According to a study presented at the ongoing 2023 American College of Surgeons Clinical Congress, researchers at Northwestern University have developed an artificial intelligence-based tool to estimate the long-term survival chances of newly diagnosed cancer patients.
Currently, the survival rate of cancer patients is estimated primarily based on their cancer stage. Researchers say that in addition to staging criteria, many other factors can influence patient survival.
The team used machine learning methods to create a prototype tool called the "Cancer Survival Calculator" and tested it on a large dataset. Initial tests estimated 5-year survival rates for patients with breast, thyroid, and pancreatic cancer.
Researchers collected information on patients diagnosed with breast, thyroid, and pancreatic cancer in a number of years. Three-quarters of the collected data was used to train a machine learning algorithm to identify patterns between diagnostic characteristics and patients' 5-year survival rates, then ranked according to the factors with the greatest impact on survival. Using the remaining data, the team employed statistical methods to test the accuracy of the prototype's survival estimates.
The team included data from 259,485 breast cancer patients, 76,624 thyroid cancer patients, and 84,514 pancreatic cancer patients. The four main factors influencing 5-year survival after diagnosis, categorized by cancer location, are as follows: breast cancer; thyroid cancer (age at diagnosis, tumor size, treatment duration, and lymph node involvement); and pancreatic cancer (cancer surgery, histological or microscopic analysis of the cancer, tumor size, and age at diagnosis).
Although some predictive factors, such as tumor size, are part of cancer staging, research results indicate that many more factors influence cancer patients' survival beyond the disease stage.
Researchers report that validation tests show the calculator is "highly accurate" in estimating survival rates for all three cancer sites. Furthermore, this "cancer survival calculator" has the advantage of a more comprehensive dataset and utilizes a new machine learning data model, resulting in faster processing speeds and more accurate risk predictions.
When assessing the prognosis of cancer patients, doctors often use the term "5-year survival rate." This is a statistical indicator, representing the probability that a patient with a certain type of cancer will survive for more than 5 years after comprehensive clinical treatment. Of course, any probability is relative to a group; each individual's situation is different, and miracles may occur, while others may end tragically. This time, researchers have developed an AI tool that adopts more comprehensive indicators to assess patients' 5-year survival rates. Of course, assessing how long a patient will live should not be the ultimate goal; identifying the key factors that influence the survival time of cancer patients may extend the lives of many more people.
