UC Researchers Develop AI-Powered Algorithm Cutting Hypertension Deaths by Nearly 40 in Trial

In a groundbreaking study published Wednesday in BMJ Open Quality, researchers at the University of California revealed that an AI-powered algorithm designed to standardize hypertension treatment has dramatically improved health outcomes for nearly 5,000 patients while preventing an estimated 38 deaths across the state’s premier academic health system. The algorithm, developed by a multidisciplinary team from UC San Francisco, UC Berkeley, and other UC institutions, was deployed in 2023 across all six UC academic health centers, where it helped increase the percentage of hypertensive patients achieving controlled blood pressure from 68.5% to 73.9%—a modest but statistically significant gain with outsized implications for public health.

How the UC Algorithm Turned the Tide on a Silent Killer

Hypertension, often dubbed the "silent killer," affects an estimated 48% of all U.S. adults—roughly 122 million people—according to the most recent data from the Centers for Disease Control and Prevention (CDC). Despite the availability of effective medications like ACE inhibitors, beta-blockers, and diuretics, only about 25% of hypertensive patients achieve adequate blood pressure control nationwide. The UC algorithm was designed to bridge this gap by embedding a stepwise, evidence-based treatment pathway directly into electronic health records (EHRs), ensuring consistent care regardless of a patient’s location within the UC system.

A Standardized, Scalable Approach to a Persistent Problem

The algorithm’s core innovation lies in its simplicity and adaptability. Developed by a team including physicians, nurses, pharmacists, and data scientists, the tool guides clinicians through a tiered treatment plan that prioritizes affordable, widely available medications first—such as thiazide diuretics or calcium channel blockers—before progressing to more specialized options if needed. This approach not only reduces clinical variability but also addresses disparities in care, which disproportionately affect low-income and minority populations who often face barriers to accessing newer, more expensive drugs.

Dr. Sandeep P. Kishore, the study’s lead author and an associate professor of medicine at UCSF, emphasized in a statement that the challenge of hypertension management isn’t a lack of scientific knowledge but rather the failure to implement existing treatments consistently. "The science is clear—we know how to control blood pressure," Kishore said. "This is all about having a system-wide focus that actually moves the needle." The algorithm’s implementation in 2023 coincided with a 5.4 percentage-point increase in patients achieving target blood pressure levels, from 68.5% to 73.9%, translating to improved health for 4,860 individuals treated across UC’s six academic health centers during the study period.

The Human and Economic Cost of Uncontrolled Hypertension

The stakes of uncontrolled hypertension are staggering. In 2023 alone, the CDC reported that high blood pressure contributed to or directly caused over 664,000 deaths in the U.S., making it one of the leading preventable causes of mortality. Beyond mortality, uncontrolled hypertension is a major driver of heart attacks, strokes, kidney failure, and cognitive decline. The annual economic burden of hypertension in the U.S. is estimated at $131 billion, including direct medical costs and lost productivity. The UC algorithm’s ability to avert an estimated 72 strokes, 48 heart attacks, and 38 deaths over just two years underscores its potential to not only save lives but also reduce healthcare spending.

From Theory to Practice: How the Algorithm Works in Real-World Settings

Unlike many AI-driven healthcare tools that remain confined to research labs, the UC hypertension algorithm was designed for immediate, real-world application. It integrates seamlessly into existing EHR systems, such as Epic, used by the UC health centers. Clinicians receive automated prompts when a patient’s blood pressure readings exceed target thresholds (typically 130/80 mmHg for most adults), guiding them through a structured decision tree to adjust medications. The algorithm also flags patients who may benefit from lifestyle interventions, such as dietary sodium reduction or increased physical activity, aligning with guidelines from the American Heart Association.

Prioritizing Affordability and Equity in Treatment

One of the algorithm’s key features is its emphasis on cost-effective medications, which addresses a critical barrier to adherence. Many patients skip doses or discontinue treatment entirely due to high out-of-pocket costs, particularly for newer classes of antihypertensives like ARBs or ACE inhibitors. By defaulting to generic options first, the algorithm ensures that financial constraints do not impede effective treatment. This approach aligns with the UC system’s mission to provide equitable care, as low-income patients are overrepresented among those with uncontrolled hypertension.

Challenges of Scaling the Model Beyond UC

While the UC algorithm’s success is promising, experts caution that its scalability may be limited by the resources required for implementation. Dr. Matthew Alexander, a cardiologist at Vanderbilt University Medical Center who was not involved in the study, noted that the model’s frequent patient monitoring and multidisciplinary collaboration—including pharmacists and nurses—might be difficult to replicate in under-resourced healthcare systems. "The sort of number of resources and different types of providers that were necessary for this more frequent checking up … that’s not clear in terms of its ability to translate to other systems," Alexander said. However, he acknowledged the approach as an "interesting" solution to a longstanding problem.

Broader Implications for Chronic Disease Management

The UC study arrives at a pivotal moment in healthcare, as hospitals and health systems nationwide grapple with rising rates of chronic diseases amid staffing shortages and budget constraints. The algorithm’s success suggests that AI-driven tools could play a larger role in standardizing care for conditions like diabetes, asthma, and heart failure, where treatment variability remains a major issue. "Even in a complex, decentralized system, you can standardize care at scale—if you keep the approach simple, affordable, and flexible, and pair it with strong clinician engagement and workflow integration," Kishore said in an emailed statement. His remarks highlight a growing consensus that the future of chronic disease management may lie in hybrid models combining technology with human oversight.

Key Takeaways: What This Means for Patients and Policymakers

• The UC hypertension algorithm improved blood pressure control in 73.9% of patients, up from 68.5%, during the 2023-2025 study period.
• Researchers estimate the tool prevented 38 deaths, 72 strokes, and 48 heart attacks across UC’s six health centers.
• By prioritizing affordable medications and reducing clinical variability, the algorithm addresses equity gaps in hypertension treatment.
• The study’s success suggests scalable potential for AI-driven tools in chronic disease management, though resource constraints may limit widespread adoption.
• Uncontrolled hypertension remains a leading cause of death in the U.S., with over 664,000 fatalities linked to the condition in 2023.

The Road Ahead: Will Other Health Systems Adopt the UC Model?

The UC’s hypertension algorithm is already generating interest beyond California. In interviews, Kishore and his team noted discussions with health systems in other states, as well as the Veterans Health Administration, about adapting the tool to their EHR platforms. However, the transition won’t be seamless. Successful implementation will require buy-in from clinicians, integration with existing workflows, and ongoing data monitoring to ensure the algorithm remains effective as treatment guidelines evolve. For now, the study serves as a proof of concept that standardized, data-driven care can yield measurable improvements in patient outcomes—even in large, decentralized health systems.

The science is clear—we know how to control blood pressure. This is all about having a system-wide focus that actually moves the needle. — Dr. Sandeep P. Kishore, lead author of the study and associate professor of medicine at UCSF

Frequently Asked Questions