Even a minor typing error can have serious, sometimes fatal, impacts on AI diagnostic results.
In healthcare, even a minor typing error by a healthcare professional can have serious, sometimes fatal, impacts on AI diagnostic results.
AI systems (such as disease diagnosis, drug prescribing, or clinical decision support systems) rely heavily on accurate data.
But when these systems are fed incorrect, incomplete, or distorted information, they can lead to misdiagnoses, the wrong medication, the wrong patient outcomes, and research or predictive models.
For example, typing “Hypotension” instead of “Hypertension” can result in the wrong blood pressure medication being prescribed.
Typing a drug name as “Prednisolone” instead of “Prednisone” can result in different dosages and side effects. Similarly, typing “No diabetes” as “Diabetes” can result in insulin being prescribed. Typing a patient’s weight as 700 kg instead of 70 kg can result in dangerously high doses of medication being prescribed.
What happens in AI?
This is because AI systems do not ask questions in return or in return. They only tell you what you ask them or enter a command. If the command or question is wrong, regression models or neural networks will give an incorrect result.
