Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models
<b>Background:</b> Current guideline-based implantable cardioverter-defibrillator (ICD) implants fail to meet the demands for precision medicine. Machine learning (ML) designed for survival analysis might facilitate personalized risk stratification. We aimed to develop explainable ML mod...
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-09-01
|
| Series: | Journal of Cardiovascular Development and Disease |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2308-3425/9/9/310 |
| _version_ | 1797486725287641088 |
|---|---|
| author | Yu Deng Sijing Cheng Hao Huang Xi Liu Yu Yu Min Gu Chi Cai Xuhua Chen Hongxia Niu Wei Hua |
| author_facet | Yu Deng Sijing Cheng Hao Huang Xi Liu Yu Yu Min Gu Chi Cai Xuhua Chen Hongxia Niu Wei Hua |
| author_sort | Yu Deng |
| collection | DOAJ |
| description | <b>Background:</b> Current guideline-based implantable cardioverter-defibrillator (ICD) implants fail to meet the demands for precision medicine. Machine learning (ML) designed for survival analysis might facilitate personalized risk stratification. We aimed to develop explainable ML models predicting mortality and the first appropriate shock and compare these to standard Cox proportional hazards (CPH) regression in ICD recipients. <b>Methods and Results:</b> Forty-five routine clinical variables were collected. Four fine-tuned ML approaches (elastic net Cox regression, random survival forests, survival support vector machine, and XGBoost) were applied and compared with the CPH model on the test set using Harrell’s C-index. Of 887 adult patients enrolled, 199 patients died (5.0 per 100 person-years) and 265 first appropriate shocks occurred (12.4 per 100 person-years) during the follow-up. Patients were randomly split into training (75%) and test (25%) sets. Among ML models predicting death, XGBoost achieved the highest accuracy and outperformed the CPH model (C-index: 0.794 vs. 0.760, <i>p</i> < 0.001). For appropriate shock, survival support vector machine showed the highest accuracy, although not statistically different from the CPH model (0.621 vs. 0.611, <i>p</i> = 0.243). The feature contribution of ML models assessed by SHAP values at individual and overall levels was in accordance with established knowledge. Accordingly, a bi-dimensional risk matrix integrating death and shock risk was built. This risk stratification framework further classified patients with different likelihoods of benefiting from ICD implant. <b>Conclusions:</b> Explainable ML models offer a promising tool to identify different risk scenarios in ICD-eligible patients and aid clinical decision making. Further evaluation is needed. |
| first_indexed | 2024-03-09T23:38:22Z |
| format | Article |
| id | doaj.art-c9741485f2894559a44d827af26d58f4 |
| institution | Directory Open Access Journal |
| issn | 2308-3425 |
| language | English |
| last_indexed | 2024-03-09T23:38:22Z |
| publishDate | 2022-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Cardiovascular Development and Disease |
| spelling | doaj.art-c9741485f2894559a44d827af26d58f42023-11-23T16:57:30ZengMDPI AGJournal of Cardiovascular Development and Disease2308-34252022-09-019931010.3390/jcdd9090310Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning ModelsYu Deng0Sijing Cheng1Hao Huang2Xi Liu3Yu Yu4Min Gu5Chi Cai6Xuhua Chen7Hongxia Niu8Wei Hua9The Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, ChinaThe Cardiac Arrhythmia Center, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, No.167 North Lishi Road, Beijing 100037, China<b>Background:</b> Current guideline-based implantable cardioverter-defibrillator (ICD) implants fail to meet the demands for precision medicine. Machine learning (ML) designed for survival analysis might facilitate personalized risk stratification. We aimed to develop explainable ML models predicting mortality and the first appropriate shock and compare these to standard Cox proportional hazards (CPH) regression in ICD recipients. <b>Methods and Results:</b> Forty-five routine clinical variables were collected. Four fine-tuned ML approaches (elastic net Cox regression, random survival forests, survival support vector machine, and XGBoost) were applied and compared with the CPH model on the test set using Harrell’s C-index. Of 887 adult patients enrolled, 199 patients died (5.0 per 100 person-years) and 265 first appropriate shocks occurred (12.4 per 100 person-years) during the follow-up. Patients were randomly split into training (75%) and test (25%) sets. Among ML models predicting death, XGBoost achieved the highest accuracy and outperformed the CPH model (C-index: 0.794 vs. 0.760, <i>p</i> < 0.001). For appropriate shock, survival support vector machine showed the highest accuracy, although not statistically different from the CPH model (0.621 vs. 0.611, <i>p</i> = 0.243). The feature contribution of ML models assessed by SHAP values at individual and overall levels was in accordance with established knowledge. Accordingly, a bi-dimensional risk matrix integrating death and shock risk was built. This risk stratification framework further classified patients with different likelihoods of benefiting from ICD implant. <b>Conclusions:</b> Explainable ML models offer a promising tool to identify different risk scenarios in ICD-eligible patients and aid clinical decision making. Further evaluation is needed.https://www.mdpi.com/2308-3425/9/9/310implantable cardioverter-defibrillatorsmachine learningmortalityfirst appropriate shockShapley Additive exPlanations valuespersonalized risk stratification |
| spellingShingle | Yu Deng Sijing Cheng Hao Huang Xi Liu Yu Yu Min Gu Chi Cai Xuhua Chen Hongxia Niu Wei Hua Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models Journal of Cardiovascular Development and Disease implantable cardioverter-defibrillators machine learning mortality first appropriate shock Shapley Additive exPlanations values personalized risk stratification |
| title | Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models |
| title_full | Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models |
| title_fullStr | Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models |
| title_full_unstemmed | Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models |
| title_short | Toward Better Risk Stratification for Implantable Cardioverter-Defibrillator Recipients: Implications of Explainable Machine Learning Models |
| title_sort | toward better risk stratification for implantable cardioverter defibrillator recipients implications of explainable machine learning models |
| topic | implantable cardioverter-defibrillators machine learning mortality first appropriate shock Shapley Additive exPlanations values personalized risk stratification |
| url | https://www.mdpi.com/2308-3425/9/9/310 |
| work_keys_str_mv | AT yudeng towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT sijingcheng towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT haohuang towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT xiliu towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT yuyu towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT mingu towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT chicai towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT xuhuachen towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT hongxianiu towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels AT weihua towardbetterriskstratificationforimplantablecardioverterdefibrillatorrecipientsimplicationsofexplainablemachinelearningmodels |