AI Predicts Wasting Syndrome in Cancer Patients
Researchers have developed an artificial intelligence (AI) system capable of predicting which cancer patients are at risk for cachexia, a life-threatening wasting syndrome that accounts for about 20% of all cancer-related deaths. The condition is characterized by systemic inflammation, severe muscle wasting, and significant weight loss, particularly in patients with advanced cancers.
Cachexia is a serious complication affecting many cancer patients, but its exact cause remains unknown. Suspected factors include inflammation, increased cancer metabolism, insulin resistance, and hormonal changes. Unlike simple weight loss, cachexia cannot be reversed through nutrition alone and requires medical treatment. Early detection is crucial as the condition becomes difficult to reverse once it progresses.
The newly developed AI system assesses the risk of cachexia by analyzing CT scans and clinical data. It first evaluates the amount of muscle in a patient’s body through imaging scans and then uses additional data to determine their risk. In tests, the AI accurately identified cachexia in 77% of cases when provided with imaging scans, demographic information, weight, height, and cancer stage. The accuracy increased to 81% when lab results were included and reached 85% when doctors’ clinical notes were added to the analysis.
This assessment enabled the AI to better predict survival odds for patients with pancreatic, colon, and ovarian cancers. The AI’s muscle assessment differed from expert radiologists’ calculations by only about 2.5% on average, demonstrating its reliability. Lead researcher Sabeen Ahmed presented these findings at the American Association for Cancer Research’s annual meeting in Chicago, highlighting the potential of AI to improve detection and treatment of cachexia.
The development of this AI technology represents a significant advancement in cancer care, offering the potential to identify patients at risk earlier and improve their quality of life through targeted interventions.
