Develop a data-driven model to predict postoperative malnutrition in patients with gynecologic cancer

Objective Craft a high-fidelity, clinically calibrated prediction model to foresee postoperative malnutrition in patients with gynecologic malignancies—empowering clinicians to pinpoint vulnerability at its earliest whisper and initiate timely, life-sustaining nutritional intervention. Methods The clinical data of 755 patients with gynecological malignant tumors who received surgical treatment in 4 hospitals in China from January 2016 to January 2021 were retrospectively analyzed. The outcomes of the diseases after surgical treatment were observed. According to whether malnutrition occurred, the patients were divided into the occurrence group (228 cases) and the non-occurrence group (527 cases). Logistic regression analysis was conducted to screen out the appropriate predictive factors. A risk prediction model for postoperative malnutrition in gynecological malignant tumors was established. Results Univariate analysis identified postoperative hypoproteinemia, FIGO stage, PSQI score, postoperative chemotherapy, preoperative PG-SGA score, and grip strength as associated with postoperative malnutrition. Multivariable logistic regression confirmed five independent predictors: postoperative hypoproteinemia, FIGO stage, PSQI score, postoperative chemotherapy, and preoperative PG-SGA score. The model achieved an AUC of 0.937 (95% CI: 0.872–1.000), with sensitivity 82.6% (67.1–98.1%) and specificity 95.1% (91.6–98.7%). Ten-fold cross-validation yielded AUC 0.921 (0.845–0.996), sensitivity 87.0% (73.2–100.0%), and specificity 88.2% (82.9–93.5%). Bootstrap validation showed AUC 0.899 (0.895–0.904), sensitivity 76.8% (75.8–77.7%), and specificity 93.7% (93.5–93.9%). Calibration was excellent (mean absolute error = 0.016); decision curve analysis demonstrated clinical net benefit. Conclusion This elegantly calibrated predictive model—anchored in five clinically resonant pillars: postoperative hypoproteinemia, FIGO stage, PSQI score, postoperative chemotherapy, and preoperative PG-SGA score—delivers exceptional discriminative power for identifying patients at heightened risk of postoperative malnutrition in gynecologic malignancy. More than a statistical tool, it serves as a clinical compass: empowering oncologists and nutrition specialists to stratify risk with precision, guide empathetic, evidence-informed consultations, prioritize high-risk individuals for timely nutritional intervention, and tailor longitudinal follow-up with foresight and fidelity.