BMC Medical Research Methodology最新文献

Background: Recurrent ischemic stroke (RIS) is a significant challenge in Malaysia, affecting approximately 33% of patients. However, studies using artificial intelligence (AI) to predict this event using real-world data remain very limited. This study aimed to develop and evaluate Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and RUSBoost models for predicting recurrent ischemic stroke using real-world data from the Malaysian National Neurology Registry.

Methods: We established a retrospective study of 7,697 enrolled patients registered in the National Neurology Registry in Malaysia (2009-2016). We developed and evaluated several machine learning models, including SVM, KNN, and RUSBoost, to predict recurrent RIS. The Synthetic Minority Over-Sampling Technique (SMOTE) was applied to the training data to handle the imbalanced data. Ten-fold cross-validation was applied to assess the robustness and accuracy of the models, and performance was evaluated using criteria of accuracy, sensitivity, specificity, PPV, and area under the ROC curve (AUC).

Results: Among the evaluated machine learning models, RUSBoost demonstrated the strongest and most clinically relevant performance when assessed on validation (test) folds under stratified ten-fold cross-validation, achieving an AUROC of 0.943, sensitivity of 86.5%, and a favourable balance between sensitivity and PPV of 40.2% on the original imbalanced dataset. Although the application of SMOTE during training improved model discrimination for RUSBoost (training-fold AUROC = 0.986). The SHAP analysis showed that age, race, glucose level, hypertension, hyperlipidemia, and duration of diabetes were the most significant factors linked to an increased risk of recurrent ischemic stroke.

Conclusion: This study demonstrates that applying machine learning models on real-world clinical data is a promising tool for predicting the risk of ischemic stroke recurrence. RUSBoost emerged as the most reliable and generalisable model for clinical risk prediction, proved effective in improving prediction accuracy and identifying patients at highest risk. While SMOTE enhanced model learning during training. The findings highlight the importance of integrating AI technologies into clinical practice to support early treatment decisions and enhance preventive interventions, opening new pathways for better patient care and reducing the health burden from recurrent stroke.

This paper demonstrates the advantages of two-level non-regular fractional factorial designs for efficiently identifying influential factors in public health studies. We show how to construct smaller non-regular designs that can reliably estimate main effects and two-factor interactions in multi-factor studies, which are common in public health research. After some background information on non-regular designs and orthogonal arrays, we use real applications to support our claim. The applications include a behavioral intervention study (location/dates unreported), a drug combination trial (UCLA Micro Systems Laboratories, dates unreported), and a non-pharmaceutical Intervention study for influenza (agent-based simulation, dates unreported). We also provide resources on how to construct an appropriate non-regular design for a given problem. Our results from the three case studies show that smaller non-regular designs can identify key factors as reliably as the larger regular designs and they can require as much as 25% fewer runs, thereby reducing the study costs without sacrificing statistical precision in the inference The conclusion is that two-level regular designs are commonly used in public health studies but two-level non-regular designs can improve efficiency in screening multiple environmental, social, and behavioral factors, conserve resources and expedite discovery of evidence-based intervention effects. More specifically, two-level non-regular fractional factorial designs are more practical as they can be more flexible, cost efficient and provide just as reliable estimates as regular full or fractional factorial designs for a public health study with many interacting factors.