What question did this study set out to answer?

Develop and evaluate CRTRE, an interpretable framework using target trial emulation principles to uncover causal association rules and decorrelate features for biomedical predictions.

December 15, 2024

CRTRE: Causal Rule Generation with Target Trial Emulation Framework

View Full Paper

Ask AI

Bookmark

Authors

JWJunda WangShanghai Jiao Tong University WLWeijian LiGuangdong Polytechnic Normal University HWHan WangHangzhou Dianzi University

Discussion

Member takes

Implication

Computational study demonstrates improved disease onset and diagnostic prediction across clinical datasets, highlighting the power of target trial emulation for interpretable AI.

Key Points

Develop and evaluate CRTRE, an interpretable framework using target trial emulation principles to uncover causal association rules and decorrelate features for biomedical predictions.
Integrated randomized clinical trial design principles to estimate the causal impact of association rules in nonlinear data environments.
Evaluated the method across six healthcare datasets, including synthetic benchmarks, three disease cohorts (Esophageal Cancer, Heart Disease, and Cauda Equina Syndrome), and MIMIC-III/IV.
Benchmarked predictive accuracy, beta error, and AUC Macro against baseline and state-of-the-art models such as DWR, SVM, KEPT, and MSMN.
Achieved a lower β error of 0.907 on synthetic benchmark data, outperforming DWR (1.024) and SVM (1.141).
Attained superior classification accuracies of 0.789 for Esophageal Cancer, 0.920 for Heart Disease, and 0.300 for Cauda Equina Syndrome onset prediction.
Yielded AUC Macro scores of 92.8 on MIMIC-III and 96.7 on MIMIC-IV for ICD code prediction, outperforming state-of-the-art KEPT and MSMN architectures.

Cite This Study

Wang et al. (2024) studied this question.

synapsesocial.com/papers/6a10a65f2badbc352a0073af https://doi.org/10.1109/bigdata62323.2024.10825951

View Full Paper

Ask AI

Bookmark

Implication

Computational study demonstrates improved disease onset and diagnostic prediction across clinical datasets, highlighting the power of target trial emulation for interpretable AI.

CRTRE: Causal Rule Generation with Target Trial Emulation Framework

Discussion

Member takes

Implication

Key Points

Cite This Study

CRTRE: Causal Rule Generation with Target Trial Emulation Framework

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider