Changes in AI-based risk scores identify women at greatest risk of developing breast cancer
Short-term changes in artificial intelligence-based breast cancer risk scores could give providers valuable insight into patients’ likelihood of developing cancer.
In research shared during the annual Society for Imaging Informatics in Medicine’s annual meeting, Mark Traill, MD, a breast radiologist at University of Michigan Health West, presented new data related to changes in patients’ AI-derived cancer risk assessments.
“Image based breast cancer risk assessment using AI algorithm has shown to outperform traditional risk models,” Traill and colleagues suggest, adding that changes in these scores over time could provide equally valuable insight.
Based on the experts’ analysis, the more an individual’s score changes, the greater the odds are of them being diagnosed with cancer in the near future.
To measure the exact effect of fluctuating risk scores, the team analyzed the scores 514 controls (including 52 diagnosed with cancer) were assigned by an AI short-term risk model based on their prior and current DBT mammograms. The model uses different mammographic features, breast density status and age to predict patients’ one-year absolute risk of being diagnosed with breast cancer. Risk score change, prior risk score, age at prior mammogram and years prior to current imaging were all incorporated into a multivariate model to calculate updated cancer risks.
The AUC for the multivariable model with risk score change was 0.88. For every 0.2 unit increase from prior to current scores, patients’ risk of being diagnosed with cancer was around two times higher. Of the patients who were diagnosed with cancer, 63.5% had an increase of at least one quartile from prior to current.
Based on these findings, the team determined that “an increased likelihood of short-term cancer diagnosis based on changes in risk scores may warrant supplemental screening or clinical interventions.”
The SIIM 2024 conference is being held now in National Harbor, Maryland and will conclude on June 29.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She joined Innovate Healthcare in 2021 and has since put her unique expertise to use in her editorial role with Health Imaging.