According to research published in European Journal of Gun,The fairness and fairness of datasets for AI-driven mammogram interpretation can be at risk by underestimating racial and ethnic diversity.
While AI promises to strengthen how mammograms are interpreted, especially in areas with limited resources, the study authors have found warning signs about dataset diversity and researcher representation in AI model development.
In this study, researchers conducted a scientific review of studies published in 2017, 2018, 2022 and 2023, and conducted a screening for breast cancer detection or using diagnostic mammograms to “train or validate AI algorithms.”
Of the 5,774 studies identified, 264 met the inclusion criteria.
“The number of studies increased from 28 in 2017-2018 to 115 in 2022-2023. This increased by 311%. Despite this growth, most patients were identified as white in only 0-25% of studies reported race/ethnicity,” the study author wrote.
Furthermore, almost all patient cohorts were born from high-income countries and there were no studies from low-income settings. Author affiliations were primarily from high-income areas, and gender imbalances were observed between the first and last authors. ”
The authors concluded that “the lack of racial, ethnic and geographical diversity in both datasets and researcher representations may undermine the generalization and fairness of AI-based mammogram interpretations.”
Furthermore, recognizing disparities through diverse dataset collection and comprehensive international collaboration is important to ensure fair progress in breast cancer care.
Research data revealed that algorithms focusing primarily on white populations could lead to inaccurate outcomes and false diagnosis of underestimated populations. Additionally, patient outcomes can be threatened and current disparities may worsen.
“The fairness of these AI tools is being questioned because they risk systematically defending a particular race, ethnic or sociodemographic group, reducing these issues, ensuring that the benefits of AI in BC imaging are equitably distributed, and ensuring that it is essential to ensure that it includes international collaborations, including international coordination, including international coordination of dataset collections, clinical research,” the study author wrote.
Bigger trends
February, Google partnered with the Women’s Cancers Institute, founded by the French Cancer Research and Treatment Center Institute Curie to study how AI tools can deal with cancer, share science-based health information, and fund postdoctoral researchers.
Two entities investigated how AI-based tools can help predict cancer progression and potential recurrence in patients, with the goal of developing more accurate and successful treatments.
Researchers focused on the difficulties in treating cancer in women, including triple-negative breast cancer, an aggressive type of breast cancer that grows faster than other types.
2024, AI Biotech Company Owkin has partnered with Pharma Giant Astrazeneca to develop an AI-driven tool designed to pre-screen breast cancer GBRCA mutations (GBRCAMs) directly from digitized pathology slides.
The purpose of this tool is to speed up and increase access to GBRCA tests, which may not be considered by some patients.
That same year, Lunit, a provider of AI-driven solutions for cancer diagnosis and treatments, is a company that provides AI-powered software that helps providers better understand cancer risks, and has joined forces to develop a comprehensive ecosystem of early cancer detection, cancer risk prediction and independent AI to improve clinical workflows.
In May of that year, Lunit acquired Volpara and integrated the AI breast health platform, including the Scorecard Breast Density Assessment Tool, into a line of AI tools for breast cancer detection.
Before acquiring Bolpara, Lunit partnered with one of the nation’s largest private health care providers to support it. Improves cancer screening capabilities in Sweden.
In 2023, Lunit signed a three-year agreement with Capio S: TGöran Hospital to supply and license the AI-powered mammography analysis software Lunit Insight MMG. The AI tool allowed hospitals to analyze breast images of approximately 78,000 patients each year.
