Addressing Bias in Artificial Intelligence

In discussions surrounding Artificial Intelligence (AI), bias has been one of the major concerns raised when contemplating AI’s use in health care. Research has shown that results produced through AI may not take into consideration factors that impact certain populations or communities because the information or data used to develop those results may not be representative of everyone.  As a result, a group of hospitals and university health technology experts have recently come together in an effort to address these shortcomings.
 
VALID AI began last year to bring together health care systems and research facilities to establish industry standards for AI.  Standing for: Vision, Alignment, Learning, Implementation and Dissemination (VALID), the effort is intended to “facilitate collaboration across the public and private sectors and promote co-transformation at the national level”. VALID AI is led by two AI digital health specialists at the University of California, Davis, Dennis Chornenky and Ashish Atreja.  Their latest call to action is to work toward integrating social determinants of health (SDOH) as social vital signs (SVS) at the point of care.  In their most recent press release, a founding member of VALID AI states that the organization’s
 

 
 
Currently, the effort has over 50 members including Cedars Sinai, Union Health, Boston Children’s Hospital and Atlantic Health System. It is the goal for members to co-host in each state sandboxes and workstreams to help facilitate addressing challenges that impede the widespread adoption of “social vital signs in clinical practice.” An ultimate goal would be development of a Gen AI SDOH Toolkit.  For more information, visit the  VALID AI website where you can also join their efforts.