Answers are grounded in this paper and cite their sources.
Demonstrates AI-enhanced risk estimation for VTE in post-treatment cervical cancer patients, suggesting improved clinical decision-making.
Venous thromboembolism (VTE) is a life-threatening complication among cervical cancer patients, yet current clinical tools rely on single-time-point laboratory values and inadequately capture the dynamic biological processes preceding thrombosis. While existing artificial intelligence (AI) methods, especially time-series deep learning models, offer high accuracy, their black-box nature and substantial computational demands make them unsuitable for the rapid and justifiable requirements of clinical practice. Using longitudinal clinical data from 1427 treated cervical cancer patients, we developed TempVTE, a lightweight and interpretable temporal modeling framework designed to extract clinically meaningful dynamics from blood test biomarkers for VTE risk estimation. Guided by feature importance from a deep learning model pre-trained on large-scale time-series data, we identify the features that contribute most significantly to VTE risk estimation. TempVTE summarizes each biomarker's trajectory using temporal descriptors—including baseline value, terminal value, and within-patient correlation—preserving dynamic information while enabling rapid inference. Experiments show that TempVTE outperforms all baseline models, achieving an AUC of 0.924 and an accuracy of 0.858. Beyond risk estimation, we found that the temporal dynamics of coagulation features are more informative than static values. Moreover, apart from well-known markers such as D-dimer, immune-related features can serve as significant indicators for the development of VTE in cervical cancer patients after treatment. The results introduce extra interpretability to the model and provide clues about VTE dynamics. This study provides a new approach for lightweight dynamic VTE risk estimation and is expected to improve the prevention, treatment, and prognosis of VTE in cervical cancer patients.
No takes yet. Share an insight, caveat, or question.
Xu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: