A Knowledge-Driven Framework for Surgical Safety Check Integration Using Speech Recognition and Speaker Verification

wen shi¹rui fan¹jian hu¹jingrong wang¹mei bian²Wei Jiang^1*

• ¹Air Force Medical University Tangdu Hospital, Xi'an, China
• ²The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an, China

Background: The WHO Surgical Safety Checklist reduces preventable errors, but real operating rooms contain overlapping speech, high ambient noise, similar vocalizations, uncertain speaker identity, and occasional omission of checklist items. ASR-only systems that ignore identity constraints and semantic–temporal structure are therefore prone to misrecognition and incorrect verification. Method: We design an integrated verification framework that couples automatic speech recognition and speaker verification (ASR+SV) with a knowledge-driven rule engine derived from the WHO checklist. Multichannel audio is processed by a Conformer-based ASR module and an ECAPA-TDNN speaker verification model, after which a rule layer enforces consistency across semantic content, speaker role, and checklist phase using an explicit ontology and conflict-resolution rules. The system generates real-time prompts in four states (“pass,” “fault,” “alarm,” “uncertain”). Performance is evaluated in high-fidelity simulated operating-room scenarios with controlled noise, speaking distances, and multi-speaker interactions, using word error rate (WER), equal error rate (EER), verification accuracy, and alarm rate. Three configurations—“ASR-only,” “ASR+SV,” and the full knowledge-driven method—are compared, and ablation experiments isolate the contribution of the rule layer. Results: Under medium-to-high noise and multi-speaker interference, the full framework reduced WER from 18.7% to 13.5% and achieved an EER of about 3.1% relative to baseline. Checklist verification accuracy reached 93.8%, while the alarm rate decreased to 2.7%. The knowledge layer corrected errors from homophones, accent drift, and role confusion by constraining “role–semantics–process” relations, maintaining robust performance at speaking distances up to 1.5 m and background noise of 60 dB. Conclusion: The proposed knowledge-driven ASR+SV framework jointly addresses semantic correctness and speaker identity while remaining interpretable and suitable for embedded deployment. It provides a technical foundation for “time-out” and “operation review” in intelligent operating rooms. Future work will include clinical pilot studies, integration with electronic medical records and multimodal OR data, and deeper analysis of privacy, accountability, and workflow acceptance.