Artificial Intelligence-Driven Nomogram Integrating Hematological Indices and Clinical Features for Predicting Distant Metastasis in Gastric Cancer
1
Oncology, Ho Chi Minh City University of Medicine and Pharmacy, Ho Chi Minh City, Viet Nam
2
Ho Chi Minh City Oncology Hospital, Ho Chi Minh City, Viet Nam
Popul. Med. 2026;8(Supplement Supplement 1):
ABSTRACT
ABSTRACT:
Gastric cancer (GC) remains a major contributor to global cancer mortality, largely due to late-stage diagnosis with distant metastasis (DM). In many low- and middle-income settings, access to advanced imaging for metastatic screening is limited, underscoring the need for low-cost risk stratification tools. Systemic inflammation plays a critical role in cancer progression, and hematological indices such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) may offer practical markers for metastatic risk assessment.[1] This study aims to develop a predictive model for DM using these routinely available biomarkers. We conducted a retrospective study of 92 patients with GC treated at Ho Chi Minh City Oncology Hospital between January and June 2023 (43 with DM, 49 without) (Ethical approval: Ref. No. 292/BVUB-HDDD). Clinicopathological variables and pre-treatment hematological markers were collected. A hybrid AI–clinical approach was applied: Least Absolute Shrinkage and Selection Operator (LASSO) regression was used for key feature selection, followed by multivariable logistic regression to ensure model interpretability. Data were split into training and test sets (70:30). Model performance was evaluated using area under the receiver operating characteristic curve (AUC), calibration analysis, Brier score, mean absolute error (MAE), sensitivity, specificity, and accuracy. A digital nomogram was implemented as an R Shiny–based clinical decision-support tool. LASSO identified NLR, PLR, and primary tumor location as key predictors of DM. The final model achieved an AUC of 0.82 in the training set and 0.80 in the test set, with 75% accuracy (95% CI: 0.59–0.91), sensitivity of 78.9%, and specificity of 66.7%. Calibration metrics indicated good predictive reliability (Brier score ≤0.17; MAE 0.062). This study demonstrates the potential of an interpretable, AI-assisted digital health tool using routine biomarkers for metastatic risk stratification in GC. The proposed nomogram offers a feasible clinical decision-support platform for health systems in resource-constrained settings, contributing to early intervention strategies.
Gastric cancer (GC) remains a major contributor to global cancer mortality, largely due to late-stage diagnosis with distant metastasis (DM). In many low- and middle-income settings, access to advanced imaging for metastatic screening is limited, underscoring the need for low-cost risk stratification tools. Systemic inflammation plays a critical role in cancer progression, and hematological indices such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) may offer practical markers for metastatic risk assessment.[1] This study aims to develop a predictive model for DM using these routinely available biomarkers. We conducted a retrospective study of 92 patients with GC treated at Ho Chi Minh City Oncology Hospital between January and June 2023 (43 with DM, 49 without) (Ethical approval: Ref. No. 292/BVUB-HDDD). Clinicopathological variables and pre-treatment hematological markers were collected. A hybrid AI–clinical approach was applied: Least Absolute Shrinkage and Selection Operator (LASSO) regression was used for key feature selection, followed by multivariable logistic regression to ensure model interpretability. Data were split into training and test sets (70:30). Model performance was evaluated using area under the receiver operating characteristic curve (AUC), calibration analysis, Brier score, mean absolute error (MAE), sensitivity, specificity, and accuracy. A digital nomogram was implemented as an R Shiny–based clinical decision-support tool. LASSO identified NLR, PLR, and primary tumor location as key predictors of DM. The final model achieved an AUC of 0.82 in the training set and 0.80 in the test set, with 75% accuracy (95% CI: 0.59–0.91), sensitivity of 78.9%, and specificity of 66.7%. Calibration metrics indicated good predictive reliability (Brier score ≤0.17; MAE 0.062). This study demonstrates the potential of an interpretable, AI-assisted digital health tool using routine biomarkers for metastatic risk stratification in GC. The proposed nomogram offers a feasible clinical decision-support platform for health systems in resource-constrained settings, contributing to early intervention strategies.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
