Leveraging GenAI for Hospital Quality Measures: Enhancing Efficiency Without Compromising Accuracy

Hospital quality measures are essential for monitoring patient safety and improving healthcare outcomes. Electronic Clinical Quality Measures (eCQMs) provide a standardized way to assess hospital performance using electronic health record (EHR) data. eCQMs can be divided into inpatient/hospital eCQMs, outpatient eCQMs and clinician eCQMs. Hospital eCQMs characterise rates of hospital complications such as obstetric complications, severe hypoglycaemia or safe prescription of medications such as anticoagulation therapy for atrial fibrillation/flutter.

eCQMs are calculated from a numerator and denominator to determine an overall performance rate, which serves as a measure of healthcare quality. However, calculating eCQMs is often time- consuming and resource-intensive, requiring manual data extraction and analysis. The variability in EHR structures across hospitals further complicates the process, making it difficult to compare and benchmark healthcare performance effectively.

My project, conducted in collaboration with Evidentli, a healthcare data startup, explored whether GenAI could automate eCQM calculations while maintaining clinical accuracy. By developing and evaluating GenAI-generated workflows for four eCQMs, we identified both the strengths and weaknesses of GenAI-driven automation. While GenAI significantly reduces manual effort, errors in patient classification and criteria application highlight the need for human oversight. However, through iterative refinements and prompt engineering, GenAI performance significantly improved, demonstrating its potential for automating quality measurement processes.

AI is revolutionising healthcare analytics, but human expertise remains crucial to refining its outputs. By optimising GenAI usage and evaluation frameworks, we can create more reliable hospital performance measures, improve patient safety, and enhance healthcare system efficiency. GenAI’s ability to efficiently analyse large-scale healthcare datasets demonstrates immense potential for scalability and interoperability between healthcare organisations. As healthcare systems continue to evolve, GenAI-driven solutions will be key to efficiently deriving insights to improve quality and safety of patient care.