How radiology departments can potentially save millions

Thursday, December 3 | 9:30 a.m.-10:30 a.m. | R3-SSIN07-6 | Room E450B

In this scientific presentation, researchers will reveal how a machine-learning calculator can help predict which patients will miss radiology appointments, potentially saving millions of dollars a year.

In a retrospective study at a large tertiary academic center, presenter Charit Tippareddy, MD, of University Hospitals Cleveland, and colleagues analyzed nearly 33,000 outpatient radiology appointments from 10,000 patients. They developed a random-forest model using 16 readily available clinical and social determinants of health variables, including the Area Deprivation Index (ADI), a measure of neighborhood socioeconomic disadvantage.

The model achieved an area under the curve of 0.76, outperforming logistic regression and LightGBM models. No-show rates differed 22-fold across predicted risk deciles, and the highest-risk 10% of patients accounted for about a quarter of all no-shows. Key predictors included distance to the facility, appointment lead time, age, comorbidities, and ADI, according to the group

Across 1.2 million annual appointments, missed imaging appointments cost the institution an estimated $57.7 million in revenue each year, the researchers found. They estimated that targeting the highest-risk 10% of patients could yield $2 million in net annual savings under a conservative scenario, and more than $6 million under higher-yield scenarios.

“Embedding no-show prediction into scheduling workflows enables targeted, equity-informed interventions that improve access to care while reducing operational and financial inefficiencies,” the researchers wrote.

If you want more details, you won’t want to be a no-show for this session.

Page 1 of 2
Next Page