Back

All text: A Novel Scoring System for Precise Severity Quantification in Severe Fever with Thrombocytopenia Syndrome: Development and Application Based on Dynamic Clinical Data

Sun, Y.; Pan, Z.; Sun, J.; Sun, Y.; Wang, W.; Liang, M.; Zhang, A.; Wu, Q.; Sheng, H.; Yang, J.

2026-02-18 infectious diseases
10.64898/2026.02.17.26346452 medRxiv
Show abstract

BackgroundSevere Fever with Thrombocytopenia Syndrome (SFTS) is an acute infectious disease with high mortality. This study aimed to develop a quantitative scoring system for grading SFTS severity using dynamic clinical data. MethodsA retrospective study included 547 confirmed SFTS patients from two hospitals. Clinical data were collected over a 14-day course (divided into four phases). Patients were grouped into survivors (n=451) and non-survivors (n=96). Statistical analyses, including Kaplan-Meier curves and log-rank tests, were performed. An external validation cohort of 44 new patients was used to validate the scoring system via C-statistic, calibration curves, and decision curve analysis (DCA). ResultsOf 547 patients, 96 (17.55%) were non-survivors. Multivariate logistic regression identified six independent prognostic factors across phases: age, platelet (PLT), aspartate aminotransferase (AST), and creatinine (Cr) (days 5-7); age, red blood cell distribution width (RDW), Cr, and lactate dehydrogenase (LDH) (days 8-10); Cr and LDH (days 11-14). A scoring system (0-11 points) was developed, stratifying patients into low (0-3), intermediate (4-7), and high (8-11) risk groups, with adverse outcome rates of 1.04%, 22.92%, and 76.04%, respectively. Kaplan-Meier curves showed significant prognostic differences (log-rank P<0.001). External validation (44 cases) confirmed excellent performance: AUC 0.810-0.952, good calibration (Hosmer-Lemeshow P>0.05), and net clinical benefit (DCA Eavg 0.068-0.098, Emax 0.422-0.559). ConclusionA dynamic SFTS severity scoring system was developed and validated. Internal and external validation confirmed its reliability and clinical utility, providing a simple, practical tool for timely assessment and early intervention.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.