Application Of Mathematical Models To Estimate The Schistosoma Japonicum Infection Status At The Population Level

Schistosomiasis is recognized as one of the major tropical parasite diseases that severely threaten people's health and affect social and economical development in the world, and it ranks second only to malaria. It is widely distributed among seventy-six countries of the tropical world. Of the three major human disease-causing species, the Oriental schistosome, S. japonicum is distributed throughout China, the Philippines and part of Indonesia. Over the past 60 years, the chinese government has made great progress to control this disease and the criteria of transmission control or interruption have been achieved in 73.11 percent of the endemic countries. However, according to the report on national epidemiological situation of schistosomiasis japonica in 2006, there are still 671,265 people infected with S. japonicum, and 60 million people are at risk. Therefore, schistosomiasis control with the ultimate goal of disease elimination is still a long-term challenge in China.Diagnosis has always been critical in schistosomiasis control. There are two aspects of diagnosis, one at individual (or clinical) and the other at the population (or epidemiology) level. Methods belong to the former category focus on individual disease identification so as to define each patient for proper interventions to be carried out. Comparatively, the population level diagnosis is of more importance because it provides community-based infection status that is vital for the implementation of public policy on disease control. It can also be applied to monitor dynamic change of endemicity, to evaluate epidemic situation and to assess control efficacy, etc. at the population level.Generally, the conventional approach to diagnosis schistosomiasis at population level is the community-based epidemiological surveys in endemic areas to determine the indicators of the population infection status, such as prevalence, intensity of infection. However, this approach may be infeasible in the long run due to huge time and monetary expenses and low compliance of study population. Due to high withdraw rate, long-term dynamic change of the disease is barely achievable. In addition, the cross-sectional investigations usually carried out in endemic areas only provide infection status at a single time-point rather than disease dynamics and prevalence pattern. Moreover, these meticulously planned trials can be easily disturbed by the external events such as flooding, global warming and other climate-related nature disasters.There are mainly three approaches for epidemiological study: observational, experimental and theoretical research. Most recently, theoretical epidemiological research, which also called epidemiological mathematical modeling, have been widely used to simulate many biological systems. It can provide useful insights into disease transmission pattern prediction and the cooperative effects of etiology, host and the environment. The complexity of disease transmission course can be described more typically, briefly and quantitatively using this approach. This enables one to better understand the epidemiology of disease and can be useful for making rational decisions for disease control. Therefore, the epidemiological mathematical modeling has been widely used to evaluate the prevalence or intensity of many infectious diseases. The most important is that this method can avoid many shortcomings of epidemiological surveys.In our study, we explored modeling method to evaluate human S. japonicum infection status at the population level. It is anticipated that this research can provide evidence for the schistosomiasis control decisions making from the theoretical aspect. There are 3 sections in the dissertation.1. The assessment of community prevalence using the prevalence of infection in school-aged children as an index in endemic areas of schistosomiasis japonica Community-based surveys are expensive and time-consuming due to the size of samples required as well as low compliance and high subject withdraw rate. Therefore, our study choose to investigate infection status in school-aged children and use it as a sentinel population. The relationship between the prevalence of S. japonicum infection in school-aged children and that in the community population were investigated in this part of study.We retrieved the age-stratified S. japonicum infection data from epidemiological studies in China that have been published in CNKI and Pubmed since 1990. Twenty-five literatures were collected in our study according to certain inclusion and exclusion criteria. The relationship of prevalences between school-aged children and community population were fitted using linear and logistic regression models which take into account variation in sample prevalences. Adjusted logistic models showed a strong relationship between the observed data on the prevalence of infection in the entire population and the prevalence in school-aged children. Even when the school-aged children are removed from the residual population to avoid effects of autocorrelation, there is still a significant relationship between prevalence of school-aged children and those of infants or adults using adjusted logistic models. It indicated that despite the wide variations in study sites and survey conducting dates, a remarkably consistent pattern emerged in the relationship between the prevalence of infection in school-aged children and the prevalence in the community.The results suggest that