- The combination of machine learning tools and genomic data can help predict the risk of abdominal aortic aneurysm (AAA), one of the most frequently fatal heart conditions - and one that is most often diagnosed only after a patient’s death.
In a new study from Stanford University, published in the journal Cell, researchers combined EHR data, machine learning, and whole-genome sequencing data to create a predictive algorithm that is 70 percent accurate at identifying individuals with elevated risk of AAA.
“Right now, genome sequencing is starting to make its mark,” said Michael Snyder, PhD, professor and chair of genetics at Stanford. “It’s being used a lot in cancer, or to solve mystery diseases. But there’s still a big open question: How much can we use it for predicting disease risk?”
The study has contributed to answering that question by identifying 60 genes and 40 functional modules that are likely to contribute to AAA risk.
To do so, Snyder and his team created a machine learning algorithm called the Hierarchical Estimate From Agnostic Learning, or HEAL, to analyze genomic and clinical data from 268 known AAA patients.
The algorithm was created without preconceptions about which genes might be likely to play a role in AAA development, the team said.
Instead, the algorithm identified novel associations in the genetic data on its own, flagging genetic components involved in regulating immune response, blood vessel development, circulation, and cell-to-cell communication.
The results confirmed previous studies that highlighted the importance of these clinical factors in AAA development, but also expanded knowledge of the molecular basis of the condition, the study said.
“Our results imply that treatments aimed at regulating these modules may have disproportionately beneficial impact on AAA development and open new avenues for future studies,” said Snyder, co-senior author Philip Tsao, PhD, professor of medicine, and other members of the research team.
The study noted that AAA is not a reversible condition, but that lifestyle changes such as hypertension control and tobacco cessation could halt the progression of the disease and reduce the risk of mortality.
“Predictions from personal genomes for the first time makes possible early assessment of this disease, and will facilitate the development of early intervention strategies based on one’s genome baseline,” the team explained.
“Therefore, in addition to a clinical test for early assessment, HEAL could also be deployed as a tool for personalized health management and disease intervention by integrating one’s personal genome and lifestyle.”
The machine learning algorithm could also be adapted to analyze genomic and clinical data for other high-risk conditions, the team said.
“The HEAL framework is potentially applicable to many complex diseases, and our overall approach is expected to be valuable in developing clinical tests that incorporate personal genome sequences into disease-risk prediction,” stated the study.
The growing availability of genome sequencing in the clinical environment will likely facilitate the ongoing development of precision medicine strategies for identifying high-risk diseases and plotting personalized treatment pathways.
The cost of genetic testing is dropping as the technology involved becomes more sophisticated, allowing some large health systems to start offering testing as a routine part of care.
Coupled with rising consumer interest in using genetic data to guide healthcare decisions, the broader implementation of genome sequencing in everyday care could have positive impacts on individuals who may not be aware of the risks they face for complex conditions.
Research like the AAA study lends support to the idea that these steps may help prevent avoidable deaths and improve quality of life for the long term, said the team.
“I see a future in which everyone will be born with their genome sequenced, or shortly thereafter,” Snyder said. “Both your single-gene and your complex disease risk will be used to predict your overall disease risk, and then you can take action based on that information.”