Machine Learning Methods Can Help Predict Lupus Hospitalizations
A new study shows that machine learning can be used to help predict risk of hospitalization in people with lupus. Machine learning refers to the process by which a machine or computer can imitate human behavior to learn and optimize complicated tasks such as statistical analysis and predictive modeling using large datasets. This is an exciting new area of research as more and more health record data is stored in computers, allowing for rapid and complex statistical analysis.
In this study, researchers used multiple machine learning techniques and applied them to the electronic health record data of nearly 2,000 individuals with lupus. They assessed people’s demographic information, laboratory test results, medications, lupus symptoms and their use of different healthcare services.
Overall, 4.6% of the people included in the study were hospitalized for lupus in their most recent year of clinical follow-up. The top predictors of hospitalization included:
- dsDNA positivity (indicating the presence of double-stranded DNA antibodies, which signal a more severe form of lupus)
- Complement C3 levels (a type of complement, or protein, that can combine with antibodies to destroy other cells)
- Blood cell counts (measures of red and white blood cell levels, which can indicate underlying disease)
- Markers of inflammation (a bodily response to injury or disease in which heat, redness and swelling are present)
- Albumin levels (a protein made by the liver which can be measured to screen for liver or kidney disease or other medical conditions)
The findings suggest that machine learning approaches may offer helpful ways to predict hospitalization risk in people with lupus.
Interested in getting research like this straight to your inbox? Subscribe to our bimonthly Inside Lupus Research email for all the latest.