Turn artificial intelligence into health care intelligence.
Roughly 80 percent of data in electronic medical records consists of physicians’ unstructured notes. To unlock this important data, as well as the data in medical devices and other formats, we need a different approach than what we use to analyze structured data.
That's one place where artificial intelligence comes in. The cutting edge of today's analytic technologies are self-learning and self-improving with techniques such as deep learning. And the opportunity to solve bigger problems using new data science methods is large.
OptumLabs® has a history of building predictive AI models with regression and machine-learning methods. For example, we’ve constructed models to predict the onset of Alzheimer's disease, of diabetes, as well as models of patient clusters with heart failure and COPD.
We are now exploring the application of deep learning to the analysis of electronic health records. We're investigating the potential for natural language processing, a technique for extracting content from words in physician notes, to feed deep learning models.
AI can help us extract better insights, faster, from big data. Successful application of AI to health care problems requires a deep understanding of the strengths and limitations of health care data, and thoughtful implementation of the results.– Christopher Hane, Vice President of Data Science, OptumLabs