Key result
Deep learning of intraoperative waveforms predicts free flap complications with an AUROC up to 0.93.
Why the study?
This study explored whether high-resolution physiological signals routinely available during surgery can be leveraged by deep learning to predict adverse events after head and neck free flap reconstruction.
Does a deep learning model analyzing intraoperative physiological signals predict adverse events in patients undergoing head and neck free flap surgery?
Population
187 patients who underwent free flap surgery
Comparison
Mamba deep learning model vs conventional logistic regression and feature-based machine learning models
Design
Retrospective study
Authors
Loading...
DL models on routine intraoperative signals may predict post-flap adverse events; leaves open prospective validation before any clinical adoption.
Observational (n=187)
Does a deep learning model analyzing intraoperative physiological signals predict adverse events in patients undergoing head and neck free flap surgery?
Effect estimate: AUROC 0.86, 0.62, and 0.93
A deep learning model utilizing intraoperative physiological waveforms demonstrated feasibility and high discriminative performance for predicting flap failure and surgical complications in head and neck free flap surgery.
Kim et al. (2026) conducted an observational in Head and neck free flap surgery (n=187). Deep learning model based on the Mamba architecture using intraoperative waveforms vs. Conventional logistic regression and feature-based machine learning models was evaluated on Flap failure, return to the operating room for exploration, and other surgical complications (AUROC 0.86, 0.62, and 0.93). A deep learning model using intraoperative waveforms predicted flap failure, re-exploration, and other complications after head and neck free flap surgery with AUROCs of 0.86, 0.62, and 0.93.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: