Risk stratification for leprosy in a hyperendemic area combining data mining and Anti-PGL-I

Abstract

Objective: To investigate the combined application of anti-PGL-1 serology and data mining in leprosy surveillance in a hyperendemic area. Methods: Epidemiological study with 280 contacts in Santarém-PA, using clustering (K-means) as 11 variables: age group, gender, degree of kinship with the index case and period of coexistence with the index case, clinical form of the index case, operational classification of the index case, number of rooms, total number of residents and existence of a dormitory with more than 2 people, family income, time of residence in the family home and serological analysis (ELISA anti-PGL-I). Results: Three risk clusters were identified (C1-C3), with C3 (high risk) associated with a higher bacillary load (median OD=0.63), poor socioeconomic conditions and 54.3% of new cases. Seropositivity was 39.6% among contacts. Conclusion: This study demonstrates that the integration of anti-PGL-I serology with data mining techniques represents an innovative and effective approach for leprosy surveillance in hyperendemic areas, allowing the identification of three distinct epidemiological clusters (C1-C3) with an increasing gradient of risk associated with bacillary load, socioeconomic conditions and characteristics of index cases. The integrated approach proved to be effective for risk stratification and targeted surveillance.