The Pittsburgh Health Data Alliance is building on its partnership with Amazon Web Services to develop new machine learning models for earlier breast cancer and depression detection, according to an Oct. 6 news release.
The alliance, which comprises UPMC, University of Pittsburgh and Carnegie Mellon University, partnered with AWS in August 2019 to collaborate on innovation projects in cancer diagnostics, precision medicine, EHRs and medical imaging.
Six things to know:
1. Through the collaboration, Shandong Wu, PhD, associate professor of radiology at University of Pittsburgh, is leading a research team that is using deep learning systems to analyze mammograms to predict short-term risk of developing breast cancer.
2. Dr. Wu and his team collected 452 de-identified normal screening mammogram images from 226 patients. Of the participants, half later developed breast cancer.
3. The researchers then used AWS machine learning models to analyze the images for characteristics that could help predict breast cancer risk. The team's two models could consistently outperform breast density measure, which is the primary imaging marker for breast cancer risk, by 33 percent and 35 percent.
4. UPMC and University of Pittsburgh researchers plan to pursue further studies with more training samples and imaging data to continue evaluating the models.
5. In a second project, Carnegie Mellon University and UPMC researchers are using AWS machine learning tools to develop sensing technologies that can automatically measure subtle changes in an individual's behavior, such as facial expressions, that can act as biomarkers for depression.
6. The researchers plan to later compare the new biomarkers with results of clinical assessments; the technology is designed to help support the clinician's decision making on diagnosis and treatment.