RSNA 2021 Imaging Informatics Preview

Clinical decision-support software gets a helping hand from AI

By Erik L. Ridley, staff writer

Sunday, November 28 | 10:30 a.m.-11:30 a.m. | SSIN01-2 | Room E350
In this talk, researchers will share positive results from providing clinicians with artificial intelligence (AI)-assisted clinical indications when ordering imaging studies using clinical decision-support (CDS) software.

Although the Protecting Access to Medicare Act requires clinicians to consult appropriate use criteria (AUC) via a qualified clinical decision-support mechanism (CDSM) when ordering advanced imaging exams for Medicare patients, this type of software only provides an order appropriateness score when the physician selects a structured indication when entering the order, according to presenter Dr. Lauren Shreve from the University of Pennsylvania in Philadelphia and colleagues.

As a result, the researchers sought to assess the impact of offering structured indications generated by AI based on the free text entered by the clinician. After the AI tool was deployed, the percentage of scored orders increased from 31.8% to 50.9% and the percentage of free-text-only orders declined from 59.3% to 31.6%.

In addition, they found that the percentage of orders containing both structured and free-text only orders increased from 6.7% to 29.6%. Users pointed to a lack of an appropriate structured indication for 36.2% of the free-text only orders submitted after the AI tool was adopted.

"Free-text imaging orders can be reduced using an AI tool; however, additional measures will be needed to address the half of orders which remain unscored," the authors wrote.

Delve further into the results by attending this Sunday afternoon presentation.

Last Updated er 11/18/2021 4:02:51 PM