CanadaHealthcare and life sciences2024 – ongoingUniversity Health Network
Pneumothorax imaging prioritization with chest-tube detection
UHN reported 87% accuracy for its pneumothorax model and added chest-tube screening
Problem
Limited overnight radiology staffing could delay urgent scan review. A pneumothorax model also identified chest tubes in patients who had already received treatment.
Approach
UHN trained its detection tool on external hospital datasets and validated it internally. The team added a chest-tube model to exclude treated cases from prioritization. Priority flags appeared in CORAL, its radiology dashboard, while CyclOps generated weekly monitoring reports.
Results
Takeaway
Chest-tube screening addressed a clinically unhelpful shortcut. Reported model accuracy did not establish faster treatment or improved patient outcomes.
Categories
- Size
- Large enterprise
- Industry
- Healthcare and life sciences
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Built in-house
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
The report recounted UHN clinicians' deployment experience. It described internal accuracy, chest-tube filtering and ongoing monitoring, without quantified clinical outcomes.
Full details are in the original sources.