Neural networks are used in this chapter for classifying the HIV status of individuals based on socioeconomic and demographic characteristics. The trained network is then used to create an error equation with one of the demographic variables as a missing input and the desired HIV status as one of the variables. The missing variable thus becomes a control variable. This control mechanism is proposed to assess the effect of education level on the HIV risk of individuals and, thereby, assist in understanding the extent to which the spread of HIV can be controlled by using the education level. An inverse neural network model and a missing data approximation model based on autoassociative neural network and genetic algorithm (ANNGA) are used for the control mechanism. Therefore, the ANNGA is used to obtain the missing input values (education level) for the first model and an inverse neural network model is then used to obtain the missing input values (education) for the second model. The two models are then compared and it is found that the proposed inverse neural network model outperforms the ANNGA model. The methodology thus shows that HIV spread can be controlled to some extent by modifying a demographic characteristic educational level.
TopIntroduction
Acquired Immunodeficiency Syndrome (AIDS) was first defined in 1982 to depict the first cases of strange immune system failures that were observed in the preceding years. The Human Immunodeficiency Virus (HIV) was soon after identified as the origin of AIDS. From the time of the detection of the virus and the disease a great deal of effort has been done to stop the spread of HIV with very little success. AIDS is currently an epidemic, which at the end of 2003 had killed an estimated 2.9 million lives. Epidemiology examines the function of host, agent and environment to elucidate the occurrence and transmission of a disease. Risk factor epidemiology examines the individual demographic and social characteristics and attempts to identify causes that position an individual at risk of acquiring a disease (Poundstone, Strathdee, & Celectano, 2004).
In this chapter, the demographic and social characteristics of the individuals and their behavior are used to establish the risk of HIV infection and this process is called “biomedical individualism” (Poundstone, Strathdee, & Celectano, 2004; Fee & Krieger, 1993; Leke et al., 2006a&b; Leke & Marwala, 2007). Because of the quick spread of the virus in the world today, especially in Sub-Saharan Africa, there has been an increase in the need to identify strategies and mechanisms for controlling the virus. This, on the other hand, seems ineffective given that the virus spread still appears to be uncontrolled particularly in the developing world. Social factors influence the risk of exposure as well as the probability of transmission of the disease. These social factors are, therefore, essential in order to comprehend and model the disease. By identifying the individual risk factors that lead to the disease, it is feasible to adjust social conditions which give rise to the disease and, therefore, design successful HIV intervention policies (Poundstone, Strathdee, & Celectano, 2004; Leke, Marwala, & Tettey, 2006; Leke, Marwala, & Manana, 2008). Using this information, a model can be constructed and then used to control the disease using conventional adaptive control strategies (Widrow & Walach, 1993; Aoyama, Doyle III, & Venkatasubramanian, 1999; Akin, Kaya, & Karakose, 2003; Acosta & Todorovich, 2003; Altinten et al., 2003; Ahn & Kha, 2007).
Adaptive control theory provides a plausible way to solve many of complex problems. Two distinct approaches can be used to control a system adaptively; the direct adaptive control and indirect adaptive control. In the direct control approach, the parameters of the controller are directly adjusted to reduce a distance of the output error. In the indirect control, the parameters of the plant are estimated as elements of a vector at any instant k; and the parameters vector of the controller is adapted based on the estimated plant vector. A general configuration of the indirect adaptive control as a self-tuning controller is shown in Figure 1 (Widrow & Walach, 1993). At each sampling instant, the input and output of the generating unit are sampled and a plant model is obtained by an on-line identification algorithm to represent the dynamic behavior of the generating unit at that instant in time. The required control signal is computed based on the identified model and various control techniques can be used to compute the control signal. All control algorithms assume that the identified model is a good approximation of the system that needs to be controlled. In this chapter, control algorithm is created using computational intelligence.
Figure 1. A general adaptive controller