|
|
AI |
|
|
|
AI, bias and healthcare |
|
|
The pros and cons of the rapid spread of AI/machine learning in healthcare, in the form of automation and decision support tools, has been a topic of interest this week. BMJ Health and Care Informatics published the report of a study looking at gender bias in the Indian Liver Patient Database, which is used to create algorithms to predict liver disease. The study found this bias would show up as an increased risk of misdiagnosis for female patients. The I newspaper website picked up on the study in an article that went on to examine the dangers of medical AI exacerbating existing inequalities.
Meanwhile, digitalhealth.net published an opinion piece from David Newey, deputy chief information officer at the Royal Marsden NHS Foundation Trust, which examined some potential sources of bias in more detail. Newey looks in particular at ‘temporal bias’ – or the risk of data, assumptions and algorithms going out of date, as information, norms and practice change. Newey argues the government should follow through on its commitment to develop a regulatory framework for the use of AI in the real-world, and that CIOs should form oversight committees to regularly review AI algorithms for applicability and bias. Otherwise, he warns, there is a risk of hard-baking-in inequality and social injustice. |
|
|
|
|
|