BrazilHealthcare and life sciences2024 – ongoingRede D'Or São Luiz
NLP screening of radiology reports for suspected lung cancer
Rede D'Or deployed AI report screening across hospitals after reporting 98% accuracy in testing
Problem
Chest scans performed for other conditions sometimes revealed suspicious lung nodules that did not receive follow-up. During the pandemic, some patients did not return for their results.
Approach
IDOR researchers developed an NLP tool to read chest CT reports for descriptions of potentially malignant nodules. They trained it using more than 20,000 reports and conducted supervised testing. Rede D'Or subsequently used the tool in several hospitals to identify patients who might have overlooked lung cancer.
Results
Takeaway
The tool searched existing reports for findings that might need follow-up. Its reported accuracy did not establish an improvement in patient outcomes.
Categories
- Size
- Large enterprise
- Industry
- Healthcare and life sciences
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Built in-house
- Goal
- Quality and risk
- Evidence
- Company's own claim
Sources
Source summaries
IDOR reported developing an NLP tool using more than 20,000 radiology reports and reaching 98% accuracy in testing. It confirmed use in several Rede D'Or hospitals to find patients with potentially overlooked lung cancer.
Full details are in the original sources.