Industry Insights AI
This round of insights explores some of the latest advancements and discussions in the application of artificial intelligence (AI) in healthcare.
AI to reduce futile donation after circulatory death (DCD) liver procurements in the USA
The non-utilisation of DCD livers are high, as donors do not progres to death after withdrawal of care, and transplant centres decline grafts if the donor does not progress to death within a specific timeframe. This results in futile procurement with associated financial and operational burdens. Stanford researchers have developed a machine-learning model to predict DCD-donor progression to death. This LightGBM model significantly outperformed surgeon’s predictions, the DCD-N score, and the Colorado Calculator, reducing the incidence of futile procurements by 60% compared with surgeon predictions. They note that further improvements are needed to reduce missed opportunities and improve the overall accuracy of LightGBM.
American Psychological Association health advisory on use of generative AI chatbots and wellness applications for mental health
More people are using generative artificial intelligence chatbots for mental health support. Although they are easy to access and low cost, these are not clinically validated, regulated or have adequate safety measures. In response, the American Psychological Association health advisory has offered eight recommendations in response to this. Key recommendations include avoiding the use of GenAI chatbots and wellness apps to deliver psychotherapy or psychological treatment and prevent unhealthy relationships and dependencies between users and Gen AI chatbots and apps. They also include recommendations for the public, policymakers, developers, clinicians and other stakeholders to help them address these concerns.
Clinicians view generative AI (GenAI) technologies as beneficial for medical decision-making, but evaluate peers who use them negatively
A recent study evaluated how physicians view the use of GenAI in medical decision making by presenting three scenarios depicting a physician: using no GenAI, GenAI as a primary decision-making tool and GenAI as a verification tool. Physicians that were more dependent on AI were viewed more negatively, rated as less clinically skilled, less competent and delivering a lower quality healthcare experience. Framing GenAI as a verification tool reduced some negative perceptions, but did not fully reduce them. These findings highlight a barrier that may have significant implications for the development and deployment of AI tools in medicine.
Want to get in touch? Contact us at https://elion.nz/get-in-touch/
