Houston Methodist Hospital researchers created an artificial intelligence-powered risk calculator that predicts the likelihood a patient will develop breast cancer and recommends whether a biopsy should be performed, according to a Nov. 9 Culture Map Houston report.
Six details:
1. Stephen Wong, PhD, biomedical engineering chair at Houston Methodist, led the development of the risk calculator, dubbed iBrisk, which builds on the national breast cancer risk and biopsy decision-making standard the Breast Imaging Reporting and Database System score.
2. As an expansion of the BI-RADS system, Dr. Wong used multiple patient data points and AI tech to create iBrisk, which integrates natural language processing, medical image analysis and deep learning on the BI-RADS patient data to make one of three recommendations: biopsy not recommended, consider biopsy or biopsy recommended.
3. The tool aims to alleviate unnecessary biopsies, which Dr. Wong said of 10,000 women biopsied, less than two will be positive while using the current BI-RADS system.
4. The researchers determined approximately 100 parameters for iBrisk to analyze including age, sex, socioeconomic data, medical history and insurance plans. After applying the deep-learning method, the AI reduced the data points to 20 risk indicators.
5. The hospital used about 11,000 cases for training and then 2,200 cases of its own data to test iBrisk. It also worked with other hospitals including MD Anderson to independently test the tool and confirm its results.
6. The risk categories are broken into low, moderate and high-risk groups. Of the low-risk groups, 99.8 percent of individuals' results have been accurate. Patients in the high-risk groups, which prompt an order for a biopsy, have seen 85.9 percent accuracy compared to radiology, which is 25 percent accurate, Dr. Wong said. The moderate risk category saw 93.4 percent accuracy.