via Swad Komanduri
The shift towards value-based payment by the Center for Medicare and Medicaid Services (CMS) is designed to create incentives for healthcare providers to find innovative ways to improve outcomes for their patients. This shift, intended to promote proactive care, depends on providers establishing practices that foster continuous improvement. By measuring outcomes and costs, rather than processes, CMS seeks to incentivize providers to form partnerships with other organizations, establish meaningful relationships with their patients, and prioritize preventative care over service volume. Succeeding in this endeavor requires accurate measurements of value, both in terms of quality and costs, so that payments align with the desired behavioral incentives.
Lessons learned from the COVID-19 pandemic can help deliver on the promises of value-based care. The advancements in real-time risk modeling developed during the pandemic can play a valuable role in achieving these goals. However, there remains much work to be done. Current models and data are missing information about social determinants of health, which may be one contributing factor behind the widespread disparities in health outcomes observed during the COVID-19 pandemic. In the absence of this information, inaccurate models risk incentivizing the wrong behaviors, which may lead to worse outcomes over time. Advances in machine learning, particularly in explainable ML, can help us better understand inaccuracies in our current models, and help shed light on the medium to long term effects of continuing to use these models.