Personalized disease risk prediction holds great promise for focused prevention, early intervention, and patient empowerment. The recent advent of polygenic risk scores now allows early risk prediction based on genetic information. However, there are multiple challenges in turning raw predictions into clinical information, ranging from statistical issues, regulatory aspects, risk communication and transferability.

Here, we describe the implementation of genetic risk prediction at MyHeritage, a global direct-to-consumer genetic testing company. Our database is one of the largest of its kind, including over three million genetic profiles and hundreds of thousands of questionnaires. We leveraged our data to develop a clinical risk score for a range of common complex conditions, including heart disease, breast cancer, type 2 diabetes, hypertension and obesity. We validated our predictions by multiple means to guarantee their accuracy and relevance.

This talk will demonstrate the process and challenges of transforming data science into an end-to-end product that empowers the lives of participants in the new world of affordable genetic testing.