Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study
Abstract Background Many studies have evaluated stroke using claims data; most of these studies have defined ischemic stroke using an operational definition following the rule-based method. Rule-based methods tend to overestimate the number of patients with ischemic stroke. Objectives We aimed to id...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2024-01-01
|
| Series: | European Journal of Medical Research |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s40001-023-01594-6 |
| _version_ | 1827388592666705920 |
|---|---|
| author | Hyunsun Lim Youngmin Park Jung Hwa Hong Ki-Bong Yoo Kwon-Duk Seo |
| author_facet | Hyunsun Lim Youngmin Park Jung Hwa Hong Ki-Bong Yoo Kwon-Duk Seo |
| author_sort | Hyunsun Lim |
| collection | DOAJ |
| description | Abstract Background Many studies have evaluated stroke using claims data; most of these studies have defined ischemic stroke using an operational definition following the rule-based method. Rule-based methods tend to overestimate the number of patients with ischemic stroke. Objectives We aimed to identify an appropriate algorithm for identifying stroke by applying machine learning (ML) techniques to analyze the claims data. Methods We obtained the data from the Korean National Health Insurance Service database, which is linked to the Ilsan Hospital database (n = 30,897). The performance of prediction models (extreme gradient boosting [XGBoost] or gated recurrent unit [GRU]) was evaluated using the area under the receiver operating characteristic curve (AUROC), the area under precision–recall curve (AUPRC), and calibration curve. Results In total, 30,897 patients were enrolled in this study, 3145 of whom (10.18%) had ischemic stroke. XGBoost, a tree-based ML technique, had the AUROC was 94.46% and AUPRC was 92.80%. GRU showed the highest accuracy (99.81%), precision (99.92%) and recall (99.69%). Conclusions We proposed recurrent neural network-based deep learning techniques to improve stroke phenotyping. This can be expected to produce rapid and more accurate results than the rule-based methods. |
| first_indexed | 2024-03-08T16:22:03Z |
| format | Article |
| id | doaj.art-149712666e1343a0baf3e827c5a63a20 |
| institution | Directory Open Access Journal |
| issn | 2047-783X |
| language | English |
| last_indexed | 2024-03-08T16:22:03Z |
| publishDate | 2024-01-01 |
| publisher | BMC |
| record_format | Article |
| series | European Journal of Medical Research |
| spelling | doaj.art-149712666e1343a0baf3e827c5a63a202024-01-07T12:17:36ZengBMCEuropean Journal of Medical Research2047-783X2024-01-012911910.1186/s40001-023-01594-6Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based studyHyunsun Lim0Youngmin Park1Jung Hwa Hong2Ki-Bong Yoo3Kwon-Duk Seo4Department of Research and Analysis, National Health Insurance Service Ilsan HospitalDepartment of Family Medicine, National Health Insurance Service Ilsan HospitalDepartment of Research and Analysis, National Health Insurance Service Ilsan HospitalDivision of Health Administration, Yonsei UniversityDepartment of Neurology, National Health Insurance Service Ilsan HospitalAbstract Background Many studies have evaluated stroke using claims data; most of these studies have defined ischemic stroke using an operational definition following the rule-based method. Rule-based methods tend to overestimate the number of patients with ischemic stroke. Objectives We aimed to identify an appropriate algorithm for identifying stroke by applying machine learning (ML) techniques to analyze the claims data. Methods We obtained the data from the Korean National Health Insurance Service database, which is linked to the Ilsan Hospital database (n = 30,897). The performance of prediction models (extreme gradient boosting [XGBoost] or gated recurrent unit [GRU]) was evaluated using the area under the receiver operating characteristic curve (AUROC), the area under precision–recall curve (AUPRC), and calibration curve. Results In total, 30,897 patients were enrolled in this study, 3145 of whom (10.18%) had ischemic stroke. XGBoost, a tree-based ML technique, had the AUROC was 94.46% and AUPRC was 92.80%. GRU showed the highest accuracy (99.81%), precision (99.92%) and recall (99.69%). Conclusions We proposed recurrent neural network-based deep learning techniques to improve stroke phenotyping. This can be expected to produce rapid and more accurate results than the rule-based methods.https://doi.org/10.1186/s40001-023-01594-6PhenotypingIschemic strokeMachine learningDeep learningInsurance claim analysis |
| spellingShingle | Hyunsun Lim Youngmin Park Jung Hwa Hong Ki-Bong Yoo Kwon-Duk Seo Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study European Journal of Medical Research Phenotyping Ischemic stroke Machine learning Deep learning Insurance claim analysis |
| title | Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study |
| title_full | Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study |
| title_fullStr | Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study |
| title_full_unstemmed | Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study |
| title_short | Use of machine learning techniques for identifying ischemic stroke instead of the rule-based methods: a nationwide population-based study |
| title_sort | use of machine learning techniques for identifying ischemic stroke instead of the rule based methods a nationwide population based study |
| topic | Phenotyping Ischemic stroke Machine learning Deep learning Insurance claim analysis |
| url | https://doi.org/10.1186/s40001-023-01594-6 |
| work_keys_str_mv | AT hyunsunlim useofmachinelearningtechniquesforidentifyingischemicstrokeinsteadoftherulebasedmethodsanationwidepopulationbasedstudy AT youngminpark useofmachinelearningtechniquesforidentifyingischemicstrokeinsteadoftherulebasedmethodsanationwidepopulationbasedstudy AT junghwahong useofmachinelearningtechniquesforidentifyingischemicstrokeinsteadoftherulebasedmethodsanationwidepopulationbasedstudy AT kibongyoo useofmachinelearningtechniquesforidentifyingischemicstrokeinsteadoftherulebasedmethodsanationwidepopulationbasedstudy AT kwondukseo useofmachinelearningtechniquesforidentifyingischemicstrokeinsteadoftherulebasedmethodsanationwidepopulationbasedstudy |