A Novel Mental Workload-driven Adaptive Training Framework in Robotic Surgical Skill Acquisition

Yang, Jing; Yu, Denny · 2026 · Crossref

DOI: 10.21203/rs.3.rs-8872917/v1

archive: archived pipeline: cataloged

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenge of limited access to structured robotic-assisted surgery (RAS) training, which often forces trainees into self-directed learning that lacks personalized feedback and may lead to inefficient skill acquisition. The authors propose a Mental Workload-driven Adaptive Training Framework (MATF) that uses real-time physiological monitoring to adjust task difficulty, keeping trainees within an optimal cognitive challenge zone. The research aims to determine if this workload-aware, closed-loop approach improves surgical skill acquisition compared to traditional self-directed practice. The study involved two phases. First, a multimodal mental workload (MWL) model was developed using data from 15 participants performing established cognitive tasks (e.g., N-back, Stroop). The model integrated EEG (theta and alpha band powers) and eye-tracker features (pupil diameter, fixation count/duration). Six machine learning models were trained using leave-one-subject-out cross-validation, comparing survey-based (NASA-TLX) and domain-knowledge-based (dual-frequency) labeling techniques. Second, the effectiveness of MATF was evaluated in a between-subject experiment with 20 novice participants using the da Vinci Research Kit. Participants were randomly assigned to either the MATF group (n=10) or a self-directed control group (n=10). The MATF system used a 15-second sliding window to stream physiological data, calculating an MWL score relative to expert benchmarks to dynamically select the next training task from a set of four fundamental RAS exercises. Performance was assessed via a complex anastomosis task before and after a 40-minute training session. Results indicated that the multimodal MWL model using domain-knowledge labeling achieved the highest classification accuracy of 83.8% (F1-score 0.82), outperforming single-modality or survey-based approaches. In the training trial, the MATF group demonstrated significantly greater improvements in anastomosis task performance, including reduced times for needle positioning, suture looping, and knot tying, and a higher number of successful knots compared to the self-directed group. Physiologically, the MATF group exhibited lower frontal theta power and higher parietal alpha power post-training, indicating neural efficiency and reduced cognitive demand. Additionally, the MATF group showed decreased fixation counts and durations, suggesting improved visual efficiency, and reported significantly lower NASA-TLX scores, confirming reduced perceived workload. The findings suggest that integrating objective, real-time mental workload monitoring into surgical training curricula can enhance learning efficiency and safety. By preventing cognitive overload and targeting specific skill gaps, MATF offers a scalable solution for personalized RAS education, particularly in settings with limited mentorship. However, the study’s reliance on novice participants and group-averaged benchmarks highlights the need for longitudinal validation in clinical residency programs to confirm generalizability to experienced surgeons.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.8-27b-gittensor on 2026-08-23; verification: pending re-verification.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• workload measurement

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: self report data