In a rapidly evolving healthcare landscape, AI has become a key focus for health systems and healthcare vendors. However, the ability to harness its full potential could be hindered by staffing challenges within IT departments.
Dennis Chornenky, chief AI adviser at Sacramento, Calif.-based UC Davis Health, told Becker's that there are gaps in IT talent across hospitals, even those that are well-resourced.
"Most hospitals struggle with securing the IT talent needed to deploy AI solutions, let alone build and design them in-house," he said.
This challenge will intensify as companies like Epic begin to roll out a wide range of AI-enabled features.
"Epic alone has something like 80 to 100 new AI-powered capabilities set to roll out soon," he said. "Health systems will need to quickly evaluate these capabilities to ensure they aren't introducing risks."
Mr. Chornenky emphasized the importance of health systems developing internal structures to manage the complexity of AI integration. One key recommendation is for IT departments to establish "model ops committees" to ensure AI deployments are safe, secure and effectively monitored.
"Without a way to quickly stress test and evaluate these AI capabilities, health systems could be introducing risks, especially from a governance perspective," Mr. Chornenky said. "The ability to evaluate and govern these technologies will be crucial as more AI-enabled features become embedded in EHR platforms like Epic."
He also stressed the importance of incorporating "red teams" within IT departments — teams dedicated to identifying vulnerabilities in AI systems
"Red teaming is becoming essential," Mr. Chornenky said. "This involves stress testing AI applications by exposing them to scenarios they may not have been trained on, to see how they respond."
As more AI functionalities are integrated into existing EHR systems, Mr. Chornenky warned that IT departments will need to adopt new strategies to keep pace.
"The influx of AI-enabled capabilities from companies like Epic will force health systems to react quickly," he said. "The only way to keep up will be to invest in the necessary talent and internal processes to evaluate and manage these technologies."