Open Access Repository

An ensemble-based feature selection methodology for case-based learning

Downloads

Downloads per month over past year

Ali, M ORCID: 0000-0002-4107-7122 2018 , 'An ensemble-based feature selection methodology for case-based learning', PhD thesis, University of Tasmania.

[img]
Preview
PDF (Whole thesis)
Ali_whole_thesi...pdf | Download (4MB)

| Preview

Abstract

Case-based learning (CBL) approach has been receiving a lot of attention in medical education, as an alternative to the traditional learning environment. This student-centric teaching methodology, exposes the medical students to practice the real-world scenarios. In order to support the learning outcomes of students, existing systems do not provide computer-based as well as experiential knowledge-based support for CBL practice. Medical literature contains textual knowledge, which can be used as a very beneficial source for the computer-based CBL practice. Therefore, designing and developing of an automated CBL approach is a challenging problem. In order, to solve this problem, the text mining domain provides the basic framework for constructing domain knowledge, where the feature selection is considered to be one of the most critical requirement to select the appropriate features.
Keeping in view these facts, this research, provides contribution, in the following areas: (1) Feature Ranking; where we propose, an innovative unified features scoring algorithm to generate a final ranked list of features, (2) Feature Selection; where we propose, an innovative threshold value selection algorithm to define a cut-off point for removing irrelevant features for the domain knowledge construction, and (3) CBL Platform; where we designed and developed, an interactive CBL system consisting of experiential as well as domain knowledge to nurture medical students for their professional learning and development. We perform both quantitative and qualitative evaluation of our proposed (1) methodology on benchmark datasets, and (2) CBL approach. The extensive experimental results show that our approach provides competitive accuracy and achieved (1) on average, more than 5% increase in f-measure and predictive accuracy as compared to state-of-the-art methods, and (2) a success rate of more than 70% for students’ interaction, group learning, solo learning, and improving clinical skills.

Item Type: Thesis - PhD
Authors/Creators:Ali, M
Keywords: feature selection, ensemble learning, classification, case-based learning, teaching methodology, medical education, clinical case
DOI / ID Number: 10.25959/100.00030874
Copyright Information:

Copyright 2018 the author

Additional Information:

The author studied for and was awarded a conjoint doctoral degree at both the University of Tasmania and Kyung Hee University.

Related URLs:
Item Statistics: View statistics for this item

Actions (login required)

Item Control Page Item Control Page
TOP