News
Article
Study: Predictive CAD Model Outperforms Standard Practice in Estimating Individual Risk
Author(s):
Key Takeaways
- The new machine learning model integrates genetics, lifestyle, and medical history for personalized CAD risk assessment, outperforming traditional age-based methods.
- The model achieved an AUC of 0.84, effectively predicting CAD risk, especially in populations typically considered low risk, such as women and younger individuals.
The coronary artery disease (CAD) predictive model help health care professionals tailor personalized treatment methods for patients.
Recently, researchers developed a machine learning model that more accurately estimates a patient’s risk of coronary artery disease (CAD) compared to current standard clinical practice, which is primarily based on age. The findings, which were published in Nature Medicine, utilized 10-year data to create a model that integrates genetics, lifestyle, and medical history, forming more personalized assessments of patients and allowing health care professionals to better tailor treatment for the individual patient.1
Image credit: 博 杨 | stock.adobe.com
CAD is a common type of heart disease that affects the main blood vessels that supply blood to the heart, resulting in a lack of blood flow to the heart, which causes chest pain and shortness of breath, among other symptoms. Complete blockages can cause a heart attack. The disease often develops over many years, but the main cause is a buildup of fats, cholesterol, and other substances in and on the artery walls. Preventative methods often involve regular exercise and smoking cessation. Treatment often involves medications, but for some patients, surgery is necessary.2
CAD is the leading cause of morbidity and mortality worldwide, according to the investigators. It is imperative that predicting individual risk is accurate for prevention. Age, according to the investigators, is the main method of estimating one’s risk of CAD. Additionally, there are effective preventative treatment methods that exist; however, the authors noted that these are often underutilized because patients do not know they are at risk of CAD until it is too late. For these reasons, the investigators aimed to integrate unmodifiable risk factors (eg, age, genetics) along with modifiable risk factors (eg, clinical and biometric measurements) into a meta-prediction framework that can create actionable and personalized CAD risk estimates for the individual patient.1,3
“We tend to observe bad outcomes for CAD in patients who are in their mid-50s and older, but the disease actually begins to develop much earlier, sometimes even while people are teenagers. There's a lot of room for us to take action,” first author Shang-Fu ‘Shaun’ Chen, former doctoral student, Scripps Research translation Institute, explained in a news release.3
For the initial development of their predictive model, the investigators used approximately 2000 predictive features, including demographic data, lifestyle factors, physical measurements, laboratory tests, medication usage, diagnoses, and genetics. Additionally, to power the meta-prediction approach, data from UK Biobank was utilized and stratified into 2 primary cohorts, including a prevalent CAD cohort used to train the models for cross-sectional prediction at baseline and prospective estimation of contributing risk factor levels and diagnoses, and an incident CAD cohort that used these baseline models as meta-features to train a final CAD incident risk prediction model.1
The investigators observed that their new model, which was composed of 15 derived meta-features with multiple embedded polygenic risk scores, had achieved an area under the curve (AUC) of about 0.84. Using an independent test cohort from the All of Us research program, an AUC of 0.81 was achieved for predicting incident 10-year CAD risk.1
Specifically, the investigated model outperformed the standard clinical model, enabling the prediction of 2 times as many CAD events. After a 10-year follow-up, approximately 62.9% of individuals that the model categorized as “highest risk” had developed CAD, compared with the lowest risk group (0.3%). The authors noted that the model’s accuracy was somewhat because of its better ability to predict CAD in populations of people who are typically deemed “low risk” (eg, women, younger individuals).1,3
“Our model can pick up individuals who would be considered at low risk of CAD due to their age, but who are actually high risk due to their underlying genetics,” senior author Ali Torkamani, PhD, professor and director of Genomics and Genome Informatics at the Scripps Research Translational Institute, said in the news release. “The higher your genetic risk for one of those traits—high cholesterol levels or high blood pressure levels or high diabetes risk—the greater benefit you get from intervening on that particular aspect through medication or lifestyle changes.”3
Although the model’s accuracy was dependent on the inclusion of multiple factors, the investigators observed that genetic predisposition was the strongest predictor of CAD risk. Not only did this include the genetic predisposition for CAD itself, but also for CAD-related conditions, such as hypertension, diabetes, and high cholesterol.1,3
“I think more precise and personalized risk prediction could motivate patients to engage in early prevention,” said Torkamani. “Our model first predicts the risk that a person will develop CAD, and then it provides information to allow personalized intervention.”3
This framework, wrote the authors, enables a generation of individualized risk reduction profiles by quantifying the potential impact of standard clinical interventions. Further, the investigators wrote that they intend to conduct a long-term clinical trial to test whether informing patients of the CAD risk can help disease prevention.1
“Compared to traditional clinical tools, the new model improved risk classification for approximately one in four individuals — helping to better identify those truly at risk while avoiding unnecessary concern for those who are not,” said Chen. “We think the most important thing is for patients to be aware of their individual risks so that they can receive the appropriate treatments and make lifestyle changes.”3
