Cargando…
Malaria temporal dynamic clustering for surveillance and intervention planning
BACKGROUND: Targeting interventions where most needed and effective is crucial for public health. Malaria control and elimination strategies increasingly rely on stratification to guide surveillance, to allocate vector control campaigns, and to prioritize access to community-based early diagnosis an...
Autores principales: | , , , , , , , , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Elsevier
2023
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10280692/ https://www.ncbi.nlm.nih.gov/pubmed/37004429 http://dx.doi.org/10.1016/j.epidem.2023.100682 |
Sumario: | BACKGROUND: Targeting interventions where most needed and effective is crucial for public health. Malaria control and elimination strategies increasingly rely on stratification to guide surveillance, to allocate vector control campaigns, and to prioritize access to community-based early diagnosis and treatment (EDT). We developed an original approach of dynamic clustering to improve local discrimination between heterogeneous malaria transmission settings. METHODS: We analysed weekly malaria incidence records obtained from community-based EDT (malaria posts) in Karen/Kayin state, Myanmar. We smoothed longitudinal incidence series over multiple seasons using functional transformation. We regrouped village incidence series into clusters using a dynamic time warping clustering and compared them to the standard, 5-category annual incidence standard stratification. RESULTS: We included 1115 villages from 2016 to 2020. We identified eleven P. falciparum and P. vivax incidence clusters which differed by amplitude, trends and seasonality. Specifically the 124 villages classified as “high transmission area” in the standard P. falciparum stratification belonged to the 11 distinct groups when accounting to inter-annual trends and intra-annual variations. Likewise for P. vivax, 399 “high transmission” villages actually corresponded to the 11 distinct dynamics. CONCLUSION: Our temporal dynamic clustering methodology is easy to implement and extracts more information than standard malaria stratification. Our method exploits longitudinal surveillance data to distinguish local dynamics, such as increasing inter-annual trends or seasonal differences, providing key information for decision-making. It is relevant to malaria strategies in other settings and to other diseases, especially when many countries deploy health information systems and collect increasing amounts of health outcome data. FUNDING: The Bill & Melinda Gates Foundation, The Global Fund against AIDS, Tuberculosis and Malaria (the Regional Artemisinin Initiative) and the Wellcome Trust funded the METF program. |
---|