AI IN MAMMOGRAPHY: +4% RECALL WITHOUT INCREASE IN CANCER DETECTION AND INCREASE IN FALSE POSITIVES
A new analysis of the practical application of AI for interpreting mammograms in women aged 40 and older shows mixed results. PCORI and UCLA conducted a randomized clinical trial to evaluate the impact of the algorithm on recall rates and diagnostic accuracy.
The results showed that the implementation of the solution increased the recall rate by approximately 4%, while not significantly increasing the number of confirmed cancer cases. An increase in false positives leads to unnecessary examinations and costs in medical facilities 💼 and creates additional stress for patients 🏥. This demonstrates the importance of balancing the expected benefits and potential risks when integrating AI solutions.
In addition, scientists point to the problem of algorithm bias: the training data set was limited in terms of various demographic indicators, which may impair the quality of diagnosis in different population groups. Experts recommend conducting large-scale clinical trials in real-world settings and carefully analyzing the economic and ethical aspects of innovations.
A qualitative assessment of AI prior to mass implementation will help managers of companies and medical institutions make informed decisions about investing in such technologies.
Source of material – https://www.statnews.com/2025/09/24/breast-cancer-artificial-intelligence-mammogram-pcori-ucla/
#artificialintelligence #healthtech #medicalai #breastcancer #pcori #ucla