the prevalence of infection in school-aged children could serve as a cost-effective predictive tool that can be used at a district/national level to identify target areas for control and to evaluate the numbers at risk of infection in China.2. Application of mathematical models based on differential equation in the transmission dynamics of schistosomiasis japonica in endemic areasThe results in the first part of our study indicate that the prevalence of infection in school-aged children could provide an index for determining community prevalence. Also, we may undertake school-based surveys rather than community-based surveys in the future to decrease participants'withdraw while lower the costs. However, like the traditional community-based surveys described above, this study still can only provide human population infection status at a single point. Disease dynamics prevalence patterns can't be observed and predicted.In this study, we use the Barbour two-host model to predict the community-based prevalence of schistosomiasis japonica in Poyang lake region in year 2006. Parameters are estimated by the use of Chinese data we are collecting on schistosomiasis japonica transmission in the Poyang lake area of Jiangxi Province in China. The results show that the one-year effects of this model are satisfactory. Comparing the effects of different control strategies on human prevalence, the model predicts that: (1) A combination of human and bovine treatment will dramatically reduce human prevalence and maintains the reduction for a longer period of time than treatment of a single host. (2) Selective human treatment accompanied by bovine vaccine and by human behavior intervention, elimination of the disease is possible.3. The establishment of a Schistosomiasis japonica Cellular Automata (SjCA) modelEquation-based models described above are performed in continuous system simulation, and the process of translating the population dynamics and stochastics into differential equation form is not easy. Furthermore, those models only tracked the average worm burden without taking the variations among individuals into account. It is important to point out that omitting the stochasticity or the discrete nature of the individuals often leads to inconsistent results. Therefore, the use of those equation-based models in the practice of schistosomiasis control is restricted. Most recently, computer stochastic simulating methods, including cellular automata (CA), have been widely used to mimic the transmission and immune dynamics of diseases. In view of this, a stochastic model based on CA (we name it SjCA, for Schistosomiasis japonica Cellular Automata) that is different from those that have been proposed most frequently in the past is introduced to describe the transmission of schistosomiasis japonica and predict the human population S. japonicum infection status in this part of our study.The SjCA model was utilized to predict the prevalence and intensity of infection as the outcomes of the selected chemotherapy carried out in Jiahu village. The inputs of the SjCA model were estimated by use the data collected in Jiahu village in 2005 or based on field data and estimated equation in expert opinion. Comparing model-predicted prevalence and intensity of S. japonicum infection with the observed data, the effects of the model are shown to be satisfactory. It is anticipated that our SjCA model can serve as a tool to simulate schistosomiasis transmission dynamics in endemic areas in China. Furthermore, SjCA model shows that coverage of selected chemotherapy should be no less than 85 percent to guarantee an effective drug control program.In summary, human S. japonicum infection status were estimated at the population level using three mathematical models in our study, in hope of having this work provides useful information for schistosomiasis control in China. We firstly proposed that the prevalence of S. japonicum infection at the population level could be estimated by the S. japonicum infection rate of school-aged children in the same community using adjusted logistic model in China. In the second part of our study, Barbour two-host model was used to predict the S. japonicum prevalence in a community one year after selective chemotherapy. The result is consistent with the observed data. We also evaluated the effects of various control strategies using the same model. In the last part of the study, we established a new stochastic simulation model which was named as SjCA model, to simulate the transmission dynamics of schistosomiasis japonica. The newly eatablished SjCA model is satisfactory judged by comparing the predicted and observed human S. japonicum infection status. It is anticipated that SjCA model can serve as a new approach to estimate S. japonicum infection status at population level in the field. The three parts of our study revealed epidemiological characteristics and patterns of schistosomiasis japonica in the survey area in China. We demonstrated three approaches for the estimation of S. japonicum infection status. The models illustrated are simulation of infection status from one-dimensional (cross-sectional analysis), to two-dimensional (community prevalence dynamics simulation), and furthermore a three-dimentional method for individual infection dynamics simulation. Among the three mathematical models, SjCA model is the best model in theory because of the most information it provides and better simulation results. This model warrants be optimized and widely applied in the transmission dynamics of many other infectious diseases.