Search papers, labs, and topics across Lattice.
This study evaluates the efficacy of risk stratification for pulmonary embolism (PE) using only medical records and biomarkers derived from computed tomography pulmonary angiography (CTPA), in the absence of commonly used blood tests. By benchmarking various methods, including the integration of vascular biomarkers and graph neural networks (GNNs) with traditional tabular models, the authors reveal that medical records and cardiac biomarkers are the most significant predictors of PE risk. Surprisingly, the addition of vascular graphs does not enhance stratification accuracy, challenging the assumption that vascular information is crucial for PE risk assessment.
Medical records and cardiac biomarkers alone outperform vascular graphs in predicting pulmonary embolism risk, raising questions about the value of complex vascular representations.
Risk stratification for pulmonary embolism (PE) is critical for clinical decision-making. Stratification guidelines are based on patient medical records, parameters measured from computed tomography pulmonary angiography (CTPA), and blood tests. However, blood tests are often missing in routine practice. This work studies whether state-of-the-art models can accurately classify risk stratification from only medical records and biomarkers extracted from CTPA images. We benchmark different approaches to combine medical records and cardiac biomarkers with rich pulmonary vascular information; we add vascular biomarkers to tabular models and apply graph neural networks (GNNs) on the vascular tree's intrinsic graph representation. We use a private dataset (n=353) with uniquely complete data for PE risk stratification. Our results show that, among global features, medical records and cardiac biomarkers are the most significant predictors, while vascular biomarkers do not further improve stratification. Even more surprising, even GNNs on vascular graphs fail to outperform strong tabular baseline on global features. We consider hypotheses, on both models and data, that could explain this suboptimal performance. Our investigation suggests that, counter-intuitively, vascular graphs might hold no discriminative information for PE risk stratification. Code is available from https://github.com/creatis-myriad/GENESIS.
