Prostate cancer is one of the leading causes of cancer-related deaths for American men. Sadly, it affects African American men at a higher rate than any other ethnic group.
This health disparity can be the result of inadequate health care limiting early cancer detection or incomplete treatments. Other factors are the invasive nature of prostate specific antigen tests, the limited availability of tests, and a tendency for false results in those screenings.
Vibhuti Gupta, Ph.D., assistant professor, computer science and data science, recently received a $1.2 million grant from AIM-AHEAD to leverage artificial intelligence and machine learning (AI/ML) to develop a multimodal framework for prostate cancer risk prediction and address health disparities.
Dr. Gupta’s project will improve on current AI/ML models that are typically trained on a single modality, either imaging, genomics or clinical data.
That approach, he says, does not sufficiently capture the heterogeneity and variability of prostate cancer malignancies to tailor medical care and improve personalized medicine.
“The model will systematically integrate clinical, demographic, genomics, and imaging data to predict various prostate cancer risk groups based on Gleason scores, clinical, and pathological stages,” says Dr. Gupta.
Dr. Gupta will improve health outcomes by incorporating Social Determinants of Health data—such as socioeconomic status and access to care—and by applying techniques to ensure that all relevant groups are accurately reflected in the data.
“If a model is built on flawed or incomplete data, it can produce poor results that negatively affect clinical treatment, so those steps are crucial for our model,” he explains.
