Bias CT-AI Practice Questions (Page 5)

Questions 21–21

1. 21application · medium

   A hospital is deploying an AI system to predict patient readmission risk. The model was trained on electronic health records (EHR) from the hospital's own patients. The data scientists discover that the model performs well for patients with private insurance but poorly for patients with Medicaid. They suspect the model is using insurance type as a proxy for socioeconomic status. Which mitigation technique is most appropriate to address this issue?

   • ACollect additional data from Medicaid patients and oversample that group
   • BApply a fairness constraint during training that equalizes false positive rates across insurance groups
   • CUse a more complex model to capture the relationship between insurance type and readmission
   • DRemove the insurance type feature from the model and retrain
   Reveal Solution Discussion Select an answer first