Artificial Intelligence in Medicine
Software can now certain cancers on a scan as accurately as an experienced radiologist, and considerably faster. Hospitals under pressure see an obvious in using it to sort urgent cases from routine ones. The difficulty is that these systems are trained on historical data, and any bias in that data will be rather than corrected. A model trained mainly on one population may badly for another. There is also the question of who is responsible when the software is and a patient is harmed. Most regulators have settled, for now, on requiring that a human clinician the final decision.
