Q&A: Closing care gaps and advancing health equity with AI-driven insights

Andrew Whitehead, VP and Head of Population Health, Bristol Myers Squibb: If our medicines are not reaching or working for real-world populations, we limit both patient impact and outcomes, and we miss the potential full value of our science.

One of the biggest obstacles is that care gaps are often invisible in traditional datasets. They are fragmented across the patient journey and are often driven by structural and social factors like socioeconomic status or transportation insecurity. To understand and close these gaps at scale, we have to think differently and rewire how decisions are made. This requires collective action and partnerships that allow us to pilot new approaches, measure the impact, and create sustainable, scalable solutions.

Gary Grad, MD, Senior Medical Oncologist, Tempus: The overarching goal is for every cancer patient who is medically fit for treatment to be identified for guideline-supported therapy at a time point that optimizes their care. Despite the best efforts of physicians and health systems, gaps in biomarker testing and optimal care persist for many reasons.

These reasons are multifactorial. We see complexities and bottlenecks in clinical workflows. There are knowledge gaps, as clinical guidelines evolve almost daily. There are also inherent biases related to a patient’s socioeconomic status, age, or other factors. For example, a physician may think they ordered a critical biomarker test, but for various operational reasons, it never gets done. The ability to identify and close these testing and care gaps in a cause-agnostic fashion can have a tremendous impact.

Dr. Grad: To accurately identify a care gap, you need deep clinical context. By integrating with the electronic health record (EHR), platforms like Tempus Next Pathways can analyze not only structured data but also a vast amount of unstructured information from lab values, radiographic reports, therapeutic histories, and clinical notes. This allows us to build a more complete, longitudinal patient story.

This comprehensive view helps us capture clinical context and decision-making insights that are often lost in traditional datasets like claims. We can surface the specific reasons a patient was not tested or why their care diverged from the recommended pathway, making these invisible factors visible and, therefore, actionable.

Andrew Whitehead: We collaborated with Tempus to deploy the Next Pathways platform across 13 health systems with the goal of improving adherence to guideline-directed biomarker testing in patients with advanced NSCLC. We wanted to identify a specific care gap, co-create an intervention, and then measure and monitor the impact in near real time. The dynamic, real-time nature of the platform was very appealing because it gave us a view of the entire patient journey, not just a single point in time. This challenged our assumptions and allowed us to be more agile in our strategy.

Andrew Whitehead: The results were incredible. Before the intervention, we found that the baseline comprehensive biomarker testing rate was only about 72%, meaning nearly one in four patients in this well-established treatment paradigm was not receiving guideline-indicated testing. After deploying the platform, we saw a 12-point absolute increase in testing, bringing the rate up to 84%.

This has significant potential for real-world impact. In this study, the intervention reduced missed treatment opportunities and accelerated time to immunotherapy or targeted therapy options, where outcomes are most sensitive. The granular, longitudinal data also showed us where testing was breaking down—by site type, timing, or patient characteristics—revealing gaps that guidelines alone do not surface. These insights are now directly informing a more targeted, equity-focused strategy.

Dr. Grad: Alert fatigue is real, but nothing is more frustrating for a physician than missing a biomarker that could have helped a patient. The key is to deliver alerts that are accurate, actionable, and integrated into the existing clinical workflow. We work with each health system to customize the platform, whether that means sending alerts directly to a physician or to a patient navigator. More importantly, this kind of intervention does more than just close individual gaps. The feedback provided to the health system helps change the entire culture of testing at the institutional level. It allows them to identify and fix the operational bottlenecks that contribute to gaps, leading to more consistent, guideline-supported care for all patients.

Dr. Grad: For providers, these platforms will continue to be essential for integrating rapidly changing guidelines into practice and ensuring no patient is left behind. For life science partners, the real-time data provides invaluable insights into physician practice patterns, treatment sequencing, and unmet needs. It can even help de-risk biomarker-based clinical trials by ensuring a relevant biomarker is available and top of mind for physicians.

Andrew Whitehead: The speed at which we operate matters because we are talking about patients’ lives. Every day counts. Real-time data allows providers to intervene and reach patients who might otherwise receive testing too late or not at all. It allows us to be more intentional and strategic with our outreach, education, and support programs. We are still learning, which is a beautiful part of this process. It allows us to be more agile and continuously evolve our strategy to meet the needs of patients.