Surgical Phase Recognition via Sufficient Statistics and RNN
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current surgical care systems face challenges in accurately identifying surgical phases due to the temporal nature of surgical procedures and the limited availability of experienced surgeons, particularly in rural areas, leading to disparities in access to quality care.
Innovation Solution
A system utilizing a sufficient statistics model combined with a recurrent neural network and convolutional neural network to analyze sensor data from surgical procedures, including video frames, to automatically identify the current surgical phase and provide predictive assistance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If surgical phase recognition relies on experienced surgeons' expertise, then accuracy in identifying surgical phases improves, but availability of such expertise deteriorates in rural and minority areas
Solution Approach 1:
The patent creates a digital copy of experienced surgeons' expertise by training an AI model on surgical video data and expert annotations. This digital replica can be deployed anywhere with internet connectivity, making expert-level surgical phase recognition available in rural and minority areas without requiring physical presence of experienced surgeons
Solution Approach 2:
The patent introduces an AI-based surgical phase recognition system as an intermediary between raw surgical video data and clinical decision-making. This intermediary processes video feeds in real-time, providing structured phase information that assists surgeons regardless of their experience level or location
2Reliability
If cognitive task analysis is used to codify surgeon knowledge into checklists, then decision-making support improves, but time consumption and scalability deteriorate
Solution Approach 1:
The patent replaces manual cognitive task analysis and checklist creation with an automated AI system that continuously processes surgical video data. The system automatically generates and updates phase recognition models without requiring manual codification of knowledge, eliminating the time-consuming nature of traditional CTA while maintaining reliability
Solution Approach 2:
The surgical phase recognition system is designed to be self-learning and self-updating through continuous exposure to surgical video data. The AI model automatically improves its performance over time without requiring external intervention for knowledge updates, enabling scalable deployment across multiple institutions
3Reliability
If surgical care is regionalized to high-volume hospitals, then care quality improves, but access to surgery deteriorates for rural and minority patients
Solution Approach 1:
The patent creates a digital copy of expert surgical knowledge that can be distributed to any hospital regardless of volume. This allows rural and minority-serving hospitals to maintain high care quality through AI assistance without requiring patients to travel to urban academic centers
Data Source
AI summary
Systems and methods are provided for identifying a current phase of a surgical procedure. Sensor data representing a time period is received and a plurality of numerical features representing the time period are generated from the sensor data. A statistical parameter representing a plurality of stored values from a memory is generated at a sufficient statistics model. An output, representing a surgical phase associated with the time period is provided at a recurrent neural network from a set of inputs that includes the plurality of numerical features and the statistical parameter.