AI is being enlisted to help fight cancer, especially when it comes to early detection. These initiatives, which would automate medical imaging analysis, are developing across several fronts at once as it relates to specific cancers. But AI is also being used to shed some light on cancer globally to try to understand why cancer survival rates differ so dramatically around the world.
Among the newest efforts is a partnership between Bristol Myers Squibb (BMS) and Microsoft to use FDA-approved AI algorithms for early detection of lung cancer. More than 80% of hospitals use Microsoft’s precision imaging network for medical imaging. The AI algorithms can automatically analyze X-ray and CT scans to identify hard-to-detect lung nodules, potentially identifying patients in the early stages of lung cancer, the leading cause of cancer-related deaths in the United States.
“By combining Microsoft’s highly scalable radiology solutions with BMS’ deep expertise in oncology and drug delivery, we’ve envisioned a unique AI-enabled workflow that helps clinicians identify patients with non-small cell lung cancer and guide them to optimal care pathways and precision therapies,” says Dr. Alexandra Goncalves, vice president and head of digital health at BMS.
Similar initiatives that employ AI’s image analysis capabilities include collaboration between 10X Genomics, a leader in single-cell and spatial biology and the Cancer Research Institute to build a 20,000 sample database that may reveal how the immune system recognizes and responds to cancer that may inform future treatment and prevention strategies.
Likewise, AI’s pattern-recognition skills are being used in dermatology to classify skin lesions, including potentially lethal melanoma. AI also is lowering the instances of false positives in mammography. Among the newest areas of AI use is the analysis of liquid biopsies that may prove to be the earliest indicator of cancer. More controversial is a discussion of whether AI agents that can automatically optimize drug design and develop therapeutic strategies have a place in medical treatment, a development that would effectively turn an AI into a doctor.
Another area where AI is poised to help fight cancer is informing patients of clinical trials that might offer life-saving treatment. A new AI platform launched by New York’s Mount Sinai Tisch Cancer Center provides better linkage to clinical trials. Mount Sinai notes that while it may see nearly 10,000 cancer patients per year, less than 10% are enrolled in clinical trials, primarily because many doctors and patients are not aware of them. Although thousands of cancer-related clinical trials that may involve everything from new vaccines and medication to new surgical procedures, past studies indicate that only about 7% of people with cancer participate in them.
Another informational approach to fighting cancer with AI’s machine learning skills is an online tool published in the Annals of Oncology which examines which policy tools or health improvements has the greatest impact on cancer survival.
“We found that access to radiotherapy, universal health coverage and economic strength were often important levers being associated with better national cancer outcomes,” said Dr. Edward Christopher Lee, co-leader of the 185-country study and a resident physician at the Memorial Sloan Kettering Cancer Center in New York City.
While AI is being used to help the fight against cancer, it is a notoriously complicated battle. Cancer evolves through layers of change that may include mutations in DNA, shifts in gene regulation, altered cell behavior and interactions with the immune system. AI tools can narrowly examine parts of cancer’s complexity but what’s really needed is an AI system that can follow a cancerous mutation as it cascades through a cell into a tissue and ultimately shapes an immune response. Two British operations hope to achieve exactly that. Dawn a national research supercomputer, and Isambard-AI are cornerstones of the UK’s Cancer AI & Supercompute Project. The hope is that their combined computational firepower will lead to the development of personalized cancer vaccines to better attack this human scourge. In the meantime, AI is helping to identify the enemy sooner.
