Just and their colleagues examined whether an artificial intelligence (AI) tool could automatically measure body composition from CT scans and help predict outcomes in patients receiving left ventricular assist devices. Preoperative CT scans from 137 patients were analyzed using a U-net convolutional neural network that segmented visceral fat, subcutaneous fat, and muscle at the L3 spinal level. The AI model produced automated tissue measurements with only occasional manual correction, highlighting consistent performance. Obesity was present in 32.1 percent of patients and sarcopenia in 70.1 percent. Higher visceral adipose tissue was linked to a greater risk of postoperative infection. Sarcopenic obese patients had a 4.2-fold increased infection risk with a 95% CI. Higher visceral and subcutaneous fat were also found in patients who died during the hospital stay using the model. The AI model in this work was not used to classify or predict outcomes directly. Instead, the AI model was used only for image segmentation.