Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities
The cost of healthcare is currently a huge burden to governments and health care organisations across the world. In South Africa, laboratory tests administered by government facilities are delivered by the National Health Laboratory Service (NHLS) regardless of payment, and hence there is a possibil...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Department of Statistical Sciences
2024
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613221421056001 |
|---|---|
| access_status_str | Open Access |
| author | Nantongo, Ssozi Margaret Eva |
| author2 | Er Sebnem |
| author_browse | Er Sebnem Nantongo, Ssozi Margaret Eva |
| author_facet | Er Sebnem Nantongo, Ssozi Margaret Eva |
| author_sort | Nantongo, Ssozi Margaret Eva |
| collection | Thesis |
| description | The cost of healthcare is currently a huge burden to governments and health care organisations across the world. In South Africa, laboratory tests administered by government facilities are delivered by the National Health Laboratory Service (NHLS) regardless of payment, and hence there is a possibility that certain tests are over ordered by doctors at government health facilities. Gatekeeping is a demand management tool utilised by facilities across the world to manage costs of laboratory testing. Electronic gatekeeping addresses inappropriate laboratory test ordering to reduce the over ordering or under ordering of tests. In South Africa, the electronic gatekeeping (eGK) system is a standardised set of rules that was developed by the National Department of Health (NDOH) as well as NHLS pathologists and clinicians from the individual provinces (NHLS, 2017; Pema et al., 2018; Smit et al., 2015). The eGK system restricts test ordering by applying a given set of rules to tests ordered by a medical official for each patient. The protocols followed by the eGK system are defined using criteria such as date or result of previous test and location/ward of patient. This project aims to identify facilities and wards that are incurring high violations of tests subject to eGK rules. Anomaly detection methods are utilised to identify these facilities and wards together with the tests that require intervention to address the high violations. Three methods were utilised for anomaly detection and included K-means clustering, isolation forests and one-class Support Vector Machines (SVM). The recommended wards for intervention were mostly the maternity related wards at major hospitals. Within these wards, there was evidence of ordering tests that violated the eGK rules more than other wards. Other wards with evidence of over-ordering and violation of eGK rules included ARV Clinic ward, cardiac wards, high care units and respiratory ICU wards. The tests that were selected for intervention in these wards included calcium, magnesium, inorganic phosphate, total protein, albumin, bilirubin tests, creactive protein, procalcitonin and rubella PCR. The facilities selected for intervention included major hospitals for example Nelson Mandela Academic Hospital, Port Elizabeth Provincial Hospital, Livingstone Hospital and Dora Nginza Hospital. In addition, district hospitals and specialised TB hospitals were selected amongst those recommended for intervention. The tests selected for intervention in these facilities included calcium, magnesium, inorganic phosphate, total protein, albumin, c-reactive protein, procalcitonin, Hepatitis B DNA and CA 15-3. The results of the analysis were compared with results from published literature, and it was found that some of the tests recommended for intervention were also highlighted by previous researchers, for example c-reactive protein tests. A comparison of the results from the K-means clustering, one-class SVM and isolation forests anomaly detection showed that the same wards, facilities, and tests were recommended for intervention. Therefore, anomaly detection is a suitable method for identification of wards and facilities that are violating test ordering rules more than other facilities. |
| format | Thesis |
| id | oai:open.uct.ac.za:11427/39743 |
| institution | University of Cape Town (South Africa) |
| language | eng |
| last_indexed | 2026-06-10T12:32:41.376Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UCTD — University of Cape Town Open Access Repository |
| publishDate | 2024 |
| publishDateRange | 2024 |
| publishDateSort | 2024 |
| publisher | Department of Statistical Sciences |
| publisherStr | Department of Statistical Sciences |
| record_format | dspace |
| source_str | UCTD — University of Cape Town Open Access Repository |
| spelling | oai:open.uct.ac.za:11427/39743 Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities Nantongo, Ssozi Margaret Eva Er Sebnem Silal Sheetal Data Science The cost of healthcare is currently a huge burden to governments and health care organisations across the world. In South Africa, laboratory tests administered by government facilities are delivered by the National Health Laboratory Service (NHLS) regardless of payment, and hence there is a possibility that certain tests are over ordered by doctors at government health facilities. Gatekeeping is a demand management tool utilised by facilities across the world to manage costs of laboratory testing. Electronic gatekeeping addresses inappropriate laboratory test ordering to reduce the over ordering or under ordering of tests. In South Africa, the electronic gatekeeping (eGK) system is a standardised set of rules that was developed by the National Department of Health (NDOH) as well as NHLS pathologists and clinicians from the individual provinces (NHLS, 2017; Pema et al., 2018; Smit et al., 2015). The eGK system restricts test ordering by applying a given set of rules to tests ordered by a medical official for each patient. The protocols followed by the eGK system are defined using criteria such as date or result of previous test and location/ward of patient. This project aims to identify facilities and wards that are incurring high violations of tests subject to eGK rules. Anomaly detection methods are utilised to identify these facilities and wards together with the tests that require intervention to address the high violations. Three methods were utilised for anomaly detection and included K-means clustering, isolation forests and one-class Support Vector Machines (SVM). The recommended wards for intervention were mostly the maternity related wards at major hospitals. Within these wards, there was evidence of ordering tests that violated the eGK rules more than other wards. Other wards with evidence of over-ordering and violation of eGK rules included ARV Clinic ward, cardiac wards, high care units and respiratory ICU wards. The tests that were selected for intervention in these wards included calcium, magnesium, inorganic phosphate, total protein, albumin, bilirubin tests, creactive protein, procalcitonin and rubella PCR. The facilities selected for intervention included major hospitals for example Nelson Mandela Academic Hospital, Port Elizabeth Provincial Hospital, Livingstone Hospital and Dora Nginza Hospital. In addition, district hospitals and specialised TB hospitals were selected amongst those recommended for intervention. The tests selected for intervention in these facilities included calcium, magnesium, inorganic phosphate, total protein, albumin, c-reactive protein, procalcitonin, Hepatitis B DNA and CA 15-3. The results of the analysis were compared with results from published literature, and it was found that some of the tests recommended for intervention were also highlighted by previous researchers, for example c-reactive protein tests. A comparison of the results from the K-means clustering, one-class SVM and isolation forests anomaly detection showed that the same wards, facilities, and tests were recommended for intervention. Therefore, anomaly detection is a suitable method for identification of wards and facilities that are violating test ordering rules more than other facilities. 2024-05-30T07:51:10Z 2024-05-30T07:51:10Z 2023 2024-05-28T08:31:47Z Thesis / Dissertation Masters MSc http://hdl.handle.net/11427/39743 eng application/pdf Department of Statistical Sciences Faculty of Science |
| spellingShingle | Data Science Nantongo, Ssozi Margaret Eva Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| thesis_degree_str | Master's |
| title | Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| title_full | Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| title_fullStr | Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| title_full_unstemmed | Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| title_short | Anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| title_sort | anomaly detection in laboratory tests subject to gatekeeping in selected health facilities |
| topic | Data Science |
| url | http://hdl.handle.net/11427/39743 |
| work_keys_str_mv | AT nantongossozimargareteva anomalydetectioninlaboratorytestssubjecttogatekeepinginselectedhealthfacilities |