Background

Integrating AI into a Remote Patient Monitoring (RPM) platform presents challenges due to the necessity of aligning AI features with the needs of physicians and patients while navigating the complexities of compliance and data privacy.

“We believe we can reframe software delivery from the ground up, where every decision, tool, and interaction is guided by contextual intelligence.” — Cadabra Studio

What We Tried (and Why)

Our approach involved enhancing the engineering team’s productivity with AI through tools like Cursor while maintaining strict security standards. The primary focus, however, was embedding AI interactively within the RPM platform, creating distributed, contextual AI touchpoints on critical interfaces rather than a single AI feature.

What Broke or Didn’t Work

The challenge was in balancing automation with meaningful doctor control, and the regulatory complexities of accessing real-time patient data. Vendor lock-in posed sustainability issues as reliance on a single Large Language Models (LLM) provider wasn’t feasible in a healthcare setting.

📌 AI systems in healthcare need to empower user decision-making, not replace it.

The Shift We Made

We designed AI features that allow for doctor interaction and co-creation, embedding AI modules contextually across interfaces. We also moved towards a vendor-flexible approach, enabling dynamic integration with multiple model types—local, cloud-based, proprietary, and open-source.

What Worked (and What Still Doesn’t)

The contextual AI modules improved interaction without disrupting workflow, providing summarization and report organization capabilities. However, navigating the compliant use and retention of patient audio and data in real-time remains an ongoing challenge.

Tradeoffs and Strategic Decisions

Option A Option B
Use a single LLM provider Integrate multiple dynamic models
Simplicity of implementation but risk of vendor lock-in Increased flexibility and compliance, at the cost of complexity

Open Questions We’re Still Exploring

  • How to effectively manage model orchestration across sensitive data domains?
  • What are best practices for ensuring context control in AI systems?
  • How to further develop trust in AI-assisted workflows?

If You’re Solving Something Similar…

If your projects intersect with AI and sensitive domains like healthcare, we’d love to hear your approaches to model orchestration, context control, and trust-building in user experiences.

Contact: hello@cadabra.studio
More at: https://cadabra.studio

🧠 Keep Exploring

If this topic resonated with your work, check out more insights from our team: