Development of an Artificial Intelligence Model to Predict 5-Year Survival Rate After Gastric Cancer Surgery

Seoul Asan Hospital Professors Lee Inseop and Kim Kyungwon Team
Accuracy Around 80%... Also Discovered Important Predictive Factors

Research results on an artificial intelligence (AI) model that can predict the 5-year survival rate of gastric cancer patients have been announced, drawing attention. Data analysis showed that exercise and diet efforts that patients can undertake themselves after surgery had a significant impact on long-term survival.

Professor Inseop Lee of the Department of Gastrointestinal Surgery and Professor Kyungwon Kim of the Department of Radiology at Seoul Asan Medical Center announced on the 16th that their research team developed an AI algorithm that predicts the 5-year survival rate at about 80% based on treatment outcomes and health status one year after surgery.

Professor Lee In-seop of the Department of Gastrointestinal Surgery at Asan Medical Center, Seoul (left), and Professor Kim Kyung-won of the Department of Radiology.
[Photo by Asan Medical Center, Seoul]

The research team trained the AI with 65 types of data, including preoperative health information, surgical, chemotherapy, and pathological information of 3,220 patients who underwent gastric cancer surgery at Seoul Asan Medical Center from 2003 to 2012, as well as commonly performed blood tests and computed tomography (CT) scans for recurrence follow-up. The algorithm development utilized data from gastric cancer patients one year after surgery. The team judged that death within one year after surgery is often due to the aggressiveness of the cancer, and since stage 2 and 3 gastric cancer patients undergo adjuvant chemotherapy for 6 months to 1 year after surgery, the patient’s condition one year post-surgery is crucial for determining long-term survival.

After creating the algorithm, internal validity was evaluated using data from 805 patients, resulting in a 76% accuracy in predicting the 5-year survival rate after gastric cancer surgery. External validity was verified with data from 590 patients who underwent gastric cancer surgery at Ajou University Hospital from 2010 to 2012, achieving a prediction accuracy of 81%.

The research team further analyzed the 65 types of patient data and found that changes in body weight, muscle mass, fat mass, and nutritional status in patients who underwent gastric cancer surgery were important factors in predicting the 5-year survival rate. They confirmed that worsening related indicators such as decreases in body weight and muscle mass, increases in fat mass, and Nutritional Risk Index (NRI) corresponded with a decline in the 5-year survival rate.

Professor Inseop Lee emphasized, "This study is significant not only because it enables the prediction of long-term outcomes after gastric cancer surgery through large-scale data analysis but also because it revealed that factors patients can self-correct, such as consistent strength training and a high-protein diet, influence long-term survival." Professor Kyungwon Kim explained, "Most treatment outcome prediction models have the limitation of not being validated with external patient groups, but this study was conducted based on both internal and external data, enhancing the reliability of the research."

This study was recently published in the international journal in the field of geriatrics, Journal of Cachexia Sarcopenia and Muscle (IF=12.063).

This content was produced with the assistance of AI translation services.