- Common risk factors for breast cancer include family history and breast density. But oncologists argue that these do not paint a complete picture of a patient’s risk.
- In a new study, AI was more accurate than standard risk models in predicting breast cancer.
- Experts say a better understanding of patient risk factors is key to improving outcomes.
breast cancer is
Arasu wanted to change that and give patients a clearer understanding of their risks.
“Traditional risk factors that we’ve known for decades include a woman’s age, family history, previous benign biopsies, estrogen exposure, and breast density,” says Allas. says. “Identifying new risk factors will help identify women who would benefit from more cancer screening, with the goal of reducing advanced breast cancer diagnoses and breast cancer deaths.”
But how?
AI, the same technology that recently made headlines on ChatGPT, could be a key help in predicting a person’s breast cancer risk, according to new research led by Arasu and released Tuesday. Radiology,Journal of the Radiological Society of North America (RSNA).
The study, which included thousands of mammograms, is an AI model of the standard clinical risk model currently used to predict a person’s five-year risk of developing breast cancer, known as the Breast Cancer Surveillance Consortium. It has been shown that there may be more than one.
“This suggests that using AI alone or in combination with current risk prediction models will provide new avenues for future risk prediction,” said Arras. say.
Breast cancer experts not involved in the study hailed the study as promising for health care providers and their patients.
“AI is expected to help radiologists detect subtle breast cancers and potentially alert patients who may be at increased risk for breast cancer within the next decade.” Yale University Breast Imaging Machine (Radiologist) and Associate Professor, Yale School of Medicine.
The research also presents new use cases for AI.
“It’s a new way of looking at artificial intelligence,” says Nina Stuzin Vincoff, M.D., director of breast imaging at Northwell Health in New York. “We always thought of it as a way to make a discovery. This is a very interesting and important way for artificial intelligence to play a role.”
Arasu explains that the study is retrospective and that it looks back on what has already happened.
Arasu and his team began by identifying more than 324,000 women who underwent mammograms at Kaiser Permanente in Northern California in 2016 and showed no signs of breast cancer.
The team narrowed the participant pool to a random subgroup of 13,628 people for analysis.
“Then we looked at which women developed breast cancer between 2016 and 2021,” Aras explains. “We found that there were 4,584 women diagnosed with breast cancer. We compared these women to a subgroup that included her 13,435 out of 324,000 women who did not develop breast cancer.”
Researchers will follow all participants through 2021.
“We evaluated five artificial intelligence algorithms to generate negative mammogram scores for women taken in 2016,” says Arasu. “These scores are intended to detect breast cancer, but now we evaluated whether these same scores can predict future cancer risk five years into the future.”
“We also used the Breast Cancer Surveillance Consortium’s BCSC clinical risk model to assess breast cancer risk based on conventional risk factors in 2016,” Arasu added.
The Breast Cancer Surveillance Consortium (BCSC) is a commonly used model to predict breast cancer risk. A risk score is calculated using self-reported information from the patient and other factors such as age, family history of breast cancer, birth history, and breast density.
Is there one critical gap?
“There are a lot of factors that go into whether you have an increased risk of developing cancer, and you may not know them,” Vinkoff said.
For example, if you were adopted or estranged from your parents, you may not know all of your family history of breast cancer.
Can AI help change that?
We looked to see whether AI or BCSC performed better in predicting which women would be diagnosed with breast cancer,” says Arras.
It happened.
“This study demonstrates the potential of AI risk assessment models to enhance the identification of average-risk patients who are more likely to develop breast cancer in the next five years,” said Andrejeva-Wright. I’m here. “Furthermore, this study applied the BCSC risk assessment model in combination with the AI risk assessment model to help identify potential patient cohorts within the average-risk population who might benefit from enhanced screening. It suggests that it can be strengthened.”
While the results of the study are encouraging, Arras said there is still much to learn, measure and improve.
“Further research is needed to see if the algorithm can be made more accurate,” says Arras. “We also need to identify appropriate ways to use this information in clinical practice.”
A radiologist agreed that the findings were interesting, but said he still had questions about their applicability to the clinic.
“What remains to be proven is whether these AI applications can be fully and effectively integrated into mainstream women’s health care,” said Chest Imaging Medical Director of the Memorial Care Breast Center on the Orange Coast and a member of the Society. Richard Leiterman, M.D., a board-certified radiologist, said. Medical Center, Fountain Valley, Calif. “This publication is based on a so-called retrospective analysis of past cases, but requires validation in an appropriate prospective clinical trial.”
Vinkov doesn’t know exactly if or when patients can expect the tool to be used as part of their mammography. But the fact that the researchers haven’t exactly reinvented the wheel of cancer risk prediction means it could be done more quickly when the time comes, she says.
“No additional testing is needed,” Vinkoff said. “This is a whole new way to use mammography to predict risk. The amazing thing is that we are already doing mammograms. I am doing.”
However, an additional factor is important for predicting rather than detecting cancers that have already developed.
“The interesting message of this article is that AI will not only help radiologists read images, but it is also possible that AI, which is not yet cancerous and therefore undiagnosable today, could develop into cancer within the next five years. It could be used to identify features in mammography,” Leiterman said.
A better understanding of patient risk factors is critical to improving outcomes.
“The earlier breast cancer is detected, the more likely it is to be cured, and the less burdensome and costly treatment is,” Leiterman said.
This appeals to Vinkov, too, who says it could reduce the need for more intensive procedures such as mastectomies in more patients.
But with the current model, patients don’t get much personalized care, Vinkov said.
“We treat everyone as if they were average,” says Vinkov. “This study suggests ways that women’s screening tests can be personalized rather than one-size-fits-all.”
More broadly, while AI may be controversial in other areas such as writing, it could have a life-saving impact on medicine and the future of breast cancer risk assessment, detection and care, Vinkov said. To tell.
“this [study] We treat women as individuals,” says Vinkov. “In healthcare in general, that’s where we want it to be: a place where everyone has access to care and screening tests that are tailored to them and their individual needs.”
