MetaCare

When my younger sibling became ill, I helped manage their medication and day-to-day care. It was difficult to know whether they were taking their medication as scheduled, and I could not be there 24 hours a day. While at work, I had to repeatedly call to check and remind them. I wanted a better way to stay involved in their care without depending on another phone call every time. MetaCare grew out of that experience. It is a mobile-first care-coordination prototype for a patient and one trusted caregiver. Its purpose is to make medication records and clinic schedules easier to share when family members cannot be in the same place, while keeping the patient in control of sharing. Patients set their own schedules and upload a medication-package photo to create a record for a scheduled medication time. With explicit consent, a caregiver connects using a short-lived, one-time code and can view dated records, original photos, schedules and shared clinic appointments. Patients can cancel records, control appointment sharing or unlink the caregiver. The interface includes calendar views and configurable reminder settings. The goal is not to replace a caregiver or a medical professional. It is to reduce the need for repeated calls by giving families a shared record they can check when apart. A photo is a user-submitted record, not proof that medication was swallowed. MetaCare does not interpret prescriptions, recommend dosages or diagnose conditions. I built the prototype from scratch during the hackathon in Visual Studio Code. By my estimate, approximately 70% of the AI-assisted development used Meta's Muse Spark, mainly for architectural design and detailed implementation, while approximately 30% used GPT for the remaining development tasks. It combines a Node.js backend, a mobile web/PWA interface and Android WebView wrapper source. Both models were used during development, not for runtime medical inference. This is an early prototype, not a clinically validated or production-ready medical system.