Axxess Sentinel

Inspiration

We were inspired by a simple problem in chronic care: wearable data exists, but meaningful early action often does not. Patients and caregivers need timely, understandable signals, not raw numbers or black-box outputs.

What it does

Axxess Sentinel streams live vitals, predicts short-term risk trends, and triggers tiered alerts (info, warning, critical). It separates observed data from forecasted data, ranks patients for caregivers, supports beneficiary notifications, and includes onboarding-based baseline risk inference plus ICD-linked insurance context.

How we built it

We built a Next.js frontend for patient and caregiver dashboards, a Node.js/Express backend for auth, routing, alerts, and prediction orchestration, and WebSocket/SSE channels for real-time updates. Data persistence uses Prisma with SQLite. We integrated a Flask ML API for heart-rate forecasting and added assistant/coaching capabilities through an LLM service with fallback handling.

Challenges we ran into

We had to balance alert sensitivity with alert fatigue, tune mock vitals to be realistic while still testable, enforce strict caregiver-patient authorization, and handle real-world API issues like ML timeouts and LLM rate limits. We also resolved persistence and schema consistency issues during database migration.

Accomplishments that we're proud of

We’re proud of the clear live-vs-predicted UX, persistent role-based mapping, testable alert pipeline, and robust fallback behavior. The system is not just a dashboard; it reflects real healthcare workflow constraints with guardrails and auditability.

What we learned

We learned that healthcare software needs reliability and clarity as much as model quality. Explainability, consent boundaries, access control, and safe failure modes are critical for user trust.

What's next for Axxess Sentinel

Next, we plan to connect real wearable APIs, improve model calibration with larger datasets, add configurable clinical rules, expand notification channels (email/SMS providers), integrate interoperability standards (FHIR/HL7), and strengthen compliance and deployment readiness for pilot use.

Built With

• audit-logging-database/orm:-sqlite
• bcrypt
• custom-css
• dotenv
• express.js
• featherless-ai-(llm)-integration-tooling:-npm
• git
• javascript
• languages:-typescript
• prisma-cli
• prisma-ml/ai:-flask-api
• python
• react
• recharts-backend:-node.js
• risk-scoring-logic
• role-based-access-control
• sql-frontend:-next.js-(app-router)
• sse-auth/security:-jwt
• timesfm-based-heart-rate-forecasting
• websocket-(ws)