Question

make a concept map about artificial issues intelligence in the health field?

Answer 1

Artificial intelligence issues in healthcare:

• The first issue is data. Training data is essential data for machine learning. In fact the volume of data needed can exceed what many organizations generate. There are many organizational barriers due to regulation, security and incompatible ways of storing, accessing and managing he data. There are also privacy issues and incorrect assumptions made preventing establishment of static ground truth thus reducing its efficacy for training.
• The second issue is lack of transparency: there are limited strategies to uncover way a decision was made by AI. Determining the relevant variables for AI is also not easy. Methods such as principal component analysis (PCA) are used to select variables from large set of variables. There are challenges in the field of medicine and aviation where there are minimum errors and decision needs to be highly confidential. The approach to mitigate this is to understand the model’s limitations and ensure consistency in decisions made by the people.
• The third issue is governance. Inserting an AI is very challenging in decision making mechanisms of the organization. There are many questions to be answered like why decisions of AI has to be trusted, feedback process to make changes in AI, is the staff capable enough of making use of AI, are there sufficient skills to understand difference between correlation and causation etc. for example, children don’t drink coffee like adults. Hence it is easy to construct relationship between height and coffee. Where people can predict that tall people can drink more coffee than short people.
• Regulatory and ethical issues: both HIPAA & GDPR place constraints on the type of data that can be made public. When and how AI can be used is a concern. As AI is more sophisticated, ethical issues arise. If AI is used to prevent some people from getting insurance there can be reasons for that. Or if AI determines that a person has some disorder by correlating public sources or taking information from social networks then an argument can be made that this violates the privacy rights or ethically questionable.