Care Porta

The Portable Health Agent is a privacy-first, on-device AI health passport engineered to resolve the critical issues of fragmented medical records, data silos, and the severe privacy risks associated with centralised healthcare systems. Currently, transitioning between different hospitals or jurisdictions often results in incomplete patient histories, redundant testing, and suboptimal clinical care. Our project empowers individuals with absolute medical data sovereignty. By fusing local, on-device artificial intelligence processing with blockchain-backed immutability, we deliver a unified and verifiable medical record that never leaves the user’s mobile device without their explicit authorisation. Key functionalities include automated consultation transcription and summarisation (processed entirely locally), a smart dashboard for proactive health insights, and granular access control that permits patients to share specific historical records via time-limited QR codes. Every generated entry is cryptographically verified on-chain, guaranteeing that the medical timeline remains strictly tamper-proof, whilst our offline-first architecture ensures seamless operation even in remote or low-connectivity environments.

Tell us about how you built this project; the nitty-gritty details. What technologies did you use? How are they pieced together? If you used any partner technologies, how did it benefit your project? Did you do anything particuarly hacky that's notable and worth mentioning? (min 280 characters)

We architected the Portable Health Agent with a strict separation between local data processing and on-chain verification to guarantee absolute privacy. The mobile application leverages local Automatic Speech Recognition (ASR) coupled with a lightweight, on-device Large Language Model (LLM). This stack records medical consultations and extracts structured, intelligent summaries entirely on the hardware, completely bypassing the need to transmit highly sensitive audio or text to external cloud APIs.

Following local AI processing, the structured summary is symmetrically encrypted and siloed within the device's secure storage. To establish systemic trust and immutability without exposing plaintext data, we integrated OpenClaw as our core partner technology. We engineered a bespoke OpenClaw skill that generates a cryptographic hash of the encrypted medical record alongside its metadata, anchoring this payload to a blockchain network. This integration seamlessly guarantees the historical integrity of the record.

A particularly 'hacky' and notable engineering feat is our ephemeral data-sharing mechanism. Rather than transmitting the underlying database, the application generates a scoped, time-limited QR code for the doctor. Upon scanning, the doctor's device queries the blockchain to verify the record's authenticity. Simultaneously, the patient's device acts as an ad-hoc local server, temporarily decrypting and serving the requested records directly to the physician's secure web viewer via a local peer-to-peer connection. This establishes a read-only, ephemeral session where the data cannot be downloaded or scraped as plaintext, elegantly circumventing the vulnerabilities of traditional centralised health data exchanges.