Although many people are dubious about artificial intelligence’s clinical capabilities, healthcare companies and researchers still are looking into how AI can affect the industry.
Two recent — but unrelated — studies tout AI’s potential to both diagnose and recommend treatment for diseases, including COVID-19.
If AI could accurately predict care management for COVID patients, that could be of major benefit to senior living and care providers. Despite moving beyond the pandemic, COVID remains a threat to residents.
The COVID study, which was conducted by a research team at the University of Chicago, trained AI on chest X-rays to see if the tool could predict when, or if, a COVID patient would need ICU treatment.
The study revealed statistically significant results for AI’s ability to predict the need for ICU care.
The findings show that AI could play a role in supporting clinical decisions based on “situations of limited resources,” the study authors state, adding that “our proposed model may be useful for efficient patient triage and for low-resourced regions that need to prioritize care, knowing who to treat immediately during a pandemic.”
Although older adults aren’t more likely than other age groups to contract COVID, they are far more likely to have more severe symptoms, or to die from the disease, especially if they are unvaccinated.
A separate study showed that ChatGPT could diagnose any disease with an average accuracy of 72%, a rate the study authors call “impressive” and approximately on par with the success rate of a recent medical school graduate.
Despite the findings, the authors noted ominously that “even small errors in clinical judgment can result in adverse outcomes.”
Beyond the initial diagnosis, ChatGPT did not show some of the biases experts are worried about, but was found to be slightly less accurate — 68% — in recommending clinical decisions like what medication to prescribe.
An overwhelming majority of older adults currently are uncomfortable with the idea of AI-led healthcare decisions.
