tetano
Editor, Senior Moderator
Sci Rep
. 2024 Nov 6;14(1):26979.
doi: 10.1038/s41598-024-78498-w. Predictive modeling of COVID-19 mortality risk in chronic kidney disease patients using multiple machine learning algorithms
Lin Luo[SUP] 1 [/SUP], Peng Gao[SUP] 1 [/SUP], Chunhui Yang[SUP] 2 [/SUP], Sha Yu[SUP] 3 [/SUP]
Affiliations
The coronavirus disease 2019 (COVID-19) has a significant impact on the global population, particularly on individuals with chronic kidney disease (CKD). COVID-19 patients with CKD will face a considerably higher risk of mortality than the general population. This study developed a predictive model for assessing mortality in COVID-19-affected CKD patients, providing personalized risk prediction to optimize clinical management and reduce mortality rates. We developed machine learning algorithms to analyze 219 patients' clinical laboratory test data retrospectively. The performance of each model was assessed using a calibration curve, decision curve analysis, and receiver operating characteristic (ROC) curve. It was found that the LightGBM model showed the most satisfied performance, with an area under the ROC curve of 0.833, sensitivity of 0.952, and specificity of 0.714. Prealbumin, neutrophil percent, respiratory index in arterial blood, half-saturated pressure of oxygen, carbon dioxide in serum, glucose, neutrophil count, and uric acid were the top 8 significant variables in the prediction model. Validation by 46 patients demonstrated acceptable accuracy. This model can serve as a powerful tool for screening CKD patients at high risk of COVID-19-related mortality and providing decision support for clinical staff, enabling efficient allocation of resources, and facilitating timely and targeted management for those who need the relevant interference urgently.
Keywords: COVID-19; Chronic kidney disease; LightGBM (light gradient-boosting machine); Machine learning algorithm; Prediction model.
. 2024 Nov 6;14(1):26979.
doi: 10.1038/s41598-024-78498-w. Predictive modeling of COVID-19 mortality risk in chronic kidney disease patients using multiple machine learning algorithms
Lin Luo[SUP] 1 [/SUP], Peng Gao[SUP] 1 [/SUP], Chunhui Yang[SUP] 2 [/SUP], Sha Yu[SUP] 3 [/SUP]
Affiliations
- PMID: 39506019
- DOI: 10.1038/s41598-024-78498-w
The coronavirus disease 2019 (COVID-19) has a significant impact on the global population, particularly on individuals with chronic kidney disease (CKD). COVID-19 patients with CKD will face a considerably higher risk of mortality than the general population. This study developed a predictive model for assessing mortality in COVID-19-affected CKD patients, providing personalized risk prediction to optimize clinical management and reduce mortality rates. We developed machine learning algorithms to analyze 219 patients' clinical laboratory test data retrospectively. The performance of each model was assessed using a calibration curve, decision curve analysis, and receiver operating characteristic (ROC) curve. It was found that the LightGBM model showed the most satisfied performance, with an area under the ROC curve of 0.833, sensitivity of 0.952, and specificity of 0.714. Prealbumin, neutrophil percent, respiratory index in arterial blood, half-saturated pressure of oxygen, carbon dioxide in serum, glucose, neutrophil count, and uric acid were the top 8 significant variables in the prediction model. Validation by 46 patients demonstrated acceptable accuracy. This model can serve as a powerful tool for screening CKD patients at high risk of COVID-19-related mortality and providing decision support for clinical staff, enabling efficient allocation of resources, and facilitating timely and targeted management for those who need the relevant interference urgently.
Keywords: COVID-19; Chronic kidney disease; LightGBM (light gradient-boosting machine); Machine learning algorithm; Prediction model.