Validated at ML4H 2025: How Olli Health’s Proprietary AI is Setting the Industry Standard

This past December, Olli Health was selected to present at Machine Learning for Health (ML4H) 2025. Getting into the demo track is a big win for our team and puts us alongside the top researchers and innovators in the healthcare AI space.

Our talk, “A versatile medical coding platform for home health agencies,” focused on how our technology is solving one of the most persistent and complex challenges in the healthcare system.

The Challenge: Home Health Coding is Hard

Home health data looks nothing like hospital data. Instead of clean, structured formats, agencies often receive scanned documents and eFaxes full of messy, inconsistent narrative text. Coders then have to connect that documentation to tens of thousands of ICD-10 codes while staying compliant with home-health specific Medicare (PDGM) rules. It’s slow, manual work and errors are easy to make.

Our Industry-Leading Solution

At ML4H, we demonstrated an AI pipeline with multiple custom built proprietary and trained models that validates our engineering team's ability to deploy state-of-the-art architectures in production environments. Crucially, our presentation also highlighted a distinct competitive advantage: a proprietary solution powered by an ever-growing engine of in-house data that general-purpose models cannot replicate.

Validated Results

The results presented at the conference demonstrate that Olli Health is not just a research project, but a platform driving real operational change. Today, we use our CodePilot+ coding and QR platform to support 100+ home health agency customers and process tens of thousands of charts each month.

• Unmatched Efficiency: Our platform delivers up to 90% time savings, transforming the workflow from manual search to expert validation.

• High Accuracy: Our models achieve impressive recall (vast majority between 0.80–1.0) and precision, ensuring that the final output is home health policy-aware and evidence-backed.

• Human-in-the-Loop By Design: We do not replace human expertise; we drastically enhance it. Our interface allows experts to quickly confirm accurate codes and reject incorrect suggestions, which completes a data feedback loop that continuously retrains and improves our models.

What’s Next

ML4H 2025 reinforced our team’s industry-leading expertise and the clear direction of the field: moving coders and quality reviewers from hunting for information to simply reviewing and validating. That’s exactly what we’re building. With the aging population set to double in the next 15 years, home health volumes will surge, making advanced coding and QA platforms that deliver higher accuracy, stronger compliance, and over 90% faster turnaround times essential.