Cardiovascular disease risk prediction using machine learning / Mohammad Najibudin Hakiim bin Haji Mohammad Midun
Material type:
TextPublication details: Bandar Seri Begawan : Universiti Teknologi Brunei, 2023. Description: 59 pages : color illustrations ; 30 cmSubject(s): -- Thesis Universiti Teknologi Brunei | -- Final Year Report Universiti Teknologi Brunei | Dissertation, Academic | Thesis writing | Dissertation Universiti Teknologi Brunei | Cardiovascular diseaseOther classification: UTB 120 REPORT, THESIS & DISSERTATION | RTDS 399 Dissertation note: Dissertation (Master of Science) - Universiti Teknologi Brunei (2023)
| Item type | Current library | Call number | Copy number | Status | Notes | Date due | Barcode |
|---|---|---|---|---|---|---|---|
Reports, Thesis & Dissertation Students
|
Universiti Teknologi Brunei Library - at level 2 | UTB 120 REPORT, THESIS & DISSERTATION, RTDS 399 (Browse shelf(Opens below)) | c. 1 | Not for loan | Reg. No. 002443_UTB [RTDS 399] | 850571 |
Submitted in fulfilment of the requirements for the degree of Master of Science
ABSTRACT One of the leading causes of deaths worldwide is CVD (cardiovascular disease). It is a disease found on the heart, blood vessels or any part of the circulatory system. Machine learning refers to the study of computer system which uses algorithms and datasets to adapt, learn, and predict the values with a bare human intervention.
This research proposes the applications of machine learning to predict CVD. Machine learning models was assembled with four different algorithms and two datasets converted from regional risk charts data with each dataset having two variants and standardised. The model is trained ten times to further clarify its performance. Performance is determined by the metrics of the Mean R'(coefficient of determination) and cross validation accuracy to measure their performance.
Best performing algorithm is Random Tree with a mean R° score range of up to 0.9807 with standard deviation of 0.0031 to 0.0069. Mean CV (Cross Validation) ranges from 0.6996 to 0.8782 with standard deviations of 0.1413 to 0.2726. The low standard deviations and high CV scores implies stability and low variance in predicting CVD. KNN (K Nearest Neighbour) and Linear Regression performs very moderately but SV (Support Vector Machine) obtained the highest mean R° scores from 0.9214 to 0.9978 with standard deviations of 0.0003 to 0.0175. However, the cross validation is very low (highest is
0.0270) and it obtained very high standard deviations when compared to the other algorithms. This indicates the presence of overfitting or lack of compatibility with the
datasets.
In conclusion, using the risk charts as substitute for as a Bruneian dataset offered a valuable perspective that CVD risk charts could be converted and used as a dataset to reliably predict CVD as shown with results. In this case, it could reliably predict the chances of a Bruneian developing CVD.
Keywords: Machine Learning, Cardiovascular Diseases, Machine Learning, Data Analytics, Medical Informatics
Dissertation (Master of Science) - Universiti Teknologi Brunei (2023)
Includes references pages 44-47
There are no comments on this title.