Abstract:

Objective: To optimize a predictive model of leprosy by including epidemiological risk factors and genetic variants in the Chinese population. Methods: 816 patients with leprosy and their 3847 contacts were recruited from 10 cities in Shandong Province, China. Epidemiological information was collected from the Leprosy Patients Registration Form and medical records. The Logistic regression technique was used to select the optimal risk features and develop predictive models based on different combinations of genetic and epidemiological risk factors. The discriminatory capability of each model was evaluated using the area under the curve (AUC). Results: By including 3 epidemiological factors and 25 variants, the discriminatory capability of the optimal predictive model of leprosy was 0.821 (95% CI: 0.801-0.842) in the discovery stage and 0.812 (95% CI:0.789-0.835) in the validation stage. In addition, the discriminatory capacity for leprosy associated with self-related factors (AUC=0.750, 95% CI:0.726-0.773) was similar to that for factors associated with index cases (AUC=0.745, 95% CI:0.718-0.772). The cut-off value was 0.202 based on the optimal sensitivity and specificity, and individuals in the high-risk group had a 8.5 times greater likelihood of developing leprosy than those in the low-risk group. Conclusion: This optimizing model based on genetic and epidemiological risk factors shows perfect discriminatory capability for leprosy, which is of great significance for the identification of high-risk populations and the implementation of precise chemical prophylaxis.

Key words: leprosy, predictive model, epidemiological factors, genetic variant