Using artificial intelligence to reduce managed care organizations' monitoring costs and knowledge diffusion: A feasibility study

Managed care organizations (MCOs) and health care in general, face two basic but competing goals: providing high-quality care, and doing so at low cost resulting in a classic principal agent problem. The approaches to mitigating a principal agent problem include: either to incentivize the agent (MCOs or providers) via pay-for-performance (P4P) or to reduce information asymmetries and/or the cost of monitoring agents' quality. A review of the research literature suggests that P4P programs may have only a weak or moderate ability to control outcomes by incentivizing positive behavioral change among providers. A complementary approach to overcoming market failure would be to reduce MCO's cost and temporal delay in monitoring providers' quality improvement processes and outcomes as well as enhancing MCO's ability to diffuse quality improvement knowledge to downstream providers. This study proposes to assess MCO's need for high-validity, low-cost monitoring of downstream provider quality improvement processes and outcomes. In collaboration with the computer scientists from the University of Maryland, Optimal Solutions Group LLC aims to utilize collective scoring knowledge as “inputs” and allow Optimal's proprietary software application, Quality Improvement Template Tracker (QIT2), to generate inferences as “outputs” in order to offer corroborated guidance. This is a feasibility study designed to establish the effectiveness of an Artificial Intelligence (AI)-generated evaluation by comparing its evaluation findings to clinician-based reviews using Krippendorff's Alpha (Kα) statistic. If feasible, an AI-generated evaluation of quality improvement processes and outcomes is an innovative approach for increasing the speed of diffusion of reliable and valid innovations, which ultimately benefit patients, providers and healthcare across the United States.