Spatio-temporal modeling of sparse geostatistical malaria sporozoite rate data using a zero inflated binomial model

Area of Science:

• Malariology
• Biostatistics
• Geostatistics

Background:

• Sporozoite rates in malaria vectors are key to transmission dynamics.
• Spatio-temporal variations in sporozoite rates significantly impact malaria transmission patterns.
• Sporozoite rate data is often sparse with numerous zeros, posing analytical challenges.

Purpose of the Study:

• To develop and compare Bayesian zero inflated binomial (ZIB) geostatistical models with standard binomial models.
• To analyze sparse sporozoite rate data from rural Western Kenya (2002-2004).
• To evaluate the predictive ability and parameter estimation of ZIB models.

Main Methods:

• Development of Bayesian zero inflated binomial (ZIB) geostatistical models.
• Application of models to sporozoite rate data from the KEMRI/CDC Health and Demographic Surveillance System (HDSS).
• Comparison of ZIB models against standard geostatistical binomial models.

Main Results:

• ZIB models demonstrated superior predictive ability compared to standard binomial models.
• ZIB models identified more significant covariates influencing sporozoite rates.
• Narrower credible intervals for parameters were achieved using ZIB models.

Conclusions:

• Bayesian zero inflated binomial geostatistical models are more effective for analyzing sparse sporozoite rate data.
• These advanced models enhance understanding of malaria transmission by improving covariate identification and prediction.
• The findings support the use of ZIB models for more accurate malaria surveillance and control strategies.