Project Overview
An NHS Foundation Trust reporting 85,000 radiology examinations per year had a radiology reporting backlog — a...
Technology Stack
Compliance & Standards
The Challenge
An NHS Foundation Trust reporting 85,000 radiology examinations per year had a radiology reporting backlog — average 12 days wait for routine chest X-ray reports, 4.2 days for urgent. The trust wanted to build an AI-assisted reporting tool: triage AI for urgent finding prioritisation (flagging potential pneumothorax, PE, pneumonia for urgent review) and a structured reporting template engine to reduce radiologist typing time. DCB0129 clinical safety, DTAC all 5 domains, MHRA medical device regulations (AI software as medical device), and UK GDPR Article 9 were mandatory. Budget: £95,000.
Our Approach
MHRA Medical Device Classification
- AI-assisted radiology reporting software is a Class IIa medical device under UK MDR 2002.
- MHRA registration required before clinical use.
Clinical Evaluation Report
- literature review of chest X-ray AI performance characteristics, comparison with intended purpose.
- CE marking (legacy) or
UKCA marking
Clickmasters worked with the trust's MHRA-registered technical file under the CE transition period.
Triage AI Algorithm
Pre-trained chest X-ray AI model (using CheXNet architecture, fine-tuned on UK NHS Trust's anonymised historical imaging data — DICOM images with radiologist reports as labels).
Output
probability score for urgent findings (pneumothorax, pneumonia, pleural effusion, cardiomegaly).
Threshold calibration
95% sensitivity for pneumothorax (missing one is catastrophic).
Worklist prioritisation
high-probability urgent findings pushed to top of reporting queue.
CSO
Consultant Radiologist (clinical safety officer must be a clinical professional for medical device AI).
Hazard Log
false negative hazard (AI misses urgent finding) — mitigated by: AI as triage support only (radiologist reviews all exams), high sensitivity threshold, and mandatory clinical review.
False positive hazard
- unnecessary urgent recall — mitigated by positive predictive value monitoring with radiologist override audit trail.
- DTAC Domain 1–5
Evidence
- Domain 1 (Clinical Safety): DCB0129 Clinical Safety Case Report.
- Domain 2 (Data Protection): Article 9 DPIA, DSP Toolkit assessment.
- Domain 3 (Technical Security): Cyber Essentials Plus, CREST pen test.
- Domain 4 (Interoperability): FHIR R4 DiagnosticReport integration with EPR, DICOM standards compliance.
- Domain 5 (Usability): radiologist user research (12 participants), WCAG 2.1 AA.
The Results
DTAC approved all 5 domains.
MHRA registered.
Clinical deployment at 20 weeks, £88,000 — under budget.
Urgent finding detection: 12 pneumothorax cases flagged in first 3 months — all confirmed by radiologist as genuine (zero false positives requiring urgent recall in first 3 months).
Routine chest X-ray reporting wait: 12 days → 7.2 days (39% improvement from worklist optimisation alone).
Radiologist reporting time: structured template reduced average report time 4.8 → 3.1 minutes.
“Twelve pneumothorax cases in three months — every one a genuine urgent finding, zero false positives. The DCB0129 process with a radiologist as CSO was absolutely right for a medical device AI. DTAC first submission pass on a Class IIa medical device is something we will be proud of for a long time." — Chief Clinical Information Officer, NHS Foundation Trust (name withheld)”
Project Details
Related Case Studies
View AllHealthTech Diabetes Management App — NHS Pilot, DTAC Approved
A UK digital health charity wanted to build a consumer-facing diabetes self-management app, seeking NHS App Li...
HealthTech Mental Health App — NHS Talking Therapies Integration
A UK digital mental health company wanted to build a self-guided CBT (Cognitive Behavioural Therapy) app for m...
HealthTech Remote Patient Monitoring Platform — UK Community Nursing
A UK community healthcare provider delivering care to 4,800 housebound patients needed a remote patient monito...
HealthTech NHS GP Appointment Booking — Integrated Care System
An NHS Integrated Care System (ICS) covering 12 GP practices and 240,000 registered patients wanted to replace...
Ready to Transform Your Business?
Let's discuss how our technical expertise can help you achieve remarkable results.