

Frontend

  • Built with Next.js and Tailwind CSS for responsive, clean UI/UX.
  • Integrated form flows, result viewers, and image upload components.

Backend & APIs

  • Developed using Node.js and Express.
  • Integrated the Perplexity Sonar API for medical reasoning and decision support.
  • Used Google Gemini models for scan preprocessing and interpretation.
  • Integrated the Google Maps API to display hospital recommendations.
  • Implemented real-time PHQ-9 and GAD-7 scoring logic.

Deployment

  • Deployed the full-stack app using Vercel.

⚠️ Challenges I ran into

  • Scan Analysis Model Tuning: Since no medical image analysis was directly available through Sonar, I had to explore alternatives. Gemini provided a viable solution, but integrating it alongside Sonar introduced complexity in managing multiple APIs and their responses.

  • Combining PHQ-9/GAD-7 with LLM Reasoning: Making the output context-aware (age, region, user notes) required careful prompt engineering and validation.

  • Accuracy & Privacy: Balancing actionable results with anonymization and privacy was a priority, especially for scans and sensitive mental health data.

  • API Limitations: Some Perplexity Sonar features had rate limits or access restrictions, requiring workarounds and batching.

🏆 Accomplishments I’m proud of

  • Developed a functioning AI-powered mental health assistant using PHQ-9, GAD-7, and Sonar's reasoning capabilities.
  • Built a medical image diagnosis feature capable of giving actionable scan insights.
  • Created a seamless, user-focused UI for simplified healthcare navigation.
  • Successfully integrated and applied Perplexity Sonar in real-world health scenarios like second opinions and drug research.

📚 What I learned

  • How to blend structured screening tools with generative AI for early diagnosis and emotional health tracking.
  • Gained a deep understanding of prompt engineering for medical applications.
  • Learned to balance interpretability, accuracy, and privacy in sensitive healthcare data.
  • Developed skills in full-stack development combining frontend, backend, and AI API integration.

🔮 What's next for Sonarive

  • 🔐 End-to-End Encryption: To fully secure patient data and scans.
  • 🤝 Collaboration with Medical Experts: To validate recommendations and improve accuracy through human-in-the-loop.
  • 🧠 Multi-language Support: Make mental health tools accessible in regional languages.
  • 📱 Mobile App Launch: Broaden reach, especially in rural/low-infra areas.
  • 🩺 Teleconsultation Feature: Enable instant connection with healthcare professionals based on user analysis.

Built With

Share this project:

Updates