HAM Study Research

Four comprehensive studies validating the Ham-based Assessment Model

The HAM Study represents a breakthrough in objective, reproducible assessment of endoscopic tip control. Through four interconnected studies, we've validated a low-cost, AI-assisted platform that enables standardized skill evaluation for endoscopists at any stage of their career.

System Overview

Core Components

  • Training Box: Reusable setup (~€200, one-time cost)
  • Equipment: Standard endoscope and processor screen
  • Medium: Disposable slice of cooked ham
  • Stencil: Free 3D-printed template with dye shapes

Assessment Protocol

Endoscopists use a snare through the endoscope channel to trace blue-dyed lines with electrosurgical precision. AI algorithms analyze performance at electrosurgery points, evaluating tip stability and generating quantitative metrics.

HAM Assessment Detailed View - Shows the snare tracing process and evaluation metrics

Detailed view of HAM assessment showing snare positioning and tracing patterns

Assessment Shapes

The assessment includes one practice shape and four test shapes of increasing circumference. While the shapes vary in size, they maintain consistent difficulty levels to ensure reliable skill assessment.

P

Practice Shape

1

Shape 1

2

Shape 2

3

Shape 3

4

Shape 4

Summary of the Four Studies

1

Study 1 – Stratification by Experience

Debels et al., Frontline Gastroenterology, 2025

This initial study demonstrated that the tip control score effectively stratifies endoscopists by experience level and role. The system successfully differentiated between:

  • Trainees
  • Standard non-interventional endoscopists
  • Interventional endoscopists

Key Findings:

  • • Interventional endoscopists outperformed both trainees and non-interventional consultants
  • • Years of experience alone did not predict superior performance
  • • Tip control stratifies endoscopists by actual skill, not seniority
Study 1: Endoscopist Performance by Role - Accuracy vs Speed scatter plot showing clear stratification between trainees, consultants, and interventional consultants

Performance stratification by endoscopist role (Study 1)Hover to zoom

2

Study 2 – Improvement Through Repetition

In a structured design with three blinded video raters, experienced endoscopists completed a comprehensive training protocol:

Pre-test

Shapes 1–4

Training

10 additional shapes

Post-test

Shapes 1–4 again

Key Findings:

  • • Significant improvement in both accuracy and speed
  • • Lower baseline performance predicted greater improvement
  • • Strong inter-rater reliability (ICC > 0.84)
  • • Demonstrates utility in monitoring skill acquisition

Improvement Curve

3

Study 3 – Realistic 3D Environment

We modified the stencil and ham setup to include 3-dimensional components, more closely simulating real endoscopic environments. This enhanced realism supports the validity of the model for real-world training and testing.

Key Findings:

  • • Enhanced realism improved speed (+7.6 mm/min) and hit frequency
  • • No increase in errors despite increased complexity
  • • Strong differentiation of performance by experience maintained
  • • Model retains key benchmarking capacities in life-like setup

3D Environment Setup

4

Study 4 – Human vs AI Scoring

This study compared blinded human rater scores to the AI-generated stability metrics, reinforcing the reliability and objectivity of the AI-based scoring algorithm for scalable assessment.

Key Findings:

  • • AI achieved high correlation with human scores (Spearman ρ = 0.78)
  • • Human model slightly outperformed AI (ρ = 0.88), but AI is scalable
  • • AI identified motion volatility and speed peaks as key predictors
  • • Vision for real-time feedback and remote benchmarking validated

AI Correlation Analysis

The Novelty and Vision

The key innovation of the HAM system lies in its ability to offer standardized benchmarkingof tip control using a validated, AI-assisted protocol. It enables educators and trainers to answer critical questions such as:

Assessment Questions

  • • What are the relative tip control skills of endoscopists in my unit?
  • • Has a trainee transitioned to standard or expert category?
  • • What is the rate of improvement over 3-6 months?
  • • Is this endoscopist likely to excel or are they plateauing?

Analysis Methods

  • • Live: Real-time scoring during procedures
  • • Video: Recorded procedures for asynchronous AI analysis
  • • Remote: Global benchmarking capabilities
  • • Longitudinal: Progress tracking over time
Back to Overview