Comparative Performance Evaluation of Artificial Neural Network and Hidden Mackov Model for Predicting Substance Abuse Neuron Disorder Rate

Abstract

Researchers in the past have adopted different techniques such as the traditional statistical models, hidden mackov models for forecasting into the distance future but this approach is becoming outdated and gradually reaching their limitation in application due to incomplete precision and accuracy, for the purpose of complete accuracy and precision, neural network model technology has been used in this application work where the problem domain mainly involves prediction model This study embraced the use of neural network model technology to forecasting precision for the control of Substance Abuse Neuron Disorder (SAND) rate. This enormous task methodology involve first the theoretical gathering of psychiatric life data from World Health Organization (WHO), then using different well-known neuron disorder diseases data were mathematically models and simulated in matrix laboratory tools with Artificial Neural Network (ANNM) and Hidden Mackov Model (HMM) as examples. Secondly, consider major variables upon which new model which is more realistically represents actual disease prediction is built. Three components are implemented: SAND; is a targeted non communicable disease with Routine surveillance, modeling the disease risk with historical data surveillance with contemporary environmental data; and Forecasting future risk with comparative evaluation using two predictive metrics models Mean Square Error (MSE), Standard Deviation (SDEV) as paramount tools in epidemiological surveillance. Average MSE and SDEV of HFSANDR for each periods was 0.7 and 0.8 which is close to (1.0) while that of AFSANDR performance was between 0.3 to 0.4 less than (0.5) which indicates the weakness of HFSANDR for prediction Finally the evaluation shows that ANN prediction model performs better than HMM with more than 18 percent in term of accuracy