Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum

Essentially, determination model for degree of injury is crucial for refining diagnostic and increasing accuracy of the Forensic and Medicolegal services. Existing models are deemed difficult in identifying the critical features. These are due to the model having insufficient of critical features an...

Full description

Saved in:

Bibliographic Details
Main Author:	Wardhana, Mohammad Hadyan
Format:	Thesis
Language:	English English
Published:	2023
Subjects:	Q Science (General) QA Mathematics
Online Access:	http://eprints.utem.edu.my/id/eprint/28268/1/Hybrid%20neural%20network%20in%20medicolegal%20degree%20of%20injury%20determination%20based%20on%20Visum%20et%20Repertum.pdf http://eprints.utem.edu.my/id/eprint/28268/2/Hybrid%20neural%20network%20in%20medicolegal%20degree%20of%20injury%20determination%20based%20on%20Visum%20et%20Repertum.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!

id	my-utem-ep.28268
record_format	uketd_dc
spelling	my-utem-ep.282682024-12-16T07:51:42Z Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum 2023 Wardhana, Mohammad Hadyan Q Science (General) QA Mathematics Essentially, determination model for degree of injury is crucial for refining diagnostic and increasing accuracy of the Forensic and Medicolegal services. Existing models are deemed difficult in identifying the critical features. These are due to the model having insufficient of critical features analysis that cause the inconsistency decision to determine degree of injury among the medical practitioners. The issue become more complex because the dataset consists of incomplete data and outliers class problem that can affects the sampling bias. The purpose of this study is to identify the characteristics and terms, develop and evaluate the Hybrid Neural Network Model (HNNM) for determining degree of injury based on Visum et Repertum (VeR) data. The VeR data consist of 289 patients’ record. The HNNM is expected to determine either the persecution victim having a minor, moderate, or serious injury which inclusively mention in Indonesian Penal Code. HNNM is developed based on the case studies at three hospitals in Pekanbaru comprise three main phases which are pre-processing, development, and performance analysis. Pre-processing phase overcomes the issue of incomplete data by performing data cleansing and data normalization. The development phase begins with utilizing Analytical Hierarchical Process (AHP) to validate the ranking for each of weight on the critical features from the experts’ opinion. Then, the selection of the critical features is chosen via Neural Network (NN) as classification algorithm and Genetic Algorithm (GA) as an optimization technique. The selected critical features are applied during the dataset training stages to improve the accuracy and reduce error of the HNNM. GA is aimed to increase the accuracy and minimize the error in the learning stages of NN. The development phase accomplished with testing stages by employing VeR dataset. The performance analysis shows the HNNM produced 98.85% accuracy level and Root Mean Square Error (RMSE) value at 0.077. In the validation stage, the questionnaires are answered by the Subject Matter Expert (SME) groups which consist of feature, implementation, and viability aspect of HNNM. Result from the questionnaires concluded that the agreement level of SMEs reaches up to 80%. Thus, the features of the HNNM are implementable and highly acceptable by the practitioner. For the future research, the HNNM need to increase the accuracy by improving the input features including lifestyle, habit, and job. 2023 Thesis http://eprints.utem.edu.my/id/eprint/28268/ http://eprints.utem.edu.my/id/eprint/28268/1/Hybrid%20neural%20network%20in%20medicolegal%20degree%20of%20injury%20determination%20based%20on%20Visum%20et%20Repertum.pdf text en public http://eprints.utem.edu.my/id/eprint/28268/2/Hybrid%20neural%20network%20in%20medicolegal%20degree%20of%20injury%20determination%20based%20on%20Visum%20et%20Repertum.pdf text en validuser https://plh.utem.edu.my/cgi-bin/koha/opac-detail.pl?biblionumber=123811 phd doctoral Universiti Teknikal Malaysia Melaka Faculty of Information and Communication Technology Hasan Basari, Abd Samad
institution	Universiti Teknikal Malaysia Melaka
collection	UTeM Repository
language	English English
advisor	Hasan Basari, Abd Samad
topic	Q Science (General) QA Mathematics
spellingShingle	Q Science (General) QA Mathematics Wardhana, Mohammad Hadyan Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum
description	Essentially, determination model for degree of injury is crucial for refining diagnostic and increasing accuracy of the Forensic and Medicolegal services. Existing models are deemed difficult in identifying the critical features. These are due to the model having insufficient of critical features analysis that cause the inconsistency decision to determine degree of injury among the medical practitioners. The issue become more complex because the dataset consists of incomplete data and outliers class problem that can affects the sampling bias. The purpose of this study is to identify the characteristics and terms, develop and evaluate the Hybrid Neural Network Model (HNNM) for determining degree of injury based on Visum et Repertum (VeR) data. The VeR data consist of 289 patients’ record. The HNNM is expected to determine either the persecution victim having a minor, moderate, or serious injury which inclusively mention in Indonesian Penal Code. HNNM is developed based on the case studies at three hospitals in Pekanbaru comprise three main phases which are pre-processing, development, and performance analysis. Pre-processing phase overcomes the issue of incomplete data by performing data cleansing and data normalization. The development phase begins with utilizing Analytical Hierarchical Process (AHP) to validate the ranking for each of weight on the critical features from the experts’ opinion. Then, the selection of the critical features is chosen via Neural Network (NN) as classification algorithm and Genetic Algorithm (GA) as an optimization technique. The selected critical features are applied during the dataset training stages to improve the accuracy and reduce error of the HNNM. GA is aimed to increase the accuracy and minimize the error in the learning stages of NN. The development phase accomplished with testing stages by employing VeR dataset. The performance analysis shows the HNNM produced 98.85% accuracy level and Root Mean Square Error (RMSE) value at 0.077. In the validation stage, the questionnaires are answered by the Subject Matter Expert (SME) groups which consist of feature, implementation, and viability aspect of HNNM. Result from the questionnaires concluded that the agreement level of SMEs reaches up to 80%. Thus, the features of the HNNM are implementable and highly acceptable by the practitioner. For the future research, the HNNM need to increase the accuracy by improving the input features including lifestyle, habit, and job.
format	Thesis
qualification_name	Doctor of Philosophy (PhD.)
qualification_level	Doctorate
author	Wardhana, Mohammad Hadyan
author_facet	Wardhana, Mohammad Hadyan
author_sort	Wardhana, Mohammad Hadyan
title	Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum
title_short	Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum
title_full	Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum
title_fullStr	Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum
title_full_unstemmed	Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum
title_sort	hybrid neural network in medicolegal degree of injury determination based on visum et repertum
granting_institution	Universiti Teknikal Malaysia Melaka
granting_department	Faculty of Information and Communication Technology
publishDate	2023
url	http://eprints.utem.edu.my/id/eprint/28268/1/Hybrid%20neural%20network%20in%20medicolegal%20degree%20of%20injury%20determination%20based%20on%20Visum%20et%20Repertum.pdf http://eprints.utem.edu.my/id/eprint/28268/2/Hybrid%20neural%20network%20in%20medicolegal%20degree%20of%20injury%20determination%20based%20on%20Visum%20et%20Repertum.pdf
_version_	1818612060781543424

Hybrid neural network in medicolegal degree of injury determination based on Visum et Repertum

Similar Items