A new predictive model is helping clinicians uncover hereditary cancer risk earlier and more accurately by standardizing how patient history is captured and used.
Many people at high risk for hereditary cancers never receive genetic testing—often because their personal and family histories are incomplete or inconsistently documented. These gaps in data can lead to missed or delayed diagnoses and, ultimately, lost opportunities for life-saving prevention and treatment. That realization motivated Sapna Syngal, MD, MPH, Director of Research in the Division of Cancer Genetics and Prevention at Dana-Farber Cancer Institute, to take action. Together with her colleagues, including Hajime Uno, PhD, Chinedu Ukaegbu, MB, BS, Miki Horiguchi, PhD, and Matthew Yurgelun,MD, FASCO, CGAF she developed a suite of predictive tools known as PREMM (Prediction Models for Mendelian Mutations). Today, PREMM models are helping clinicians at Dana-Farber and beyond identify high-risk patients more efficiently and equitably—transforming how hereditary cancer risk is detected and managed.
Standardizing the Missing Piece: Patient History
“Because patients typically aren’t asked their individual and family health history in a systematic way, a lot of that essential information isn’t captured in the medical record,” Syngal explains. “Even advanced AI tools can’t help if the data simply doesn’t exist in a retrievable format.”
PREMM tackles this longstanding challenge by offering a scalable, standardized approach to capturing the right details of a patient’s history—whether entered by providers or directly by patients themselves. The model ensures that critical information is complete, structured, and ready for use in predictive algorithms and clinical decision-making.
“It’s about collecting the right elements of a patient’s history in a standardized way,” Syngal says. “That’s what makes the prediction accurate and actionable.”
The Engine Behind PREMM
At the core of the PREMM models lies several logistic regression algorithms—what Syngal calls “the brain of the system.” The models process straightforward, patient-derived data such as age, cancer type, and family history to calculate a personalized risk score for inherited cancer syndromes.
“PREMM isn’t powered by deep learning or black-box AI,” she notes. “The algorithms are transparent, validated models built from tens of thousands of real patient records.”
The most widely used version, PREMM5, estimates the likelihood that a person carries mutations in any of the five genes associated with Lynch syndrome, a hereditary condition that significantly raises the risk for several cancers. The team has also expanded the model into PREMMPlus, which assesses risk across 18 cancer susceptibility genes—broadening its reach and clinical value.
Validation and Clinical Impact
PREMM’s utility has been confirmed through numerous peer-reviewed studies. A 2017 paper in the Journal of Clinical Oncology detailed the development and validation of PREMM5, cementing its role in hereditary cancer risk assessment and influencing national screening guidelines.
More recently, a 2023 study in Familial Cancer demonstrated that PREMM5 could distinguish Lynch syndrome–associated colorectal cancers from sporadic cases with 100% sensitivity—showing its potential as a cost-effective alternative to tumor-based screening.
Institutional and Research Support
PREMM’s success reflects both scientific collaboration and institutional investment. Dana-Farber’s Center for Cancer Genetics and Prevention provided the clinical foundation—robust patient data, expert oversight, and access to individuals with diverse genetic backgrounds—while the institute’s informatics and analytics teams helped design and refine the tool’s algorithmic structure.
Ongoing development has been supported by multiple National Institutes of Health (NIH) and National Cancer Institute (NCI) grants, enabling implementation and validation studies in real-world settings, including Denver Health, Logan Health in Montana, and Kaiser Permanente Mid-Atlantic.
Embedding PREMM Into Everyday Care
PREMM is more than a research innovation—it’s changing policy and practice. A PREMM score above 2.5 is now recognized by many insurers as a validated indicator for genetic testing eligibility.
“Before, it was easy to deny testing based on vague family history,” Syngal says. “Now, PREMM provides a quantitative score that helps patients get approved for testing—especially for genes beyond BRCA.”
Syngal’s team is now working to integrate PREMM directly into electronic medical records (EMRs) so that risk assessment becomes part of routine care, not a specialized or siloed process.
“Right now, patients often have to see a genetic counselor just to get their history taken,” she explains. “We’re trying to take that whole front piece and make it easier—and make it patient-facing.”
Once integrated, PREMM could automatically generate risk scores and trigger referrals for genetic counseling, seamlessly embedding prevention into the clinical workflow.
What’s Next: AI, App Development, and Broader Access
While PREMM doesn’t currently rely on artificial intelligence, Syngal sees clear potential for synergy in the future. “If we could use AI to pull structured data from EMRs—like age and cancer history—that would be a huge advance,” she says.
To further expand access, her team is also partnering with an external vendor to develop a commercial app that enables patients to input their own history directly, streamlining the assessment and referral process across institutions.
“We’re not just building models,” Syngal emphasizes. “We’re building pathways to make sure patients get the care they need, when they need it.”
Director, Lynch Syndrome Center; Director of Clinical Research, Dana-Farber Cancer Institute Center for
Cancer Genetics & Prevention;
Senior Physician, Dana-Farber;
Associate Professor of Medicine,
Harvard Medical School
Biostatistician, Division of Population Sciences, Department of Medical Oncology, Dana-Farber
Research Associate in Medicine, Dana-Farber
Assistant Professor, Harvard Medical School and Researcher, Dana-Farber
Director of Research in the Division of Cancer Genetics and Prevention, Dana-Farber
