Research and Products:
Predicting Valley Fever Incidence

Research Methods

An exploratory data analysis was conducted during the early part of the project to determine the basic relationship between climate and incidence. This portion of the study involved a bivariate analysis of individual climate variables, including temperature (minimum, maximum, dew point, and average), precipitation, wind speed, and the Palmer Drought Severity Index (Figure 3). A composite analysis was also conducted, which examined the climate conditions leading up to months with especially high and especially low incidence (Figure 4).

The results of the exploratory analysis informed the development of monthly multivariate predictive models. For each month, temperature and precipitation variables at varying time scales up to four years prior to the month being predicted were identified from the exploratory analysis. These variables were entered into a backward stepwise regression. The resulting models were tested on independent data to evaluate their ability to predict incidence (Figure 5).

Figure 3. October temperature and incidence the following August, Kern County 1989-1998. [more details]

 

Figure 4. January % of total annual incidence, Pima County 1948-1998. [more details]

 

Figure 5. Mean monthly percentage of total annual cocci. [more details]

 

< Project Overview

Status of Project >