Researchers at Northwestern University in Chicago have created an artificial intelligence tool designed to forecast cancer patients’ survival chances. The study shows that outcome depends not only on how far the disease has progressed but also on a range of other influencing factors. The research is presented at the American College of Surgeons Clinical Congress and is accessible through their proceedings.
The investigation analyzed large patient cohorts: 259,485 individuals with breast cancer, 76,624 with thyroid cancer, and 84,514 with pancreatic cancer. The results indicate that survival is shaped by multiple variables, including age, tumor size, and the specific treatments administered. Each cancer type shows a unique set of impactful factors; for instance, thyroid cancer outcomes hinge on whether lymph nodes are involved, while breast cancer prognosis can be affected by hormone receptor status and responsiveness to certain therapies.
About three quarters of the dataset were used to train machine learning models. The remainder of the data served to validate the calculator’s predictive performance. The tool demonstrated reliable ability to estimate whether a patient would survive the upcoming nine to ten months when tested on unseen cases.
Traditionally, survival estimates for cancer patients have relied heavily on the stage of disease. The Northwestern tool adds layers of information by incorporating biomarkers and treatment features that were not part of earlier survival calculators, offering a more nuanced prognosis for individual patients in North America and beyond.
In related findings, studies have linked social and environmental factors to cancer outcomes. For example, data indicate that living alone can be associated with higher mortality risk in cancer patients, emphasizing that real-world context matters alongside biological indicators.