Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model
Abstract Background Hyperkalemia is a common complication of chronic kidney disease (CKD). Hyperkalemia is associated with mortality, CKD progression, hospitalization, and high healthcare costs in patients with CKD. We developed a machine learning model to predict hyperkalemia in patients with advan...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2023-06-01
|
| Series: | BMC Nephrology |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s12882-023-03227-w |
| _version_ | 1797801486084734976 |
|---|---|
| author | Hsin-Hsiung Chang Jung-Hsien Chiang Chun-Chieh Tsai Ping-Fang Chiu |
| author_facet | Hsin-Hsiung Chang Jung-Hsien Chiang Chun-Chieh Tsai Ping-Fang Chiu |
| author_sort | Hsin-Hsiung Chang |
| collection | DOAJ |
| description | Abstract Background Hyperkalemia is a common complication of chronic kidney disease (CKD). Hyperkalemia is associated with mortality, CKD progression, hospitalization, and high healthcare costs in patients with CKD. We developed a machine learning model to predict hyperkalemia in patients with advanced CKD at an outpatient clinic. Methods This retrospective study included 1,965 advanced CKD patients between January 1, 2010, and December 31, 2020 in Taiwan. We randomly divided all patients into the training (75%) and testing (25%) datasets. The primary outcome was to predict hyperkalemia (K+ > 5.5 mEq/L) in the next clinic vist. Two nephrologists were enrolled in a human-machine competition. The area under the receiver operating characteristic curves (AUCs), sensitivity, specificity, and accuracy were used to evaluate the performance of XGBoost and conventional logistic regression models with that of these physicians. Results In a human-machine competition of hyperkalemia prediction, the AUC, PPV, and accuracy of the XGBoost model were 0.867 (95% confidence interval: 0.840–0.894), 0.700, and 0.933, which was significantly better than that of our clinicians. There were four variables that were chosen as high-ranking variables in XGBoost and logistic regression models, including hemoglobin, the serum potassium level in the previous visit, angiotensin receptor blocker use, and calcium polystyrene sulfonate use. Conclusions The XGBoost model provided better predictive performance for hyperkalemia than physicians at the outpatient clinic. |
| first_indexed | 2024-03-13T04:51:12Z |
| format | Article |
| id | doaj.art-1aa849e0546848f786f275ca75e111c4 |
| institution | Directory Open Access Journal |
| issn | 1471-2369 |
| language | English |
| last_indexed | 2024-03-13T04:51:12Z |
| publishDate | 2023-06-01 |
| publisher | BMC |
| record_format | Article |
| series | BMC Nephrology |
| spelling | doaj.art-1aa849e0546848f786f275ca75e111c42023-06-18T11:08:57ZengBMCBMC Nephrology1471-23692023-06-012411810.1186/s12882-023-03227-wPredicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost modelHsin-Hsiung Chang0Jung-Hsien Chiang1Chun-Chieh Tsai2Ping-Fang Chiu3Division of Nephrology, Department of Internal Medicine, Antai Medical Care Corporation Antai Tian-Sheng Memorial HospitalDepartment of Computer Science and Information Engineering, National Cheng Kung UniversityDivision of Nephrology, Department of Internal Medicine, Changhua Christian HospitalDivision of Nephrology, Department of Internal Medicine, Changhua Christian HospitalAbstract Background Hyperkalemia is a common complication of chronic kidney disease (CKD). Hyperkalemia is associated with mortality, CKD progression, hospitalization, and high healthcare costs in patients with CKD. We developed a machine learning model to predict hyperkalemia in patients with advanced CKD at an outpatient clinic. Methods This retrospective study included 1,965 advanced CKD patients between January 1, 2010, and December 31, 2020 in Taiwan. We randomly divided all patients into the training (75%) and testing (25%) datasets. The primary outcome was to predict hyperkalemia (K+ > 5.5 mEq/L) in the next clinic vist. Two nephrologists were enrolled in a human-machine competition. The area under the receiver operating characteristic curves (AUCs), sensitivity, specificity, and accuracy were used to evaluate the performance of XGBoost and conventional logistic regression models with that of these physicians. Results In a human-machine competition of hyperkalemia prediction, the AUC, PPV, and accuracy of the XGBoost model were 0.867 (95% confidence interval: 0.840–0.894), 0.700, and 0.933, which was significantly better than that of our clinicians. There were four variables that were chosen as high-ranking variables in XGBoost and logistic regression models, including hemoglobin, the serum potassium level in the previous visit, angiotensin receptor blocker use, and calcium polystyrene sulfonate use. Conclusions The XGBoost model provided better predictive performance for hyperkalemia than physicians at the outpatient clinic.https://doi.org/10.1186/s12882-023-03227-wMachine learningHyperkalemiaChronic kidney disease |
| spellingShingle | Hsin-Hsiung Chang Jung-Hsien Chiang Chun-Chieh Tsai Ping-Fang Chiu Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model BMC Nephrology Machine learning Hyperkalemia Chronic kidney disease |
| title | Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model |
| title_full | Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model |
| title_fullStr | Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model |
| title_full_unstemmed | Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model |
| title_short | Predicting hyperkalemia in patients with advanced chronic kidney disease using the XGBoost model |
| title_sort | predicting hyperkalemia in patients with advanced chronic kidney disease using the xgboost model |
| topic | Machine learning Hyperkalemia Chronic kidney disease |
| url | https://doi.org/10.1186/s12882-023-03227-w |
| work_keys_str_mv | AT hsinhsiungchang predictinghyperkalemiainpatientswithadvancedchronickidneydiseaseusingthexgboostmodel AT junghsienchiang predictinghyperkalemiainpatientswithadvancedchronickidneydiseaseusingthexgboostmodel AT chunchiehtsai predictinghyperkalemiainpatientswithadvancedchronickidneydiseaseusingthexgboostmodel AT pingfangchiu predictinghyperkalemiainpatientswithadvancedchronickidneydiseaseusingthexgboostmodel |