Higher time-dependent area under the curves seen for five AI algorithms compared with Breast Cancer Surveillance Consortium risk model
By Elana Gotkine HealthDay Reporter
THURSDAY, June 8, 2023 (HealthDay News) — Mammography artificial intelligence (AI) algorithms perform better than the Breast Cancer Surveillance Consortium (BCSC) risk model for predicting breast cancer risk, according to a study published online June 6 in Radiology.
Vignesh A. Arasu, M.D., Ph.D., from Kaiser Permanente Northern California in Oakland, and colleagues compared existing mammography AI algorithms and the BCSC risk model for predicting five-year breast cancer risk in a retrospective case-cohort study of women with a negative screening mammographic examination in 2016 who were followed until 2021. A random subcohort of 13,628 patients was selected from a cohort of 324,009 eligible women, regardless of cancer status, to which all additional patients with breast cancer were added (additional 4,391 women). Continuous scores were generated using the index screening mammographic examination as input for five AI algorithms; these were compared with the BCSC clinical risk score. Risk estimates were calculated using a time-dependent area under the receiver operating characteristic curve (AUC).
The researchers found that the time dependent AUC for BCSC was 0.61 for incident cancers at zero to five years. Higher time-dependent AUCs were seen for the AI algorithms versus the BCSC, ranging from 0.63 to 0.67. For combined BCSC and AI models, time-dependent AUCs were slightly higher than AI alone (0.66 to 0.68).
“AI for cancer risk prediction offers us the opportunity to individualize every woman’s care, which isn’t systematically available,” Arasu said in a statement. “It’s a tool that could help us provide personalized, precision medicine on a national level.”
Several authors disclosed ties to industry.
Copyright © 2023 HealthDay. All rights reserved.
