sas-based computer-generated randomization algorithm  (SAS institute)

Journal: Healthcare

Article Title: Screening for Left Ventricular Hypertrophy Using Artificial Intelligence Algorithms Based on 12 Leads of the Electrocardiogram—Applicable in Clinical Practice?—Critical Literature Review with Meta-Analysis

doi: 10.3390/healthcare13040408

Figure Lengend Snippet: Comparison of available AI applications analyzing ECG.

Article Snippet: Salazar, 2021 [ ] , Computer-based ECG model , 458 ECG standard and non-standard parameters; 25 mm/s velocity and 10 mm/mV sensitivity , - ECG computer-based data (Philips DXL-16 algorithm) and the C5.0 ML algorithm - based on decision tree structure: - first step: T voltage I (cut-off: 0.055 mV) - second step: QRS PPK aVL or aVF (cut-off: 1.235 mV and 0.178 mV, respectively) - gradient boosting machine (XGBoost) as the core algorithm - Hyperparameter tuning via grid search , ✓ feature selection using LASSO regression ✓ focused on optimizing traditional ECG criteria ✓ 5-fold cross-validation ✓ employed SHAP values for feature importance analysis.

Techniques: Comparison, Biomarker Discovery, Diagnostic Assay

Journal: Healthcare

Article Title: Screening for Left Ventricular Hypertrophy Using Artificial Intelligence Algorithms Based on 12 Leads of the Electrocardiogram—Applicable in Clinical Practice?—Critical Literature Review with Meta-Analysis

doi: 10.3390/healthcare13040408

Figure Lengend Snippet: Excluded studies.

Article Snippet: Salazar, 2021 [ ] , Computer-based ECG model , 458 ECG standard and non-standard parameters; 25 mm/s velocity and 10 mm/mV sensitivity , - ECG computer-based data (Philips DXL-16 algorithm) and the C5.0 ML algorithm - based on decision tree structure: - first step: T voltage I (cut-off: 0.055 mV) - second step: QRS PPK aVL or aVF (cut-off: 1.235 mV and 0.178 mV, respectively) - gradient boosting machine (XGBoost) as the core algorithm - Hyperparameter tuning via grid search , ✓ feature selection using LASSO regression ✓ focused on optimizing traditional ECG criteria ✓ 5-fold cross-validation ✓ employed SHAP values for feature importance analysis.

Techniques:

Journal: Healthcare

Article Title: Screening for Left Ventricular Hypertrophy Using Artificial Intelligence Algorithms Based on 12 Leads of the Electrocardiogram—Applicable in Clinical Practice?—Critical Literature Review with Meta-Analysis

doi: 10.3390/healthcare13040408

Figure Lengend Snippet: Included studies and their characteristics.

Article Snippet: Salazar, 2021 [ ] , Computer-based ECG model , 458 ECG standard and non-standard parameters; 25 mm/s velocity and 10 mm/mV sensitivity , - ECG computer-based data (Philips DXL-16 algorithm) and the C5.0 ML algorithm - based on decision tree structure: - first step: T voltage I (cut-off: 0.055 mV) - second step: QRS PPK aVL or aVF (cut-off: 1.235 mV and 0.178 mV, respectively) - gradient boosting machine (XGBoost) as the core algorithm - Hyperparameter tuning via grid search , ✓ feature selection using LASSO regression ✓ focused on optimizing traditional ECG criteria ✓ 5-fold cross-validation ✓ employed SHAP values for feature importance analysis.

Techniques: Biomarker Discovery

Journal: Healthcare

Article Title: Screening for Left Ventricular Hypertrophy Using Artificial Intelligence Algorithms Based on 12 Leads of the Electrocardiogram—Applicable in Clinical Practice?—Critical Literature Review with Meta-Analysis

doi: 10.3390/healthcare13040408

Figure Lengend Snippet: Comparison of AI models utilized for each selected study.

Article Snippet: Salazar, 2021 [ ] , Computer-based ECG model , 458 ECG standard and non-standard parameters; 25 mm/s velocity and 10 mm/mV sensitivity , - ECG computer-based data (Philips DXL-16 algorithm) and the C5.0 ML algorithm - based on decision tree structure: - first step: T voltage I (cut-off: 0.055 mV) - second step: QRS PPK aVL or aVF (cut-off: 1.235 mV and 0.178 mV, respectively) - gradient boosting machine (XGBoost) as the core algorithm - Hyperparameter tuning via grid search , ✓ feature selection using LASSO regression ✓ focused on optimizing traditional ECG criteria ✓ 5-fold cross-validation ✓ employed SHAP values for feature importance analysis.

Techniques: Comparison, Activation Assay, Extraction, Biomarker Discovery, Variant Assay, Selection

Journal: Healthcare

Article Title: Screening for Left Ventricular Hypertrophy Using Artificial Intelligence Algorithms Based on 12 Leads of the Electrocardiogram—Applicable in Clinical Practice?—Critical Literature Review with Meta-Analysis

doi: 10.3390/healthcare13040408

Figure Lengend Snippet: Comparison of AI models.

Article Snippet: Salazar, 2021 [ ] , Computer-based ECG model , 458 ECG standard and non-standard parameters; 25 mm/s velocity and 10 mm/mV sensitivity , - ECG computer-based data (Philips DXL-16 algorithm) and the C5.0 ML algorithm - based on decision tree structure: - first step: T voltage I (cut-off: 0.055 mV) - second step: QRS PPK aVL or aVF (cut-off: 1.235 mV and 0.178 mV, respectively) - gradient boosting machine (XGBoost) as the core algorithm - Hyperparameter tuning via grid search , ✓ feature selection using LASSO regression ✓ focused on optimizing traditional ECG criteria ✓ 5-fold cross-validation ✓ employed SHAP values for feature importance analysis.

Techniques: Comparison