Multi-class classification model for dementia, heart failure, and aorticv valve disorder

Name: Khomkrit Yongcharoenchaiyasit

keyword: Aortic Valve Disorder

MeSH: Cardiovascular system -- Diseases

Classification :.DDC: e-thesis

; Cardiovascular Disease

MeSH: Dementia

MeSH: Ensemble learning (Machine learning) -- 307073

MeSH: Heart failure -- 47749

MeSH: Machine learning

MeSH: Aortic valve -- 309917

LCSH: Cardiovascular system -- Diseases

LCSH: Dementia

LCSH: Ensemble learning (Machine learning) -- 307073

LCSH: Heart failure -- 47749

LCSH: Machine learning

Abstract: Dementia is a group of symptoms that affect memory, analytical thinking, and social abilities in the elderly population. Early diagnosis of dementia is crucial for reducing or mitigating mortality rates. Machine learning algorithms, given the wide availability of electronic health records, have proven capable of disease prediction. However, developing effective disease predictive models presents a challenge due to imbalanced, low-dimensional, and complex data. For this study, the multi-class classification model was proposed to distinguish dementia from its potential causes, including heart failure and aortic valve disorder. The extreme gradient boosting (XGBoost)-based model was proposed for multi-class classification of dementia, heart failure, and aortic valve disorder in the elderly population due to its strong capability to handle complex data and achieve high predictive accuracy. The model was trained and tested on the data collected from Chiang Rai Prachanukroh Hospital and Thoeng Hospital in Chiang Rai, Thailand, consisting of 21,201 records. To address low dimensional data, various prior knowledge-based feature creation techniques were employed to create additional features. In addition, the borderline synthetic minority over-sampling technique was applied to address the class imbalance issue. The performance of the proposed XGBoost-based model was compared with several existing methods, including support vector machine, k-nearest neighbors, decision tree, random forest, extra trees, gradient boosting, and TabNet. Hyperparameter optimization was performed for all models. Classification performance was evaluated using the confusion matrix-based metrics including precision, recall, F1 score, accuracy, area under the receiver operating characteristic curve, and area under the precision-recall curve. In conclusion, the XGBoost-based model achieved superior performance compared to other models based on these metrics.

LCSH: Aortic valve -- 309917

Mae Fah Luang University. Learning Resources and Educational Media Centre

Address: CHIANG RAI

Email: library@mfu.ac.th

Name: Punnarumol Temdee

Role: Advisor

Created: 2023

Modified: 2024-08-08

Issued: 2024-08-08

วิทยานิพนธ์/Thesis

application/pdf

CallNumber: e-thesis

eng

DegreeName: Master of Engineering

Level: Master's Degree

Descipline: Computer Engineering

Grantor: Mae Fah Luang University

©copyrights Mae Fah Luang University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	138127.pdf	8.56 MB	4	2025-10-03 15:27:34

ใช้เวลา

0.019472 วินาที