Abstract

Provider profiling is a critical tool for assessing the quality, cost, safety, and effectiveness of health care delivery, serving the needs of payers, providers, and patients. However, conventional profiling methods face significant limitations, including inflexible modeling, interpretability challenges, and inadequate attention to equity, which hinder evidence-based accountability. This article reviews existing provider profiling methods, highlighting their pitfalls, and introduces a novel design-based approach that leverages artificial intelligence (AI) to address these limitations. By integrating advanced AI techniques—such as causal inference, machine learning, natural language processing, and large language models—this framework enables the construction of versatile performance benchmarks tailored to diverse profiling objectives. The article also discusses the challenges of adopting AI in provider profiling, including algorithmic disparities, data privacy concerns, and the need for transparency, cost-effectiveness, and fairness. When implemented responsibly, AI-assisted provider profiling holds significant promise for advancing health care quality and equity, though its adoption will require collaboration among stakeholders and careful attention to ethical and operational considerations.

 

Citation

Wu, W., Díaz, I. & Horwitz, L.I. Design-based provider profiling with artificial intelligence to enhance quality and equity in health care delivery. Health Serv Outcomes Res Method (2025). https://doi.org/10.1007/s10742-025-00355-8