In the U.S., 1 in 8 women will be diagnosed with breast cancer and it is the second leading cause of cancer death in women. But traditional breast cancer risk assessment tools demonstrate poor or biased performance that is unequal across different populations.
MIRAI is an open source, state-of-the-art deep learning model that can provide a personalized risk score up to 5 years in advance just by analyzing a patient’s mammogram, maintaining high accuracy across diverse populations.
MIRAI has been validated on over 1.9 million mammograms from patients all over the world. This validation process is key to ensuring that MIRAI maintains high performance across an array of local populations and that it can be safely used on real patients.
MIRAI is a deep learning model that can analyze a patient’s mammogram to accurately predict the patient’s risk of developing breast cancer in the next 5 years.
MIRAI assigns a personalized risk score to the mammogram, helping clinicians determine when a patient should return for their next screening.
Who is MIRAI for?
Breast cancer is the #1 most common cancer in women worldwide, causing 670,000 deaths in 2022. Half of breast cancers develop in women who have no identifiable breast cancer risk factors other than gender and age.
Moreover, survival is widely inequitable — nearly 80% of deaths from breast cancer occur in low- and middle-income countries.
The team behind MIRAI
Jameel Clinic AI faculty lead Regina Barzilay was inspired to build MIRAI after her own breast cancer diagnosis when she learned there were no clinically-available AI models that assess breast cancer risk. Joined by then-student Adam Yala and Mass General Brigham radiologists, Barzilay led the team to develop the pioneering technology behind MIRAI, which is widely considered state-of-the-art.
Democratizing world-class technology
47 hospitals around the world have installed MIRAI for the purposes of screening and risk assessment.
Thanks to the support of Wellcome Trust, we have been able to expand the deployment of MIRAI to include hospitals serving under-resourced regions.
The scale and computational power needed for bold ideas to work is not possible without the support of grants, foundations, and donors. Your support would enable MIT Jameel Clinic to make a better, healthier future for all.
Artificial intelligence–based risk-stratified programs were estimated to be cost-saving and increase QALYs compared with the current screening program. A screening schedule…was estimated to give yearly net monetary benefits within the NHS of approximately £60.4 (US $77.3) million and £85.3 (US $109.2) million, with QALY values set at £20 000 (US $25 600) and £30 000 (US $38 400), respectively.JAMA NETWORK OPEN VOLUME 7 NO. 9 / SEPTEMBER 2024
[MIRAI] showed good performance in a high-risk external dataset enriched for African American patients, benign breast disease, and BRCA mutation carriers, and study findings suggest that the model performance is likely driven by the detection of precancerous changes.RADIOLOGY ARTIFICIAL INTELLIGENCE VOLUME 20 5(6) / NOVEMBER 2023
Of the risk-stratified regimens considered, we found using three screening intervals (1, 3 or 4 years) using Mirai 3-year risk thresholds (<1.56% for low risk, and ≥5.06% for high risk) was promising, and expected to reduce the number of advanced cancers diagnosed by approximately 18 advanced cancers per 1000 cancers diagnosed with triennial screening.npj Digital Medicine VOLUME 6, NUMBER 223 / NOVEMBER 2023
AI provides a powerful way to stratify women for clinical considerations that necessitate shorter time horizons such as risk-based screening and supplemental imaging.RADIOLOGY VOLUME 307, ISSUE 5 / JUNE 2023
Overall, 3-year risk was strongly associated with development of both future interval cancer and screen-detected cancer. There was no evidence of a significant difference in model performance between interval and screen-detected cancers (AUC, 0.69 vs 0.67; P = .085) or invasive cancer versus DCIS.RADIOLOGY VOLUME 307, ISSUE 5 / JUNE 2023
This study concluded that Mirai has the potential to replace current breast cancer risk assessment models recommended by clinical guidelines for magnetic resonance imaging screening.Journal of Clinical Oncology VOLUME 40, NUMBER 20 / APRIL 2022
For patients with breast cancer, learning about the pathology can help them regain control of their lives, especially when they are shown the hopeful data of healing and recovery.
Dr. Luz Fernanda Sua Villegas, MD, PhD Pathologist Fundación Valle del Lili
At IRST, we firmly believe that integrating deep learning technologies like SYBIL and MIRAI in clinical practice, will radically improve patient care. The use of AI algorithms by physicians and medical physicists can capture this opportunity and lead to the future of precision medicine because the utilization of imaging data is now only partially exploiting its full potential.
Dr. Giacomo Feliciani, PhD Medical Physics Unit IRCCS Istituto Romagnolo per lo Studio dei Tumori “Dino Amadori” IRST