Ensuring Non-discriminatory Use of Artificial Intelligence in Medicine
The WSMA adopts a position on the use of artificial intelligence in
medicine, stating that any AI that is patient-facing should be transparent
to patients so that they know an automated system is being used. (Res B-3,
A-23)
The WSMA supports the development of physician-guided guidelines defining
“patient-facing AI” that will consider direct patient interactions, AI’s
role in information retrieval and interpretation, and its influence on
decision-making to strike a balance between transparency to patients and
the feasibility and necessity of disclosing AI’s involvement when
physicians remain the final decision-makers, reviewers, or editors of the
information provided to patients. (Res B-3, A-23)
The WSMA supports efforts to align AI systems with the dynamic medical
field, fostering collaboration and ongoing updates for effective health
care AI integration. (Res B-3, A-23)
The WSMA supports requiring patient-facing AI systems to be interpretable
to the extent possible by human coders and users to ensure that the system
operates within reasonable parameters and aligns with the intended design
while avoiding obvious failures, misdiagnoses, or adverse consequences due
to algorithmic decision-making under uncertainty and helps avoid
malalignment. (Res B-3, A-23)
The WSMA encourages physicians to actively engage in a continuous feedback
loop with AI systems to improve accuracy, learn from and correct errors,
and continuously enhance the performance of patient-facing AI because
physicians' feedback will be instrumental in addressing errors and
optimizing AI utilization in health care settings. (Res B-3, A-23)
Return to the WSMA Policy Compendium index
Abbreviations for House of Delegates report origination:
EC – Executive Committee; BT – Board of Trustees; CPA – Council on
Professional Affairs; JC – Judicial Council; CHS – Community and Health
Services