Project overview

Untreated heart failure (HF) tends to progress over time. A machine learning (ML) tool has been developed to identify when a patient's condition advances from chronic (Stage C HF) to advanced (Stage D HF). This transition is critical, as it significantly impacts treatment options, often leads to delayed referrals, and is associated with higher mortality rates. ML has the potential to support clinical workflows, such as the one described here, by enabling real-time chart reviews and streamlining early referrals for specialist evaluation in heart failure. However, most ML-based clinical tools fail to integrate into real-world settings because they often lack thorough understanding of existing workflows and the specific user requirements. As such, my goal was to engage clinicians in using ML algorithms within the electronic health record (EHR) to identify patients with advanced heart failure (HF).

Process and Methods

Thematic analysis, Interdisciplinary literature reviews, Semi-structured interviews with nurses, primary care physicians, cardiologists, administrators, and quality and data experts.

Deliverables

Understanding of end user requirements of clinical-ML, academic case study paper around the invisible clinical labor driving the successful integration of AI in healthcare.

⚠️ Jan 2025: Under construction, more details to come soon