New AI Tool Spots Silent Killer In Cancer Patients
A newly created artificial intelligence tool can predict which cancer patients are at risk of developing a life-threatening condition known as cachexia, according to recent research.
Cachexia is responsible for roughly 20% of cancer-related deaths and is marked by severe muscle wasting, systemic inflammation, and significant weight loss, explained Sabeen Ahmed, a graduate researcher at the University of South Florida, in a press release. Though the exact cause of cachexia remains unclear, experts suspect factors like inflammation, metabolic changes from cancer, insulin resistance, and hormonal shifts contribute to its onset, according to the National Cancer Institute (NCI).
Because cachexia cannot be reversed simply through nutrition and becomes difficult to treat once it progresses, early detection is crucial. Interventions — both lifestyle adjustments and medication — can help slow muscle loss, improve metabolism, and boost patients’ quality of life if initiated early enough, Ahmed noted. However, existing detection methods based on weight loss, clinical observations, and indirect biomarkers often prove inconsistent and are usually too late to be effective.
In this study, researchers trained an AI system to predict cachexia risk by analyzing CT scans along with other clinical information. The AI first assesses skeletal muscle volume from imaging scans and then combines that with data such as a patient’s demographics, weight, height, cancer stage, lab results, and clinical notes. When only imaging and basic patient information were used, the AI correctly identified cachexia in 77% of cases. Its accuracy rose to 81% with lab data and reached 85% when clinical notes were added. The AI also proved highly reliable, with muscle mass measurements differing from expert radiologists’ assessments by an average of just 2.48%.
These promising results, presented by Ahmed at the American Association for Cancer Research’s annual meeting in Chicago, suggest that AI could significantly improve early detection and treatment strategies for cancer-related cachexia.
Discussion about this post