Evaluation of the Clinical Impact of Machine Learning-Based Risk Classification Using Blood Analysis on Iron Deficiency Detection

What this Trial Is About

Researchers are evaluating whether an AI-based risk notification system integrated into routine clinical care can improve the detection of iron deficiency in adults attending Internal Medicine, Family Medicine, and Hematology/Oncology clinics at China Medical University Hospital in Taiwan. This pragmatic randomized controlled trial aims to see if displaying AI-generated risk classifications to physicians affects iron deficiency detection rates, ordering of ferritin tests, and diagnostic efficiency, comparing outcomes between AI-assisted and standard care groups. The study involves two groups: the Prompt Group, where physicians see AI-generated risk hints categorizing patients as high or low risk for iron deficiency, and the Control Group, where physicians receive standard lab results without AI risk information. For high-risk patients in the Prompt Group, the system checks for recent iron testing and suggests consideration if none was done. The AI tool supports clinical decisions but does not mandate testing or treatment. Randomization occurs automatically after routine complete blood count (CBC) results become available. Participants undergo no additional procedures beyond usual outpatient care and routine CBC testing ordered by their doctors. The AI analyzes existing CBC data within the hospital system without extra patient input. Researchers will monitor iron deficiency detection rates overall and among patients with and without anemia, ferritin test ordering rates, diagnostic yield of ferritin testing, and cost-effectiveness. All data handling complies with institutional privacy policies, and participation lasts during routine outpatient visits without extra visits or tests.

Assessing Clinical Impact of AI for Iron Deficiency