Latest / AI Ethics with Fexingo: Bias, Safety, and Responsible Artificial Intelligence
When Your AI Diagnosis Misses Rare Disease Patients
Lucas and Luna explore how AI diagnostic tools systematically underdiagnose patients with rare diseases—those affecting fewer than 200,000 people. They break down the data imbalance problem: rare diseases collectively affect 300 million people worldwide, but AI training datasets overwhelmingly favor common conditions. The hosts examine a 2025 study from the Journal of the American Medical Informatics Association showing that diagnostic AI models misdiagnose rare diseases at rates 30% higher than common ones, with false-negative rates reaching 60% for conditions with fewer than 1,000…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.