AI, Ethics, and Healthcare
According to the AMA Journal of Ethics, “An artificially intelligent computer program can now diagnose skin cancer more accurately than a board-certified dermatologist,” and it can do so much more efficiently. All the program requires is some training data to make it smarter than a doctor and testing data to ensure its skills are up to par.
All fields of Artificial Intelligence have the capability to be applied in every field of healthcare. Using neural networks to diagnose diseases based on x-ray training data to NLP is used to classify diseases based on lab tests and write-ups. However, there should be some ethical considerations before having hospitals dive into every new technology being created.
One large consideration that needs to be made is ensuring patients' privacy when creating models based on clinical data. For example, when clinical data is used when building these AI models, the data should be given to researchers and developers only after patient identification has been removed, and those handling the clinical data should be trustworthy, in that they won’t sell this data to the highest bidder. Also, patients should definitely have a say in the usage of their data when making these clinical datasets, and they should have the right to say no if they don’t want researchers to see their data. Another deliberation that should take place is that the patient data should not be owned by any specific entity. Instead, it should be shared amongst all those who choose to use it. Private data should be safeguarded at all times, and if the data comes to light, then it should be discarded.
A further element of AI ethics in healthcare that has to be considered is the impact of AI on patient safety, and how we can guarantee the safety of patients, as a first priority. To make certain that patients are safe, the first step is to train the AI models with unbiased, and untampered data, so that the model has correct and unfiltered data to make a diagnosis that can lead to the patient’s survival. Secondly, training data should be from a diverse and varying field of patients, so that model makes a prediction with a vast amount of biodiverse data, instead of having to make a prediction based only on one race, or gender, or any other trait. Informing patients of incorrect diagnoses should also be an important part of the medical process.
While AI can be used in a large portion of the healthcare industry, the usage of AI should be under continuous scrutiny to ensure the safety and privacy of the patient. Legislation to guarantee these basic rights should also be put into place, to further help AI be used safely, and correctly in practice across hospitals around the world.
Thank you for reading the first part of my series! In the next part, we will explore the ethics of using AI in another field, perhaps regarding social media.
Arjun Santhanakrishnan is a Student Ambassador in the Inspirit AI Student Ambassadors Program. Inspirit AI is a pre-collegiate enrichment program that exposes curious high school students globally to AI through live online classes. Learn more at https://www.inspiritai.com/.