Evaluating the benefits of machine learning for diagnosing deep vein thrombosis compared with gold standard ultrasound: a feasibility study
2024

Using AI to Diagnose Deep Vein Thrombosis with Ultrasound

Sample size: 91 publication 10 minutes Evidence: moderate

Author Information

Author(s): Nothnagel Kerstin, Aslam Mohammed Farid

Primary Institution: University of Bristol

Hypothesis

Can AI-guided point-of-care ultrasound effectively diagnose deep vein thrombosis by non-specialists?

Conclusion

AI-guided ultrasound can help non-specialists diagnose deep vein thrombosis effectively, potentially reducing the need for formal scans.

Supporting Evidence

  • 91% of scans had sufficient quality for diagnosis.
  • Sensitivity was 100% and specificity was 90.57%.
  • 53% of participants were classified as low risk, potentially avoiding formal scans.

Takeaway

This study shows that a smartphone app can help doctors without special training take pictures of veins to check for blood clots.

Methodology

Patients underwent AI-guided ultrasound scans followed by formal scans, with images reviewed remotely by specialists.

Potential Biases

Potential bias from the non-specialist operators and the retrospective nature of the study.

Limitations

The study had a small number of positive DVT cases and was retrospective, requiring further validation.

Participant Demographics

The participants were predominantly older females with an average age of 69.7 years.

Statistical Information

P-Value

0.0001

Confidence Interval

95% CI = 99.12% to 100%

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.3399/BJGPO.2024.0057

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication