MedVault
Own your medical records.
Demo
About
The problem it solves I grew up in India, where going to the doctor often means walking into a small clinic inside someone’s home. There’s one room. A doctor at a desk. Chairs outside filled with patients waiting their turn. There is usually no digital record waiting for you. If you have a serious condition, your “medical history” is a paper folder you carry yourself — prescriptions, lab reports, scan results. If you forget it, the doctor has nothing. If you lose it, your history is gone. If the doctor asks what medication you’re taking, you might not know the name. Sometimes people bring the empty blister foil of old medicines as proof of what they were prescribed. That foil becomes the record. When you get blood work done, you carry the paper results back by hand. If you want a second opinion, you walk to another clinic and show them the same papers. Nothing is shared digitally. Nothing is structured. Everything depends on what the patient remembers or physically carries. This is not just an India problem. Around the world, especially in decentralized and private healthcare systems, patients are the couriers of their own history. MedVault solves this. --- ### What MedVault Does MedVault gives every patient a secure, structured medical vault that they control. Patients can: - Store their medical history digitally - Share specific documents with a doctor - Share records by time range (e.g., last 3 years) - Revoke access at any time - Maintain a clearly marked emergency summary - See a full audit log of who accessed their records Doctors can: - Upload visit notes and test results - Add follow-up notes - Access only what the patient explicitly shares - Contribute to a patient’s history without owning it --- ### Why This Matters MedVault makes healthcare: - **Safer** — doctors see structured history instead of guessing from memory or packaging - **More flexible** — patients can get second opinions without carrying folders - **More accountable** — emergency access is immediate but logged - **More private** — consent is explicit and revocable - **More interoperable** — structured records align with global standards without centralizing ownership MedVault is not another hospital system. It is a **patient-sovereign continuity layer** — built for real-world healthcare where records are fragmented, mobile, and human. Use of AI tools and agents MedVault does not currently deploy autonomous AI agents inside the live system. However, the architecture is intentionally designed to support permissioned AI assistance within the Canton/Daml privacy model. AI is treated as a scoped service agent — not a system authority. --- ### Current State At this stage, MedVault focuses on: - Privacy-preserving record storage - Explicit consent enforcement - Structured data visibility - Governance and auditability No automated AI analysis is performed on patient data by default. --- ### Designed AI Integration (Planned Modules) The system is architected to support two AI-assisted workflows in future iterations: #### Provider Documentation Assistance AI can assist providers by: - Converting free-text notes into structured formats - Providing predictive text during documentation - Extracting structured data from uploaded handwritten notes - Suggesting standardized terminology mappings All outputs would require explicit provider confirmation before being finalized. --- #### Patient-Authorized Longitudinal Review With explicit patient consent, AI may be used to: - Detect trends in lab results - Flag potential medication interactions - Highlight recurring symptoms - Generate structured summaries for provider review This would not be automated diagnosis, but pattern assistance. AI access would be: - Explicitly permissioned - Time-scoped - Logged - Revocable - Limited to the documents granted under contract visibility rules --- ### AI Within the Privacy Model Within the Canton/Daml framework: - AI operates as a permissioned service participant. - It cannot independently retrieve records. - It cannot access data outside explicitly granted scopes. - All AI interactions are auditable through contract logs. This ensures that AI remains assistive and bounded by patient sovereignty. --- MedVault’s philosophy is simple: AI may enhance clarity and safety — but it must operate within enforceable consent boundaries. Privacy comes first.
How it's made
Challenges we ran into A key challenge we faced was handling updates to patient-owned health records in Daml. When a contract is updated, the original contract is archived and a new contract is created with a new contract ID. When a doctor added a prescription to a patient's health record, the original record contract was archived. The newly created contract had a different contract ID, which the doctor did not have. Storing the contract ID on access templates did not work either — by the time a pharmacy attempted to dispense medication, the stored ID was already stale due to earlier operations. To solve this, we redesigned the model so that the "healthRecordCid" is passed as a choice parameter at exercise time rather than stored on templates. Each dashboard queries the ledger for the patient's current HealthRecord contract ID immediately before exercising a choice, ensuring it always has the latest reference. This preserves the HealthRecord as the single source of truth while working within Daml’s archive-and-create model.