SwasthyaLogs AI
Streamlining Healthcare, One Note at a Time
Created on 7th September 2025
•
SwasthyaLogs AI
Streamlining Healthcare, One Note at a Time
The problem SwasthyaLogs AI solves
Problem:
Doctors and medical staff spend a lot of time manually recording patient visits, medications, vitals, and other clinical data. Traditional record-keeping is time-consuming, error-prone, and often delayed, which can lead to incomplete patient histories and inefficient care.
Solution:
SwasthyaLogs AI is an AI-powered, digital, audio-based medical record system that converts multi-lingual voice recordings into structured, real-time patient data. By automating data entry and integrating it with a dynamic frontend dashboard, SwasthyaLogs AI reduces paperwork, minimizes errors, and ensures instant access to patient records.
Link to the presentation - https://drive.google.com/drive/folders/115pWVk51FwjgneSr6oWaUnuIfWB1OP1X?usp=drive_link
Workflow:
- Doctors record patient conversation notes as audio and upload via Slack channel.
- n8n workflow detects new uploads and triggers automation.
- Audio files are securely stored in Google Drive.
- Audio file is passed to Elevenlabs and transcript is generated from its STT (Speech-to-text) model.
- The transcript is then fed to AI model (ChatGPT) to extract important information from it (E.g.: Basic Patient details, patient vitals like blood pressure, heart rate, doctor name, visit time, symptoms, medication suggested, dosage, summary, etc.)
- Supabase backend is updated with this patient information, visits, vitals, and medications, etc.
- The frontend portal reflects the updates dynamically in real-time for doctors and admins. They can access it by logging in the portal.
- Patient automatically receives a message with details like the summary of the visit, medication and dosage, etc.
Benefits:
- Saves significant time for medical staff.
- Improves accuracy and completeness of patient records.
- Provides real-time access to patient information.
- Enables better patient tracking, medication management, and visit history.
- Frees doctors to focus on patient care rather than documentation.
Tools Used:
- ElevenLabs – Multi-language AI audio processing.
- n8n – Workflow automation between Slack, Google Drive, and Supabase.
- Supabase – Real-time backend for database and storage.
- Google Drive – Cloud storage for audio and medical documents.
- Slack – Trigger platform for audio uploads.
- Rocket – AI tool used to generate the initial frontend structure.
- Windsurf – AI code editor used to refine and integrate frontend code with workflows.
Challenges I ran into
1. Website Template Idea Generation
- Challenge: As a robotics engineer, I didn’t have much experience with frontend development and was short on ideas for a professional UI.
- Solution: I used Rocket as a vibecoding tool to generate excellent templates for the website, which I later refined and customized to fit the SwasthyaLogs AI workflow.
2. Dynamic Frontend Integration
- Challenge: Initially, the Rocket-generated frontend was static and relied on cloud runtime scripts (rocket-web.js), making it impossible to run offline and integrate with Supabase.
- Solution: I removed all Rocket dependencies, refactored the code to be fully self-contained, and ensured proper ES6 imports for JavaScript modules using the AI-assisted Windsurf editor. All dynamic content now works using the Supabase JS SDK.
3. Real-time Data Synchronization
- Challenge: Patient, doctor, and visit data needed to update in real-time on the frontend.
- Solution: I implemented Supabase real-time subscriptions for all tables. This ensures that any change in the backend is immediately reflected on the frontend dashboard and patient views.
4. Handling Existing Records in Workflows
- Challenge: n8n workflows were inserting duplicate patients and doctors if they already existed in Supabase.
- Solution: I modified the workflows to check for existing records before insertion. I used conditional IF nodes to fetch existing data and pass it downstream, ensuring idempotent inserts while keeping the data flow intact.
5. Complex Data Structures & Expressions
- Challenge: Extracting, transforming, and displaying nested data (e.g., patient vitals, medications, summaries) in the frontend required multi-step operations.
- Solution: I used IIFE patterns in n8n expressions to handle array filtering, object transformations, and conditional logic cleanly. This enabled dynamic rendering of complex patient data without breaking workflows.
6. Local Deployment vs Subdomain Hosting
- Challenge: The site worked locally, but subdomain deployment caused caching issues and inconsistencies between preview links and the subdomain version.
- Solution: I implemented hard reload instructions, added a _headers file for cache control on Netlify, and used versioning query strings for JS/CSS to ensure the subdomain always loads the latest deploy.
7. Supabase Table & Schema Adjustments
- Challenge: Minor mismatches between frontend expectations and Supabase schema (e.g., visit_ts vs ts, missing photo column in patients).
- Solution: I updated frontend code to align with the current schema, added missing tables (patient_vitals, patient_medications), and ensured foreign key constraints were respected for workflow consistency.
8. Asset & Image Loading Issues
- Challenge: Images stored locally or on external URLs sometimes failed to load on the deployed site, causing 404 errors.
- Solution: I verified and defined all relative paths for assets, updated broken URLs, and ensured all images used proper CDN or local references for smooth display across devices.
Tracks Applied (3)
All Registered Participants
Windsurf
Vibecoding your Idea | Powered by Rocket.new
Rocket
Voice AI & Conversational Products | Powered by Elevenlabs
ElevenLabs
Discussion
Builders also viewed
See more projects on Devfolio