Today, we published a peer-reviewed study in Nature’s Scientific Reports comparing the performance of Waymark Signal^TM, the company’s proprietary machine learning technology, to conventional Medicaid risk models. The study found that Waymark Signal was 90 percent accurate in predicting avoidable emergency room (ER) and hospital utilization for patients receiving Medicaid — stronger performance than leading Medicaid risk models in the field.  

Waymark Signal combines data on social risk factors and patient risk trajectories with healthcare utilization to identify patients at risk for preventable ER and hospital visits. As one of the largest and most representative comparisons of Medicaid risk models to date, the study assessed Waymark Signal’s machine learning approach against traditional regression models for Medicaid, which rely primarily on patient demographics, healthcare diagnostic codes, and medications to predict at-risk patients. Researchers found that Waymark Signal was 3x better at identifying at-risk patients and 10x better at predicting costs compared to conventional models.

"This study demonstrates the potential for machine learning models like Waymark Signal to more accurately identify at-risk patients and drive more effective interventions,” said Sadiq Y. Patel, MSW, Ph.D., an author of the study and Data Science Lead for Waymark. “By enabling care teams to better recognize and act on social and clinical risk factors, Waymark Signal can help them intervene to prevent avoidable disease complications, ER visits, and hospitalizations that negatively impact both patient health and costs."

Additionally, the study found that Waymark Signal also reversed the Black-White prediction bias observed in most risk models. Because Black individuals typically have less access to higher-cost tertiary care centers, traditional cost-based models often under-predict their future costs and assume the lower costs reflect lower health needs. Waymark Signal reversed this bias, demonstrating higher sensitivity for Black patients’ needs and offering one approach for a more equitable application of machine learning in Medicaid risk modeling.

"This study makes an important contribution to the underserved area of Medicaid risk prediction,” said Will Shrank, MD, Venture Partner at Andreessen Horowitz and former Director of Evaluation, Innovation Center for the Centers for Medicare & Medicaid Services (CMS). “While Medicare and commercial insurers have benefitted from numerous risk modeling advances, Medicaid programs have seen far fewer peer-reviewed studies achieving such a significant performance gain. By tripling sensitivity without increasing false alerts, the authors' innovative approach using machine learning provides valuable guidance for more effectively addressing care disparities through personalized outreach to those enrolled in Medicaid."

Waymark works directly with Medicaid health plans and primary care providers (PCPs) to deliver technology-enabled, community-based care for people enrolled in Medicaid. The company’s local care teams use Waymark Signal to identify and outreach patients at-risk of avoidable ER and hospital utilization. As a public benefit company, Waymark has published the key methodological advancements from this study to enable Medicaid programs to apply them to their own data and populations.  

"This study shows that advanced data science and tools like Waymark Signal offer the potential to help us deliver more equitable and effective care that improves outcomes for the chronically underserved,” said Sanjay Basu, MD, Ph.D., Co-Founder and Head of Clinical at Waymark. “Ultimately, that’s why we founded Waymark.”

Researchers used CMS data from 2017-2019 across 26 states and Washington, D.C., to compare the predictive power of Waymark Signal to conventional Medicaid risk models. The states used in the study were identified based on CMS’ Data Quality Atlas, which assesses each state's enrollment benchmarks, claim volume, and data completeness.

“Waymark’s community-based care teams have been using Signal to identify and outreach at-risk Medicaid members in partnership with our health plan and provider partners,” said Aaron Baum, Ph.D., Analytics and Economics Lead at Waymark. “These findings reaffirm what we've also seen in our data: that leveraging our data science capabilities can play a critical role in creating more accessible and equitable pathways to better health for underserved communities."

The full article titled “Prediction of Non-Emergent Acute Care Utilization and Cost Among Patients Receiving Medicaid” was published in Scientific Reports, a peer-reviewed journal published by Nature. The authors for this article were Sadiq Y. Patel, MS, Ph.D., of Waymark; Aaron Baum, Ph.D., of Waymark; and Sanjay Basu, MD, Ph.D., of Waymark.