Real-Time Heart Rate Monitoring Built Into a Home Cardiac Rehab App

Stable, real-time heart rate data without a custom pipeline

Rather than building a custom Bluetooth data pipeline and handling every device's frequency quirks in-house, we integrated Terra API to normalize heart rate data from Bluetooth-enabled monitors into a consistent, real-time stream. That gave the app a stable connection during workouts and heart rate data in a format it could evaluate against a patient's target zone as the session was happening, not after the fact.

• Normalized heart rate data from Bluetooth monitors reporting at incompatible frequencies
• Maintained a stable Bluetooth connection through an active exercise session
• Evaluated target-zone status in the moment, not after the session ended
• Removed the need for a custom in-house Bluetooth data pipeline

Progression that mirrors a clinical setting, not a game

We built the app around the clinical team's exercise design: 15-minute sessions first, with 30-minute and then 45-minute tiers unlocking only once a patient showed they were ready for more, with easy, moderate, and advanced levels within each tier. The progression logic mirrors how cardiac rehab actually works in a clinical setting, where intensity increases gradually and under supervision. The app holds that same discipline even when no clinician is in the room.

A clinical progress record, not a fitness dashboard

Daily, weekly, and monthly tracking, with a monthly view showing total exercise minutes and average heart rate per session. For a rehab app, this is the record a patient and their care team use to see whether recovery is actually progressing, built to answer that question specifically rather than just show activity graphs.

Cutting smartwatch integration to protect the core feature

Apple Watch, Samsung, and Fitbit integration were explored early in the engagement, then cut. Each platform would have added its own SDK and its own edge cases to manage, without improving the Bluetooth heart rate monitoring the clinical model actually depended on. We flagged it as a real scope decision rather than quietly dropping it, so the client could revisit it later if it made sense.