INTRODUCTIONSarcopenia negatively impacts surgical outcomes in gastrointestinal cancer patients, yet practical preoperative screening tools are lacking. The CRP/ALB ratio, a novel biomarker of systemic inflammation and nutritional status, may enhance sarcopenia prediction but remains underexplored in surgical oncology. This study aims to identify the predictors for preoperative sarcopenia prediction in gastric and colorectal cancer patients.METHODSThis retrospective study analyzed 145 patients undergoing curative surgery (2019-2021). Sarcopenia was defined by sex-specific CT-measured L3 skeletal muscle index (cutoffs, male ≤ 40.8 cm2/m2; female ≤ 34.9 cm2/m2). Multivariable logistic regression identified predictors, with model performance assessed via ROC analysis and Cohen's Kappa.RESULTSThe cohort (median age 64 years; 73.8% male) comprised 66 gastric (45.5%) and 79 colorectal (54.5%) cancer patients, with 29 (20%) diagnosed with sarcopenia. Sarcopenic patients exhibited a higher NRS 2002 score (P < 0.001), lower PNI score (P < 0.05), and higher CRP/ALB ratio (P < 0.05). Multivariate logistic regression analysis results showed that CRP/ALB ratio (OR = 3.084, 95% CI 1.071-8.882, P = 0.037), age (OR = 1.074, 95% CI 1.021-1.130, P = 0.006), and BMI (OR = 0.667, 95% CI 0.542-0.820, P = 0.000) were associated with the increased risk of sarcopenia. The combined model achieved superior discrimination (AUC = 0.854, 95% CI 0.770-0.937), yielding 75.86% sensitivity and 84.82% specificity at optimal cutoff value - 1.0340, and a Cohen's Kappa coefficient of 0.542 when compared to CT results.CONCLUSIONThe CRP/ALB ratio combined with BMI and age is utilized as a convenient and effective tool for preoperative sarcopenia screening. This model-driven approach provides robust strategies to facilitate preoperative interventions, optimize perioperative care, and enhance long-term oncological outcomes for patients undergoing gastric and colorectal cancer surgery.