mediQuack

mediQuack

Comment

Inspiration

Also, to bridge the gap between patients and doctor, we implemented email feature using SMTP library for patient users to send their standardized BPM graph to the email that user inputs. This can be doctor and medical providers' email address so that they promptly receive much needed medical records such as BPM graph.

We want you to be calm and de-stressed, DigitalOcean Gradient AI will select meditation song just for you, let's all keep our calm and mediate and keep our breathing and pulse under control!

mediQuack was born from the idea that your health vitals should be as verifiable as a financial transaction, allowing users to "anchor" their physical well-being to a permanent record.

What it does

mediQuack is a mobile health ecosystem that:

Captures & Streams: Measures Heart Rate (BPM) and Breathing Rate (RPM) via the device camera.

Warehouses: Streams raw biometric data into Snowflake for long-term, secure storage.

Analyzes: Uses Snowflake Cortex AI to generate personalized wellness insights directly from the data warehouse.

Visualizes: Generates high-fidelity, dual-axis vitals reports via a Python backend and emails them to the user.

Anchors: Commits a cryptographic summary (averages) of vitals to the Solana Devnet using the Memo program, creating an immutable "Vital Sign Certificate."

Meditation: We have 3 copyright-free meditation songs, Digital Ocean Gradient AI will look at your vitals data and recommend one song and play it! Stay calm and de-stressed with meditation sounds!

How we built it

iOS App: Built with SwiftUI and Swift Charts. We utilized a custom camera-processing engine to detect pulse and respiration.

Backend: A FastAPI server running on Python.

Data: Snowflake serves as our primary source of truth. We used the Snowflake Python Connector to handle high-frequency vitals ingestion and retrieval.

AI: Snowflake Cortex (Mistral-Large) processes vitals to provide coaching insights without the data ever leaving the Snowflake security perimeter.

Blockchain: Solana (Solders/Solana-Py) allows us to publish on-chain proof of health data with sub-second finality and near-zero fees.

DevOps: Tunneling via ngrok to connect the physical iOS device to our local development environment.

MLH Track - Solana

mediQuack utilizes the Solana Devnet to create a decentralized "Trust Layer" for health data. While Snowflake provides the storage, Solana provides the proof. We implemented a custom anchoring system that calculates a cryptographic summary of a user's biometrics and commits it to the blockchain via the Solana Memo Program.

By leveraging Solana’s sub-second finality and near-zero transaction fees, mediQuack makes it economically viable for users to "stamp" their health records onto a public ledger. Every vitals scan generates a unique transaction signature, allowing patients to provide doctors with immutable, timestamped proof of their heart and breathing rates that cannot be retroactively altered or forged.

MLH Track - Presage

mediQuack is a "Human Sensing" application at its core. We turned the standard smartphone camera into a sophisticated biometric sensor that captures Contactless Heart Rate (BPM) and Respiration Rate (RPM).

Our implementation focuses on the "Human Sensing Layer" by extracting subtle color and movement changes in the user's skin and chest to monitor vitals without the need for wearable hardware. By digitizing the human body's physical signals and streaming them directly into the data cloud, we are demonstrating a future where healthcare is accessible, non-invasive, and integrated into our daily digital interactions.

MLH Track - DigitalOcean

To power the mediQuack ecosystem, we utilized DigitalOcean’s high-performance infrastructure to host our specialized Python backend. Our FastAPI server, which orchestrates the complex logic between the iOS app, Snowflake, and Solana, is deployed as a robust service on DigitalOcean.

Furthermore, we leveraged DigitalOcean's Gradient™ AI tools (Llama LLM) to provide chatbot feature along with Cortex AI, and also Gradient AI recommends meditation song for you based on your breathing and pulse rate, let's all meditate and calm down.

MLH Track - Snowflake API

mediQuack pushes the boundaries of the Snowflake Data Cloud by treating it not just as a warehouse, but as an intelligent engine. We utilized the Snowflake Python Connector to build a high-frequency ingestion pipeline for real-time biometric streaming and used Snowflake API to use the SQL to save data.

The centerpiece of our Snowflake integration is Snowflake Cortex AI. Instead of exporting sensitive health data to external third-party LLMs, we utilized Cortex (Mistral-Large) directly within the Snowflake security perimeter. This allows mediQuack to generate "AI Wellness Insights" and "Breathing Tips" based on the samples of a user's heart rate, ensuring that advanced machine learning happens exactly where the data lives—maximizing both speed and data privacy.

Meditation Songs

Meditation songs from below website, it's copyright free music

https://pixabay.com/music/search/meditation/

Challenges we ran into

Scaling the Visuals: Mapping BPM (scale of 60-100) and RPM (scale of 12-20) on a single graph made the data look flat. we solved this by implementing a dual-axis Matplotlib rendering engine on the backend.

Timezone Synchronization: Converting raw Snowflake timestamps into readable "Human Time" that matched the local user's timezone required careful handling of UTC-to-Local conversion in Pandas.

On-Chain Constraints: Solana Memo transactions have size limits, so we had to optimize our JSON payloads to include the most critical vitals data while keeping the transaction lightweight.

Accomplishments that we're proud of Successfully integrating four major technologies (Snowflake, Cortex AI, Solana, and iOS) into a single, cohesive demo.

Achieving a "One-Tap" flow that generates a graph, emails it, and anchors it to the blockchain simultaneously.

Building a UI that makes complex data (like raw timestamps) feel intuitive and "medical-grade."

What we learned

We learned the power of Snowflake Cortex for on-the-fly data analysis—it's significantly faster than traditional API calls to external LLMs. We also gained deep experience in the Solana transaction lifecycle and how to handle multipart form data for image transfers between mobile and backend.

What's next for mediQuack

NFT Minting: Moving from simple Memos to full Metaplex Core NFTs of health "snapshots."

Doctor-Patient Portals: Allowing healthcare providers to verify these Solana-anchored records via a web dashboard.

Edge Processing: Moving more of the AI analysis to the device for even lower latency.

Built With

Share this project:

Updates