Researchers at Northwestern University’s McGaw Medical Center have developed an artificial intelligence (AI) model that could revolutionize breast cancer treatment.
With six years of experience in the intersection of AI and pathology, Mohamed Tageldin, a resident physician at Northwestern, is part of the team that has created an AI model aimed at providing more precise long-term outcome predictions for breast cancer patients. This breakthrough comes at a time when the medical field is embracing AI’s potential to support doctors in improving patient care.
Personalized treatment and reduced side effects
The AI model designed specifically for breast cancer aims to provide patients with more personalized treatment recommendations and greater agency in choosing their treatment plans.
One notable advantage of this AI-driven approach is its potential to spare patients from unnecessary and harsh chemotherapy treatments, according to a report published in late November.
Traditional prognosis methods used by pathologists sometimes place patients into higher-risk categories when they could actually benefit from shorter and less intense treatment plans.
By utilizing this AI model, researchers hope to reevaluate and reclassify patients, potentially reducing the duration and intensity of chemotherapy while maintaining clinical efficacy.
A novel approach to pathology
Unlike human pathologists and previous AI models, this algorithm assesses patients by analyzing both cancerous and noncancerous cells, including immune cells, in its prognosis.
Noncancerous cells play a crucial role in inhibiting cancer growth and shaping tumor boundaries, ultimately contributing to improved long-term outcomes for patients. However, these noncancerous cells are often challenging to analyze visually, making it difficult for doctors to determine the appropriate treatment approach.
It’s important to note that the AI model is not intended to replace pathologists but to complement their expertise. Pathologists grade cancer cells’ appearance and predict their growth, and this AI tool aims to enhance their confidence in the grades they assign to oncologists who then collaborate with patients on treatment plans.
The significance of diverse data
The algorithm’s training relied on data from 3,177 breast cancer patients, collected through a partnership with the American Cancer Society’s Cancer Prevention Studies program. This program involves individuals donating their cancer tissue before diagnosis, resulting in high-resolution digital images of the removed tissue that are then incorporated into the dataset.
A team of around 40 doctors, residents, and researchers from around the world analyzed these tissue samples to train the algorithm on cell analysis. This diverse dataset is essential, as it represents a wide spectrum of patient tissues, including those from low-income and rural areas, which can be underrepresented in traditional academic medical institution data.
The AI model’s potential extends beyond well-equipped medical centers. With access to a microscope with a camera and an internet connection, doctors anywhere in the world can use this technology to provide patients with more accurate grades and treatment options. This has the potential to significantly benefit patients in lower-income areas with limited access to specialized pathologists.
The future of AI in medicine
While this AI model represents a significant leap forward in breast cancer treatment, further evaluation is needed, including clinical trials and addressing operational challenges. If approved for clinical use, this model could serve as a template for other cancer types as well.
The American Cancer Society is cautiously optimistic about AI’s role in cancer research, emphasizing that doctors will continue to play a vital role in patient care. AI can serve as a powerful tool to assist doctors in making more informed decisions and ultimately improving patient outcomes.
