Machine Learning Analysis of Time-Dependent Features for Predicting Adverse Events During Hemodialysis Therapy: Model Development and Validation Study

Data-driven, two-stage machine learning algorithm-based prediction scheme for assessing 1-year and 3-year mortality risk in chronic hemodialysis patients

Life expectancy is likely to be substantially reduced in patients undergoing chronic hemodialysis (CHD). However, machine learning (ML) may predict the risk factors of mortality in patients with CHD by analyzing the serum laboratory data from regular dialysis routine. This study aimed to establish the mortality prediction model of CHD patients by adopting two-stage ML algorithm-based prediction scheme, combined with importance of risk factors identified by different ML methods. This is a retrospective, observational cohort study. We included 800 patients undergoing CHD between December 2006 and December 2012 in Shin-Kong Wu Ho-Su Memorial Hospital. This study analyzed laboratory data including 44 indicators. We used five ML methods, namely, logistic regression (LGR), decision tree (DT), random forest (RF), gradient boosting (GB), and eXtreme gradient boosting (XGB), to develop a two-stage ML algorithm-based prediction scheme and evaluate the important factors that predict CHD mortality. LGR served as a bench method. Regarding the validation and testing datasets from 1- and 3-year mortality prediction model, the RF had better accuracy and area-under-curve results among the five different ML methods. The stepwise RF model, which incorporates the most important factors of CHD mortality risk based on the average rank from DT, RF, GB, and XGB, exhibited superior predictive performance compared to LGR in predicting mortality among CHD patients over both 1-year and 3-year periods. We had developed a two-stage ML algorithm-based prediction scheme by implementing the stepwise RF that demonstrated satisfactory performance in predicting mortality in patients with CHD over 1- and 3-year periods. The findings of this study can offer valuable information to nephrologists, enhancing patient-centered decision-making and increasing awareness about risky laboratory data, particularly for patients with a high short-term mortality risk.

Machine learning-based intradialytic hypotension prediction of patients undergoing hemodialysis: A multicenter retrospective study

BACKGROUND AND OBJECTIVE

Intradialytic hypotension (IDH) is closely associated with adverse clinical outcomes in HD-patients. An IDH predictor model is important for IDH risk screening and clinical decision-making. In this study, we used Machine learning (ML) to develop IDH model for risk prediction in HD patients.

METHODS

62,227 dialysis sessions were randomly partitioned into training data (70%), test data (20%), and validation data (10%). IDH-A model based on twenty-seven variables was constructed for risk prediction for the next HD treatment. IDH-B model based on ten variables from 64,870 dialysis sessions was developed for risk assessment before each HD treatment. Light Gradient Boosting Machine (LightGBM), Linear Discriminant Analysis, support vector machines, XGBoost, TabNet, and multilayer perceptron were used to develop the predictor model.

RESULTS

In IDH-A model, we identified the LightGBM method as the best-performing and interpretable model with C- statistics of 0.82 in Fall30Nadir90 definitions, which was higher than those obtained using the other models (P<0.01). In other IDH standards of Nadir90, Nadir100, Fall20, Fall30, and Fall20Nadir90, the LightGBM method had a performance with C- statistics ranged 0.77 to 0.89. As a complementary application, the LightGBM model in IDH-B model achieved C- statistics of 0.68 in Fall30Nadir90 definitions and 0.69 to 0.78 in the other five IDH standards, which were also higher than the other methods, respectively.

CONCLUSION

Use ML, we identified the LightGBM method as the good-performing and interpretable model. We identified the top variables as the high-risk factors for IDH incident in HD-patient. IDH-A and IDH-B model can usefully complement each other for risk prediction and further facilitate timely intervention through applied into different clinical setting.

Explainable Machine Learning Techniques To Predict Amiodarone-Induced Thyroid Dysfunction Risk: Multicenter, Retrospective Study With External Validation

BACKGROUND

Machine learning offers new solutions for predicting life-threatening, unpredictable amiodarone-induced thyroid dysfunction. Traditional regression approaches for adverse-effect prediction without time-series consideration of features have yielded suboptimal predictions. Machine learning algorithms with multiple data sets at different time points may generate better performance in predicting adverse effects.

