Abstract
The Qinghai-Tibet Plateau is one of the regions with the highest prevalence of cerebral echinococcosis worldwide. Investigating the risk factors for cerebral echinococcosis in Ganzi and developing an effective clinical risk prediction model is crucial for enhancing disease prevention and control in the region. Study participants were selected from the People's Hospital of Ganzi Tibetan Autonomous Prefecture between January 2016 and December 2023. In this subject, 152 cases were diagnosed with cerebral echinococcosis (case group), while 580 cases were diagnosed with non-Echinococcus infection (control group). The chi-square test revealed significant differences in the component ratios for occupation, residential altitude, presence or absence of comorbidities with other sites of echinococcal infections, hypoproteinaemia, and tuberculosis (all P < 0.001). Logistic regression identified five variables-occupation, residential altitude, tuberculosis, hypovitaminosis, and combined infection in other sites-as risk factors for cerebral echinococcosis (P < 0.005). The predictive nomogram assigned the following scores: farmers (0), public officials or students (5.81), and herdsmen (60.624). The scores for the presence or absence of infections in other sites were 100 and 0, respectively. The scores for the presence or absence of hypoproteinaemia were 58.99 and 0, respectively. The scores for the presence or absence of tuberculosis were 54.65 and 0, respectively. The scores for altitudes ≤3000 m and >3000 m were 53.32 and 0, respectively. Internal validation of the nomogram's receiver operating characteristic curve using the Bootstrap method with 500 repeated samples showed the area under the curve was 0.920 (95% confidence intervals: 0.887-0.952). A confusion matrix was constructed using the true infection values, revealing a maximum Youden index of 0.76, sensitivity of 0.763, and specificity of 0.887. Internal validation using the Bootstrap method with 500 repeated samples showed that the calibration curve closely approximated the ideal curve, indicating that the model was well-calibrated. The Hosmer-Lemeshow goodness-of-fit test showed that χ² = 10.234, P > 0.05, further confirming that the model was well-calibrated. The decision curve analysis indicated that the model's best applicability for cerebral echinococcosis infection thresholds lies between 0.02 and 0.99. The nomogram model developed in this study for human brain echinococcosis infection demonstrated strong identification and predictive capabilities. The receiver operating characteristic curves and calibration plots confirmed the model's high accuracy and consistency, further supporting its effectiveness. By identifying high-risk groups and protective factors for cerebral echinococcosis, the model offers a solid scientific foundation for the development of targeted prevention and control strategies.