Unboxing the machine learning black box in healthcare applications
Published:
Machine learning models are being quickly adopted in various industry applications because of their outstanding performance compared to traditional models such as linear regression. However, the black-box nature of many machine learning models casts doubts on their usability in practice, especially in the healthcare field. As noted by Lipton (2017), “the doctor just won’t accept that!” When health or even life is at stake, it is hard to trust a black-box algorithm that does not provide sufficient explanations for its decisions.