AI Models Aim to Detect Undiagnosed Disease Earlier
Advanced AI models can analyze EHR data to identify patients with undiagnosed conditions before symptoms appear. This early detection is becoming more valuable as new therapies can delay or cure disease. Health IT leaders must prioritize where to deploy these models and turn insights into action.
Roughly 35.5 million U.S. adults have chronic kidney disease, with about 90 percent unaware. Since the U.S. Preventive Services Task Force does not recommend population screening, most organizations lack proactive kidney programs, leaving patients undiagnosed until later stages. New genetic tests can identify disease etiology, and newer drug classes can slow progression, making early detection increasingly valuable.
CMS's LEAD model extends risk windows for accountable care organizations, meaning earlier intervention yields greater financial benefit. Each disease demands a distinct program design—CKD requires population-scale approaches while adult-onset type 1 diabetes needs different workflows—so health IT leaders must carefully prioritize where to deploy predictive models.
This story could reshape how healthcare systems allocate resources, potentially shifting focus from reactive treatment toward proactive screening. Patients with undiagnosed conditions like CKD or type 1 diabetes may receive earlier interventions, possibly avoiding severe complications and reducing long-term costs. However, health IT leaders face difficult prioritization decisions, and success depends on integrating models into clinical workflows. If deployed effectively, such systems could improve outcomes for millions, though disparities in access may persist.