Machine learning-based decision support model for selecting intra-arterial therapies for unresectable hepatocellular carcinoma: A national real-world evidence-based study.

IMPORTANCE: Intra-arterial therapies(IATs) are promising options for unresectable hepatocellular carcinoma(HCC). Stratifying the prognostic risk before administering IAT is important for clinical decision-making and for designing future clinical trials. OBJECTIVE: To develop and validate a machine learning(ML)-based decision support model(MLDSM) for recommending IAT modalities for unresectable HCC. DESIGN, SETTING, AND PARTICIPANTS: Between October 2014 and October 2022, a total of 2,959 patients with HCC who underwent initial IATs were enroled retrospectively from 13 tertiary hospitals. These patients were divided into the training cohort (n = 1700), validation cohort (n = 428), and test cohort (n = 200). MAIN OUTCOMES AND MEASURES: Thirty-two clinical variables were input, and five supervised ML algorithms, including eXtreme Gradient Boosting (XGBoost), Categorical Gradient Boosting (CatBoost), Gradient Boosting Decision Tree (GBDT), Light Gradient Boosting Machine (LGBM) and Random Forest (RF), were compared using the areas under the receiver operating characteristic curve (AUC) with the DeLong test. RESULTS: A total of 1856 patients were assigned to the IAT alone Group(I-A), and 1103 patients were assigned to the IAT combination Group(I-C). The 12-month death rates were 31.9% (352/1103) in the I-A group and 50.4% (936/1856) in the I-C group. For the test cohort, in the I-C group, the CatBoost model achieved the best discrimination when 30 variables were input, with an AUC of 0.776 (95% confidence intervals [CI], 0.833-0.868). In the I-A group, the LGBM model achieved the best discrimination when 24 variables were input, with an AUC of 0.776 (95% CI, 0.833-0.868). According to the decision trees, BCLC grade, local therapy, and diameter as top three variables were used to guide clinical decisions between IAT modalities. CONCLUSIONS AND RELEVANCE: The MLDSM can accurately stratify prognostic risk for HCC patients who received IATs, thus helping physicians to make decisions about IAT and providing guidance for surveillance strategies in clinical practice.

Preoperative prediction for pathological grade of hepatocellular carcinoma via machine learning-based radiomics.

OBJECTIVE: To investigate the efficacy of contrast-enhanced computed tomography (CECT)-based radiomics signatures for preoperative prediction of pathological grades of hepatocellular carcinoma (HCC) via machine learning. METHODS: In this single-center retrospective study, data collected from 297 consecutive subjects with HCC were allocated to training dataset (n = 237) and test dataset (n = 60). Manual segmentation of lesion sites was performed with ITK-SNAP, the radiomics features were extracted by the Pyradiomics, and radiomics signatures were synthesized using recursive feature elimination (RFE) method. The prediction models for pathological grading of HCC were established by using eXtreme Gradient Boosting (XGBoost). The performance of the models was evaluated using the AUC along with 95% confidence intervals (CIs) and standard deviation, sensitivity, specificity, and accuracy. RESULTS: The radiomics signatures were found highly efficient for machine learning to differentiate high-grade HCC from low-grade HCC. For the clinical factors, when they were merely applied to train a machine learning model, the model achieved an AUC of 0.6698, along with 95% CI and standard deviation of 0.5307-0.8089 and 0.0710, respectively (sensitivity, 0.6522; specificity, 0.4595; accuracy, 0.5333). Meanwhile, when the radiomics signatures were applied in association with clinical factors to train a machine learning model, the performance of the model remarkably increased with AUC of 0.8014, along with 95% CI and standard deviation of 0.6899-0.9129 and 0.0569, respectively (sensitivity, 0.6522; specificity, 0.7297; accuracy, 0.7000). CONCLUSIONS: The radiomics signatures could non-invasively explore the underlying association between CECT images and pathological grades of HCC. KEY POINTS: • The radiomics signatures may non-invasively explore the underlying association between CECT images and pathological grades of HCC via machine learning. • The radiomics signatures of CECT images may enhance the prediction performance of pathological grading of HCC, and further validation is required. • The features extracted from arterial phase CECT images may be more reliable than venous phase CECT images for predicting pathological grades of HCC.

A Novel Machine Learning Model for Predicting Stroke-Associated Pneumonia After Spontaneous Intracerebral Hemorrhage.

BACKGROUND: Pneumonia is one of the most common complications after spontaneous intracerebral hemorrhage (sICH), i.e., stroke-associated pneumonia (SAP). Timely identification of targeted patients is beneficial to reduce poor prognosis. So far, there is no consensus on SAP prediction, and application of existing predictors is limited. The aim of this study was to develop a machine learning model to predict SAP after sICH. METHODS: We retrospectively reviewed 748 patients diagnosed with sICH and collected data from 4 dimensions-demographic features, clinical features, medical history, and laboratory tests. Five machine learning algorithms-logistic regression, gradient boosting decision tree, random forest, extreme gradient boosting, and category boosting-were used to build and validate the predictive model. We also applied recursive feature elimination with cross-validation to obtain the best feature combination for each model. Predictive performance was evaluated by area under the receiver operating characteristic curve. RESULTS: SAP was diagnosed in 237 patients. The model developed by category boosting yielded the most satisfactory outcomes overall with area under the receiver operating characteristic curves in the training set and test set of 0.8307 and 0.8178, respectively. CONCLUSIONS: The incidence of SAP after sICH in our center was 31.68%. Machine learning could potentially provide assistance in the prediction of SAP after sICH.