fbpx Skip to main content

Researchers assessed the accuracy of algorithm identification of individuals with type 1 diabetes who self-reported their race.

Identification of individuals with type 1 diabetes for biomedical research is increasingly being done with the use of automated algorithms and electronic health records. However, the accuracy of the algorithms is unknown when accounting for self-reported race. A study published in the journal Diabetes Care sought to determine if polygenic scores improve the identification of individuals with type 1 diabetes.

Algorithm Results Before and After Polygenic Scoring Were Compared With Medical Record Reviews

The biobanks of two large hospital-based biobanks, Mass General Brigham [MGB] and BioMe, were used. An established automated algorithm identified individuals with type 1 diabetes. Medical record reviews were used to validate the type 1 diabetes diagnosis. Two published polygenic scores for type 1 diabetes were implemented occurring in individuals of European or African ancestry. The researchers assessed how individuals were classified by the algorithm before and after incorporating polygenic scores.

Self-Reported Non-White Individuals Were More Likely to Be Incorrectly Assigned a Type 1 Diabetes Diagnosis

The study showed that the automated algorithm was more likely to incorrectly assign a diagnosis of type 1 diabetes in self-reported non-White individuals than in self-reported White individuals. When polygenic scores were incorporated into the MGB Biobank the positive predictive value of the type 1 diabetes algorithm increased from 70 to 97% for self-reported White individuals. This means that 97% of those predicted to have type 1 diabetes actually had type 1 diabetes. There was an increase from 53% to 100% for self-reported non-White individuals. Similar results were found in BioMe.

Disparity Could Potentially Be Reduced With the Use of Polygenic Scores

You May Also Like::  Strategies for Reducing Health Disparities in 2023

Automated phenotyping algorithms may cause further health disparities. This is due to the increased risk of incorrectly classifying individuals from underrepresented populations. This study determined that it is possible to use polygenic scores to improve the performance of phenotyping algorithms and potentially reduce this disparity.


Deutsch, A. J., Stalbow, L., Majarian, T. D., Mercader, J. M., Manning, A. K., Florez, J. C., Loos, R. J. F., & Udler, M. S. (2023). Polygenic Scores Help Reduce Racial Disparities in Predictive Accuracy of Automated Type 1 Diabetes Classification Algorithms. Diabetes Care. https://doi.org/10.2337/dc22-1833