The model combines estimates of skeletal muscle with routinely collected clinical data and unstructured notes and predicts the individual likelihood of cachexia.