Identification and analysis of vulnerable populations for malaria based on K-prototypes clustering

Abstract

Malaria is a serious public health threat in Yunnan Province of China and has been frequently reported in some endemic regions, such as Tengchong County, with high morbidity. It is essential to analyze the characteristics of malaria cases and identify vulnerable populations. Previous studies about vulnerable populations have mostly used a statistical grouping method to count frequence from a single aspect rather than defined clustered groups. Based on descriptive analysis of the temporal variation and demographic structure of the populations with malaria infection, we used a k-prototypes clustering algorithm to cluster vulnerable populations in Tengchong County in three dimensions, according to sex, age, and occupation. The results indicated that a high incidence of malaria occurred mainly in young male farmers and young or middle-aged male migrant workers. Imported cases, low education level, lack of mosquito bite prevention, and risk behaviors contributed to the high malaria incidence in these groups. Double verification ensured the reliability of this method and reasonability of the results. In addition, we highlighted the importance of targeting prevention and control of malaria for vulnerable groups. We provided suggestions of policies and measures to be implemented by regional governments and at household and individual levels for farmers and migrant workers respectively. Using the k-prototypes clustering algorithm, we efficiently identified those populations at greatest risk of malaria infection. Our results may serve as scientific guidance for targeted malaria prevention and control in Yunnan Province.