How Machine Learning Strengthens Real-World Data and Disease Identification: 5 Use Cases in 5 Minutes

By Mike Munsell, PhD Director of Research, Panalgo

Machine learning is a powerful tool in the life sciences industry, making evidence generation more efficient and precise and bolstering real-world data (RWD) to identify diseases earlier, trigger timely interventions, and recommend screenings and physician referrals. Our blog series, How Life Sciences is Leveraging Machine Learning: 5 Use Cases in 5 Minutes, explores five recent studies that used machine learning to analyze RWD or improve clinical practice.

This edition of 5 in 5 includes examples of using machine learning to predict a patient’s functional status in electronic health records (EHR) data, using natural language processing to identify social determinants of health data from EHR, using claims data to better understand the clinical predictors of anaphylaxis, using machine learning to better diagnose and treat Gauchers disease, and using computer-based EHR prompts to recommend smoking cessation to cancer patients. Each of these use cases illustrates the power of machine learning and RWD in predicting disease states and improving patient outcomes.

  1. Duke researchers used machine learning to create an algorithm that helps predict a patient’s functional status in EHR data. Functional status greatly improves RWD analyses by allowing researchers to study disease severity over time, however, it is generally not captured in administrative data sets. This algorithm has the potential to improve the accuracy of RWE by broadening our understanding of an interventions potential effect.
  2. NYU, Michigan, and Johns Hopkins researchers used natural language processing to extract and categorize social determinants of health data from EHR. Similar to the above example, RWE can be improved by understanding these important variables that impact access to care, however, they are rarely available in practice.
  3. The FDA and CMS used claims data to train a model for the identification of anaphylaxis to better understand its clinical predictors.
  4. Sanofi used Optum data and machine learning to better predict Gaucher disease to help with early diagnosis and treatment.
  5. The University of Pennsylvania tested algorithms in practice for assisting clinicians. A computer-based EHR ‘nudge’ (i.e., prompt) was used to recommend smoking cessation to cancer patients, with successful results.

While these examples tackle complex problems, machine learning enabled researchers to get to insights faster, even simplifying or streamlining the study process. Machine learning is a valuable instrument in healthcare and healthcare analytics, and it’s important that we see it as such.

Panalgo’s IHD Data Science module is a powerful machine learning analytics tool built on our self-service, point-and-click platform that allows users to uncover new insights and produce more accurate prediction and segmentation models, similar to the analyses outlined above. If you would like to learn more about how you can leverage machine learning analytics with the IHD Data Science module, contact us today.