Inspiration
Clinical trial teams still spend a lot of time on routine adherence calls, and side effects often get reported late. Apps and portals sound good on paper, but a lot of patients — especially older ones — just pick up the phone. When I saw CALL-E, it felt like the right channel for that problem: outbound voice check-ins that can feed straight into a coordinator workflow.
What it does
Cura is a Django coordinator portal for outpatient trial monitoring. It:
- manages trials and patients
- places CALL-E check-in calls when they’re due (and lets a coordinator start one from the patient record)
- builds each call task with dose, protocol, and prior notes
- ingests structured results (adherence, symptoms, transcript)
- raises alerts when something looks high-risk, so the coordinator can review and acknowledge
How I built it
I built Cura as a Django app with a coordinator portal (trials, patients, alerts, monitoring, call detail). Outbound calls go through the CALL-E Python SDK (calle-ai). After a call, results come back through webhook/ingestion and triage rules decide whether clinician review is needed.
For demos and development without a live dial, I added dry-run scheduling and a simulate_webhook path so the completed check-in and alert flow still works end to end.
Challenges I ran into
- Turning a real clinical workflow into something judges can follow in under three minutes
- Making CALL-E results useful in a UI (not just “call placed”), with transcripts, triage, and acknowledgements
- Destination limits on live dialing for some regions — so I kept the Kenya site story in the product, and show the completed results path when a live ring isn’t available
- Keeping secrets, real numbers, and PHI out of the public repo while still making the project easy to run
What I learned
I learned how to wire a real product surface around CALL-E — not just place a call, but schedule context, ingest structured outcomes, and give coordinators a clear place to act. I also got sharper about demo design: show the workflow clearly, and be honest about platform limits.
What’s next
I’d like to add stronger scheduling/ops controls, better multi-site support, and a public demo environment judges can click through without setup.
Log in or sign up for Devpost to join the conversation.