Figure 1

Workflow architecture of SmartForceps platform from data recording to modeling and visualization. Forces of tool-tissue interaction along with de-identified case information were uploaded to a HIPAA-compliant data storage and analytics platform. Force data were manually segmented and labeled by listening to the surgeon’s voice recordings, where surgeon names, surgical tasks, and important incidents were narrated. The AI modeling architecture included Auto Data Preprocessing (e.g., Data Balancing, Outlier Removal, Data Transformation, etc.), Feature Engineering, Data Modeling (T-U-Net for force profile segmentation (T-U-Net: Time-series-U-Net); XGBoost, LSTM and FTFIT (Force Time-series Feature-based InceptionTime) for pattern recognition), and Modeling Optimization and Performance Evaluation, which were integrated into the cloud platform to generate performance evaluation reports to the surgical team. A detailed description of selected processes in the figure has been described in the Supplementary Materials. Visualization was created in Microsoft PowerPoint version 16.49 with the icons obtained from a Google search: e.g., https://www.iconfinder.com.