Using electronic health records to predict health risks associated with obesity

It’s widely accepted that being overweight or obese has a negative impact on a person’s health, especially in the long run. However, it’s not clear for a given individual what exact health challenges he or she will most likely face in the future and what would be the best treatment to avoid them.

Novo Nordisk develops medicines to treat obesity as well as associated diseases, such as diabetes, heart disease, and kidney disease. In order to better understand these diseases, select the right patients for clinical trials, and negotiate product prices, the company mines electronic health records (EHRs) from both UK and US healthcare providers to discover common characteristics of patients with obesity, learn about other diseases they have developed, and learn the medications they’re being treated with. This historical data is used to build machine learning models to predict subjects’ risk of developing certain health issues in the future.

In an organization that traditionally works with data generated in experiments or clinical trials that were designed to answer one specific question, it’s challenging to convince coworkers of the value of mining observational data. Furthermore, there are a number of technical challenges that need to be addressed, from choosing the right IT infrastructure and the most efficient data organization to selecting the most feasible analysis method.

Volker Schnecke explains how data from EHRs are used to improve our understanding of obesity as a disease and its complications and support the discovery and development of new medicines.