| dc.description.abstract | The focus of this project is to apply time series analysis in modelling and forecasting of the malaria
mortality and morbidity based on data from Health Management Information System for Kisii level 5
Hospital between years 2000 to 2009 and relate the intensity of Malaria to seasonality as well as the
climatic variables.
Exploratory Data Analysis was used to uncover the structure of the data and decomposition approach
was applied for analysis. Diagnostic tests were done to the residuals to determine whether they obeyed
the model's assumptions and to test for model adequacy. Data points were taken at quarterly intervals
and analyzed for autocorrelation, trends or seasonal variations. Analysis of the trends per quarter was
done and Holt Winters approach was used to forecast Malaria in the district.
In total 25,890 patients confirmed of malaria were seen from the year 2000 to 2009 of which 644
succumbed to the disease. From the analysis, the odds ratio of mortality of children decreased per year by 39.95%.the results suggest that, there is a quadratic trend in the proportion of mortality per year by 5.13%. The odds ratio of children mortality in quarter 2,3, and 4 were 39.1%, 37.71% and 53.73% higher "than those of quarter 1 respectively. The odds'ratio of mothers' mortality increases with each year by
13%. For every unit increase of humidity, mortality of children increases by 1.6% while for every unit
increase of temperature decreases mor-tality by 14.6%.
Time series is a powerful tool for analyzing, evaluating the current achievements, comparing, and
forecasting the future based on data that is arranged chronologically. Relative humidity and
temperature should be included in early warning system of rnalaria'in Kenya. | en |
