Lee, Carmen Kar Hang, Tse, Ying Kei ![]() |
Preview |
PDF
- Accepted Post-Print Version
Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (529kB) | Preview |
Abstract
The digitalisation in healthcare opens opportunities for more effective chronic disease management. Digitalised medical records are valuable data sources for identifying high-risk patients and facilitating early clinical intervention. However, the liberation of data has plagued adoption amongst physicians as massive data mean more difficult to identify important knowledge from the data. In the cervical cancer context, many patients are adherence to prescription medications only when symptoms appear, beyond the earlier point-in-time of the disease progression. Regular screening is the only way to detect abnormal cells that may develop into cancer if left untreated. Yet, without a comprehensive understanding of the relationship between risk factors and healthcare outcomes, inappropriate screening procedures may be conducted, lengthening the treatment process. Delay in the treatment process may have an irreversible influence on patients’ conditions as chronic diseases progress. This study demonstrates a data-mining framework which extracts knowledge that can advance cervical cancer screening processes in the form of association rules and improves the generalisation potential of the rules for deployment. The knowledge discovered serves as an additional supplement for physicians’ experience and uncovers appropriate screening strategies based on patients’ risk factors, increasing the chances of high-risk patients getting treated for cervical pre-cancers.
Item Type: | Article |
---|---|
Date Type: | Publication |
Status: | Published |
Schools: | Business (Including Economics) |
Publisher: | Elsevier |
ISSN: | 0040-1625 |
Date of First Compliant Deposit: | 21 September 2020 |
Date of Acceptance: | 19 September 2020 |
Last Modified: | 07 Nov 2023 21:48 |
URI: | https://orca.cardiff.ac.uk/id/eprint/134971 |
Citation Data
Cited 10 times in Scopus. View in Scopus. Powered By Scopus® Data
Actions (repository staff only)
![]() |
Edit Item |