| ObjectiveIn order to provide the scientific data for preventing strategies, the epidemic trend of bacteroidal dysentery in Fuzhou, the relation of bacteroidal dysentery and meteorological factors and a predicating model for forecasting of bacteroidal dysentery have been studied.MethodsThe data about demography documents, bacteroidal dysentery occurrence, social and economic situation and climate changes during the period between 1987 and 2006 in Fuzhou have been collected. Then the area was divided into four areas, city zones, plain areas, mountainous zones and costal areas in terms of geographical position and social economy situation.The relationships of meteorological factors and bacteroidal dysentery have been analyzed with correlations analysis of partial in Fuzhou 5 areas. The forecasting model for bacteroidal dysentery occurrence has been established by using the Neural Network Toolbox of Matlab7.2 software package. In the studies of forecasting model, the data of bacteroidal dysentery and meteorological factors from 1987 to 2006 in 5 Fuzhou areas have been chosen to analysis. The established forecasting model also has been tested. Results1. Epidemic Research(1) In the past 20 years, there have been 15,378 reported bacteroidal dysentery cases out of a population of 113,725,105, with an average incidence of 13.52/100,000 yearly. The incidence rate of Fuzhou was low level in China during the period, the highest was 28.86/100,000 in 1987, while the lowest was 6.31/100,000 in 2001. If the districts were listed in the order of incidence of disease, from highest to lowest, it comeed out as follow: city zones, plain areas, mountainous areas and coastal lines. (2) The epidemic trend decresed in the 20 years and could be compartmentalized into four stages in periodic fluctuations. It kept decreased from late 1980s to early 1990s and gradually went up until mid 1990s.Then it went down after again and stayed stable since 2000. The epidemic trends of the four districts had charactic and slightly different from that of the whole area. (3) Bacillary dysentery occurred in every age group, and the highest incidence rate was among children under 5 years. The percentage of male was higher than that of female. (4) Local epidemic trend of bacteroidal dysentery appears in seasonal patterns: the peak was usually during August and September, while the valley was February.2. Climate AnalyzedThe temperature in Fuzhou had been getting higher year by year during the period, with the average temperature was 20.01±0.50℃. The average annual sunshine length and humidity was amount to 1618.38±132.11 hours and 77.70±2.08%, respectively. In addition, the average rainfall per year was 1479.70±300.34 mm. All the meteorological factors apparently followed seasonal patterns.Comparing meteorological factors of distinct zones in same period: 1) Temperature: it was hottest in the city zone and coldest in coastal areas, while those of mountainous zones and plain were similar with the average level of the whole area. 2) Precipitation: the amount of rainfall in city zone was lowest compared to others. Costal areas and mountainous zones had more precipitation than average.Besides plain had as much rain as the average of whole area. 3) Relative Humidity: the lowest humidity was in city zone and the highest in coastal zones. Mountainous areas and plains just had similar humidity with the average in the whole area. 4) Sunshine Length: the four areas enjoyed the same sunshine-length in general. 5) Air Pressure: the air pressure in city zone was lowest one and that of mountainous areas was near the average. However, the air pressure in costal areas and plain were higher than average.3. Partial Correlation AnalysisIn the certain economic preconditions, there was a positive correlation between monthly the average temperature,monthly average sunshine-length and the incidence rate of bacteroidal dysentery per month. The relative coefficients were as follow: temperature was 0.442～0.668(P <0.01), sunshine-length was 0.360～0.510(P <0.01). Besides, there was a negative correlation between air pressure and the incidence rate of bacteroidal dysentery, whose relative coefficient was -0.369～-0.606(P <0.01). As for the amount of rainfall, there was no significant difference (P >0.05)between that in costal areas and the incidence rate per month, while positive correlation in other areas, with the relative coefficient was 0.170～0.267(P <0.01).In addition, there was positive correlation only between relative humidity in plain and the monthly incidence rate of bacteroidal dysentery (r=0.228,P <0.01), for others had no significant difference (P >0.05).4. BP Neural Network ModelThe BP network model was established by using the data of diseases and meteorological factors. After training the neural network, the error of performance decreased and the coefficient of regression all biger than 0.8. Verified by fact data of bacteroidal dysentery in 2007, the average absolute error rate of 5 model was 17.30%～18.19% respectively, the efficiency of forecasting for bacteroidal dysentery was 83.33%～91.67% respectively.ConclusionDuring 1987 and 2006, the epidemic trend of bacteroidal dysentery in Fuzhou was obviously seasonal pattern. It decreased gradually, but still remained at a certain rate and in the shape of periodic fluctuations. In Fuzhou, which was still a place with low incidence rate, the age group under 10 years old has the highest incidence. Meanwhile, the incidence rate of city zone exceeded those in other areas.There was a certain relationship between local meteorological factors and bacteroidal dysentery. The correlation coefficient of temperature was the biggest. It was concluded that the relation of meteorological factors and bacteroidal dysentery incidence rate could be combined to establish a BP neural network model.BP neural network model was feasible to analysis the relation of meteorological factors and bacteroidal dysentery, the percentage of correct prediction was above 80%. BP neural network model could be used as a new effective method for forecasting of bacteroidal dysentery by the meteorological factors.
|
PDF Full Text Request