Publication:
A new regression modelling approach and its application in biostatistics

Loading...
Thumbnail Image
Date
2024-07
Authors
Eusufzai, Sumaiya Zabin
Journal Title
Journal ISSN
Volume Title
Publisher
Research Projects
Organizational Units
Journal Issue
Abstract
This research aims to develop an integrated methodology that will be formulated within a Multilayer Feedforward Neural Network (MLFFNN) framework and logistic regression. The mean absolute deviation and predicted mean square error will be utilised to evaluate the performance of the integrated model, and it serves as a yardstick to determine the accuracy and efficiency of the forecasting that is achieved as a result. The urgency of better significant results serves as a motivation for this study. The objective of this study is to develop and implement an integrated model combining Bootstrap and MLFFNN with logistic regression modelling (LRM) to achieve better prediction accuracy and interpretability. The integrated method used in this study is based on the principles of Bootstrap, LRM, and MLFFNN. The accuracy of the proposed technique is assessed using the Predicted Mean Squared Error Neural Network (PMSE.net), Mean Absolute Deviance (MAD), and the accuracy percentage. Each of these components acts as a benchmark for assessing the precision and effectiveness of the created model. A t-test was used to explore the difference between actual and predicted values from the models. Data analysis was conducted using the R program and SPSS version 26. Two case studies from dental public health have been used to validate this newly developed integrated model, i) dental caries among preschool children, and ii) The case study for oral health knowledge among mothers of preschool children. The incorporation of bootstrapping, MLFFNN, and logistic regression in an integrated approach enhances the accuracy of parameter estimation and addresses the uncertain relationship between dependent and independent variables. In the case study focusing on dental caries among preschool children, the Mean Absolute Deviation (MAD) is 0.0221126 and the Predicted Mean Squared Error (PMSE) is 0.07909. A paired sample t-test reveals no significant difference between the actual and predicted values, with means and standard deviations as follows: Actual (Mean [SD] = 0.30 [0.483]) and Predicted (Mean [SD] = 0.31 [0.373]); df = -0.067(9); p-value > 0.05. In the study concerning oral health knowledge among mothers of preschool children, the MAD is 0.05303337, and the PMSE is 0.053033. Results from the paired sample t-test indicate no significant difference between actual and predicted values, with means and standard deviations as follows: Actual (Mean [SD] = 0.600 [0.940]) and Predicted (Mean [SD] = 0.940 [0.030]); df = -2.154(9). This study’s findings will considerably contribute to epidemiological methodology research, particularly relationship mapping, by introducing an integrated model. Concerning MAD, PMSE, and p-value, these indicate both models showed high accuracy in outcome prediction. The significance of the produced syntax outcome will suggest a more accurate decision-making process in disease prevention.
Description
Keywords
Citation