AIC-Driven Spatial Hierarchical Clustering: Case Study for Malaria Prediction in Northern Thailand

Targeted intervention and resource allocation are essential in effective control of infectious diseases, particularly those like malaria that tend to occur in remote areas. Disease prediction models can help support targeted intervention, particularly if they have fine spatial resolution. But there is typically a tradeoff between spatial resolution and predictability of the time series of infection. In this paper we present a systematic method to quantify the relationship between spatial resolution and predictability of disease and to help provide guidance in selection of appropriate spatial resolution. We introduce a complexity-based approach to spatial hierarchical clustering. We show that use of reduction in Akaike Information Criterion (AIC) as a clustering criterion leads to significantly more rapid improvement in predictability than spatial clustering alone. We evaluate our approach with two years of malaria case data from northern Thailand.