OBJECTIVE

We aimed to develop and validate machine learning models for forecasting individualized amiodarone-induced thyroid dysfunction risk and to optimize a machine learning-based risk stratification scheme with a resampling method and readjustment of the clinically derived decision thresholds.

METHODS

This study developed machine learning models using multicenter, delinked electronic health records. It included patients receiving amiodarone from January 2013 to December 2017. The training set was composed of data from Taipei Medical University Hospital and Wan Fang Hospital, while data from Taipei Medical University Shuang Ho Hospital were used as the external test set. The study collected stationary features at baseline and dynamic features at the first, second, third, sixth, ninth, 12th, 15th, 18th, and 21st months after amiodarone initiation. We used 16 machine learning models, including extreme gradient boosting, adaptive boosting, k-nearest neighbor, and logistic regression models, along with an original resampling method and 3 other resampling methods, including oversampling with the borderline-synthesized minority oversampling technique, undersampling-edited nearest neighbor, and over- and undersampling hybrid methods. The model performance was compared based on accuracy; Precision, recall, F₁-score, geometric mean, area under the curve of the receiver operating characteristic curve (AUROC), and the area under the precision-recall curve (AUPRC). Feature importance was determined by the best model. The decision threshold was readjusted to identify the best cutoff value and a Kaplan-Meier survival analysis was performed.

RESULTS

The training set contained 4075 patients from Taipei Medical University Hospital and Wan Fang Hospital, of whom 583 (14.3%) developed amiodarone-induced thyroid dysfunction, while the external test set included 2422 patients from Taipei Medical University Shuang Ho Hospital, of whom 275 (11.4%) developed amiodarone-induced thyroid dysfunction. The extreme gradient boosting oversampling machine learning model demonstrated the best predictive outcomes among all 16 models. The accuracy; Precision, recall, F₁-score, G-mean, AUPRC, and AUROC were 0.923, 0.632, 0.756, 0.688, 0.845, 0.751, and 0.934, respectively. After readjusting the cutoff, the best value was 0.627, and the F₁-score reached 0.699. The best threshold was able to classify 286 of 2422 patients (11.8%) as high-risk subjects, among which 275 were true-positive patients in the testing set. A shorter treatment duration; higher levels of thyroid-stimulating hormone and high-density lipoprotein cholesterol; and lower levels of free thyroxin, alkaline phosphatase, and low-density lipoprotein were the most important features.

CONCLUSIONS

Machine learning models combined with resampling methods can predict amiodarone-induced thyroid dysfunction and serve as a support tool for individualized risk prediction and clinical decision support.

Early prediction of hemodialysis complications employing ensemble techniques

Background and objectives

Hemodialysis complications remain a critical threat among dialysis patients. They result in sudden termination of the session which impacts the efficiency of dialysis. As intra-dialytic complications are the result of the interplay of multiple factors, artificial intelligence can aid in their early prediction. This research aims to compare different machine learning tools for the early prediction of the most frequent hemodialysis complications with high performance, using the fewest predictors for easier practical implementation.

Methods

Fifty different variables were recorded during 6000 hemodialysis sessions performed in a regional dialysis unit in Egypt. The filter technique was used to extract the most relevant features. Then, five individual classifiers and three ensemble approaches were implemented to predict the occurrence of intra-dialytic complications. Different subsets of 25, 12 and 6 from the 50 collected features were tested.

Results

Random forest yielded the highest accuracy of 98% with the least training time using 12 features in a balanced dataset, while the gradient boosting allowed obtaining the highest F1-score of 94%, 92%, and 78% in the prediction of hypotension, hypertension, and dyspnea, respectively, in imbalanced datasets.

Conclusion

Applying different machine learning algorithms to big datasets can improve accuracy, reduce training time and model complexity allowing simple implementation in clinical practice. Our models can help nephrologists predict and possibly prevent dialysis complications.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12938-022-01044-0.