Back to Swatches

NHS - Medical Bone Imaging Pathology Detection

Fabric Score 2.5
PublicHealthNational Health Service UKGenAI modelUser / Operator / Front-line Staff ×1

Workflow Diagram

NHS - Medical Bone Imaging Pathology Detection workflow diagram

Fabric Score

ValueEfficacySecurityRiskExternalitiesEfficiency2.22.33.72.23.01.92.5Fabric Score

Task

Patient undergoes radiography - example here is x-ray scan. The AI system evaluates the image and detects pathology - e.g. fracture. It can then flag this to the clinician and add to priority list.

Intent

The AI system aims to identify pathology more efficiently and flag this to clinicians.

AI Workflow

Input

A radiograph (x-ray) is taken of the patient.

Process

The AI system detects pathology.

Output

The AI system generates a report and adds to the worklist. It prioritises on Worklist the ones which have a pathology. Human (A+E or Radiologist) decides whether this is accurate and accepts or rejects the report.

Human Oversight Level

Human-Led with AI Assistance

Institutional Oversight Examples

  • •Data protection regulation
  • •Clinician oversight
  • •Medical Device approved

Risk

Wrong pathology could be picked up and bias clinician decision making.

Output Modification Telemetry

  • Factual errors / hallucinations80%
  • Unfaithful outputs (policy, tone, or style misalignment)10%
  • Irrelevant information5%
  • Internal inconsistency5%

Transversal Metrics

Grouped by Fabric dimension.

Efficacy

Accuracy60 %

Efficiency

Modification Rate80 %
Modification Time15 min/output
Verification Time15 min
Rejection Rate30 %
Operational Friction10 hrs/wk
Implementation Overhead200 hrs
Governance Overhead200 hrs
Time to Launch12 wks

Value

Effort Reduction15 %

Risk

Reliability5 incident/month
Autonomy90 %