Chaochao Chen,* Zhengxian Su,* Yuwei Zheng, Minya Jin, Xiaojie Bi

Department of Laboratory Medicine, Taizhou Hospital of Zhejiang Province Affiliated to Wenzhou Medical University, Linhai, 317000, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Xiaojie Bi, Department of Laboratory Medicine, Taizhou Hospital of Zhejiang Province Affiliated to Wenzhou Medical University, No. 150, Ximen Street, Linhai, 317000, People’s Republic of China, Tel +86 13757693182, Email bixj@enzemed.com Minya Jin, Department of Laboratory Medicine, Taizhou Hospital of Zhejiang Province Affiliated to Wenzhou Medical University, No. 150, Ximen Street, Linhai, 317000, People’s Republic of China, Tel +86 13757693182, Email jinmy@enzemed.com

Purpose: Sepsis has high mortality and progresses rapidly, requiring early diagnosis; traditional scoring and lab parameters are limited in non-ICU settings, highlighting the need for biomarker integration and continuous monitoring to enhance diagnostic accuracy.
Patients and Methods: A retrospective analysis of 1,098 patients at Taizhou Hospital of Zhejiang Province identified sepsis and non-sepsis groups per Sepsis 3.0 criteria, Logistic regression analyses were used to identify the risk factors. A dynamic nomogram was built, and predictive accuracy was evaluated using calibration and decision curves. External validation for 94 patients occurred from January to March 2024, using Receiver operating characteristic (ROC) curve analysis for diagnostic evaluation.
Results: Multivariate logistic regression analysis revealed eight independent risk factors significantly associated with sepsis development: hypertension (odds ratio [OR] = 1.6278, 95% confidence interval [CI], 1.2079– 2.1937), renal insufficiency (OR=1.7002, 95% CI, 1.2840– 2.2513), cardiac insufficiency (OR=1.8927, 95% CI, 1.2979– 2.7599), interleukin-6 levels (OR=1.0003 95% CI, 1.0002– 1.0005), basophil percentage (OR=0.4319, 95% CI, 0.2353– 0.7926), platelet-to-lymphocyte ratio (PLR) (OR=1.0025, 95% CI, 1.0011– 1.0040), platelet count (PLT) (OR=0.9939, 95% CI, 0.9912– 0.9959) and D-dimer levels (OR=1.0796, 95% CI, 1.0273– 1.1347). The prognostic nomogram showed significant discriminative power, with a concordance index of 0.746 (95% CI 0.709– 0.772). ROC analysis further revealed a negative predictive value (NPV) of 0.832 and a positive predictive value (PPV) of 0.511. Decision curve analysis validated the clinical utility of the model, demonstrating a substantial net benefit for predicting disease progression within a clinically relevant probability threshold range of 30% - 70%. The model maintained satisfactory discriminative performance in external validation, demonstrating an area under the curve (AUC) of 0.663 (95% CI, 0.549– 0.776). The interactive web-based nomogram is available at https://bixiaojie-1987.shinyapps.io/DynNomapp/.
Conclusion: This web-based dynamic nomogram incorporating eight clinically readily available predictors demonstrates robust diagnostic performance for sepsis, which helps doctors make quicker decisions by providing real-time risk assessments for each patient in non-ICU departments.

Keywords: web-based dynamic nomogram, sepsis diagnosis, non-ICU settings, clinical biomarkers, model validation