Ethics of Deep Learning in Health

How accurate are algorithms compared to doctors? In some areas like cancer detection, algorithms and AI do a better job than doctors. In the area of diagnosing a patient based on the way they walk, talk, smell, etc. algorithms do a poor job compared to doctors. As time goes by, algorithms and AI will only improve. In fact, one deep learning algorithm has already been approved by the FDA: Artery’s deep learning DeepVentricle algorithm analyzes MRI images.

Is deep learning in Health Care ethical? Let’s consider the Utilitarian Approach. Would it do the most good and least harm if deep learning is used in health care? The parties involved are the doctors and the patients. Good that comes from using deep learning in health care includes improved reading of medical images like MRI’s, improved cancer detection, and reduced costs in areas like diagnosis. Harm that comes from using deep learning in health care includes decisions being made without reasoning being documented. Another harmful aspect of deep learning in health care is that innovations may be hidden inside deep learning’s black box. Overall, deep learning in Health Care is ethical because a patient’s health outweighs any costs of deep learning.

Are there any ethical issues about the use of patient data? Yes, an ethical issue is: should a patient have control over how their data is used? Let’s consider the Rights Approach. Which action respects the rights of all who have a stake in the decision? While making all health care data public would be a great tool for research, the patient’s privacy is also important. In order to best respect the rights all involved, a patient’s data should only be used with permission.