Using AI to Make Precision Medicine More Accessible

In the world of precision medicine, molecular diagnostics (MDx) have the power to transform patient care by using genetic information to guide individualized approaches to care and improve clinical outcomes. Yet access to these advanced tests often lags their development, primarily due to cost and uneven insurance coverage. Department of Clinical Pharmacy Assistant Professor Emily A. Hammad Mrig, PhD, is tackling this issue by using artificial intelligence (AI) tools to identify patients and solutions that make MDx more affordable and equitable. 

Supported by a Career Development Award (K01) from the National Human Genome Research Institute (NHGRI), Mrig’s research focuses on circulating tumor DNA (ctDNA) testing — a molecular diagnostic that exemplifies the broader challenges of implementing new diagnostics. For patients with advanced-stage, non-small cell lung cancer, ctDNA testing can deliver faster molecular results through a simple blood draw rather than an invasive tissue biopsy, potentially shortening the time to treatment. However, barriers that delay or prevent payment for testing can worsen outcomes.  

Identifying barriers in access to MDx coverage and payment options is a critical first step in understanding who gets access to these tests and why. To overcome this long-standing challenge, Mrig developed an AI tool using natural language processing (NLP) to analyze and extract information from unstructured data, like diagnostic results in PDFs. 

“The tool is designed to look through all of the patient records, identify which patients got this diagnostic, see what their results were, and then tie that back to structured data like billing and coding information,” Mrig said. “From there, you can identify payment pathways: Was it paid for out of pocket? By a financial assistance program? Or by insurance?” 

Breaking down barriers 

Despite its transformative potential, ctDNA testing (and many other MDx technologies) often remains out of reach for underserved populations due to cost, insurance coverage gaps and other systemic barriers. Mrig’s goal is to understand these barriers and, by working with patients and stakeholders at UCSF and in the community, lay the groundwork for policy or process interventions. 

“Clinical guidelines really guide insurance coverage,” she said. “But technology moves so quickly that clinicians are using it before the guidelines catch up, which means insurance hasn’t caught up, either.” The result? Patients may face delays in testing or struggle to pay out of pocket, blocking access to potentially lifesaving technology. 

“My passion is equity. That’s what drives everything I do,” said Mrig, who is also a member of the Center for Translational and Policy Research on Precision Medicine (TRANSPERS). “How can we replicate what we’re doing well at UCSF to give patients in community settings the same opportunities?” she added, referencing UCSF Specialty Pharmacy services as a successful institutional model. 

Embracing AI for equity 

While Mrig initially hesitated to adopt AI tools for her research, she ultimately saw the potential for NLP to tackle complex and persistent methodological and data problems. “I looked for every other way to address this data issue. Then I realized NLP could allow me to do something I’ve wanted to do since my dissertation: conduct a large, empirical analysis of [insurance and cost-related] access barriers to molecular diagnostics.” 

Mrig said NLP doesn’t replace human insight but rather enhances it. “This is qualitative work in a different form. It allows us to get at the richness of the data — like socioeconomic factors, patient preferences and the nuances of care decisions,” she said. “But we have to be careful not to build bias into these tools. That’s why I’m integrating data from UCSF and Zuckerberg San Francisco General Hospital and Trauma Center, to reflect a diverse patient population. Otherwise, the project wouldn’t make sense, because we need to develop solutions that work not just in academic medical centers, but also in community health care settings, where most patients receive their care.” 

Insurance coverage and centering the patient voice 

Mrig said she has been interested in understanding the intersection of insurance coverage and financial issues related to genetic testing since she was a PhD student because she sees how it connects to a larger conversation among scientists, policymakers and others about who should be getting the tests. 

“Studies show that population-based testing could identify a much larger group of at-risk patients. But broader testing raises concerns for payers about covering follow-up care,” she said. “Working with payers has shown me how coverage decisions are made and why certain barriers persist. The challenge lies in balancing cost concerns with the clear benefits of prevention and early detection.” 

As Mrig looks to the future, she emphasizes the importance of centering patient voices in her work. “We hear a lot from clinicians, scientists and researchers, but the patient perspective is often missing,” she said. “It’s not just about targeted therapies. It’s about understanding the molecular diagnostics that guide them and the systems that deliver them to patients.”