A collaboration between Sandia and the CDC data science group is pioneering the use of machine learning (ML) for epidemiological purposes in collaboration with researchers at the University of New Mexico (UNM) to improve methods of analysis for public health related data.
Together, the partners have developed powerful new ML algorithms that excel at finding patterns in textual data, such as physician notes, despite abbreviations, misspellings, and variations in words and phrases used at different hospitals. ML-based data fusion—combining different categories of information such as text and numbers—can better find subtle patterns in health data.
New ML methods detect some anomalies in health data better than traditional methods; the algorithms can also adapt as health issues and indicators change over time. Methods developed within this collaboration could eventually better protect the nation from disease threats. Researchers are now working to identify other public health issues that could benefit from these methods. Artificial intelligence, supported by ML methods, can potentially provide faster and more accurate results for a host of health-data analysis problems, such as identifying autism spectrum disorder from students’ behavioral evaluations or finding early signs of suicidal intent from patient records to enable earlier intervention.