Case study
External validation of RBfracture™ on South American data
Share this case study
About One Laudos
Using RBfracture™ in high volume teleradiology in South America
One Laudos, a teleradiology and radiology management company in Brazil, partnered with Radiobotics to assess the diagnostic performance of RBfracture™ in high-volume settings.
The aim was to show the potential for using RBfracture as a clinical decision support tool for trauma X-rays on South American patients.
Can an AI-powered solution like RBfracture™ that automatically detects trauma-related findings improve diagnostic performance in a high-throughput setting?
Purpose and scale
Audit details
The audit included an evaluation of RBfracture™ through a focused retrospective review of 912 MSK cases
Radiographs were independently annotated by qualified annotators who, supported by the original radiological report, analysed the radiographs and identified fractures in the images. In case of disagreement between the two annotators, adjudication was conducted by a third independent annotator who chose which annotator was right, supported by the original radiological report. In case of disagreement among all three annotators, the study was discarded.
Of the 912 cases analysed, there were discrepancies in 187 cases (20%). Of the 187 cases where there was a discrepancy, RBfracture was correct in 157 cases (84%).
Results
We saw a reduction in missed fractures by 88% — a significant decrease in missed fractures by teleradiology clinicians
97%
Accuracy
97%
Sensitivity
96%
Specificity
In the media:
Ferramenta de IA inédita na América Latina ajuda no diagnóstico de fraturas
The value in teleradiology
The value of RBfracture™
RBfracture is trained on emergency data from +1.300 hospitals around the world, ensuring high performance across diverse patient populations
Reduces missed fractures
Improves patient outcomes
clinician skills
Measuring the standalone performance of RBfracture™
Predictive values indicate how accurate RBfracture was in correctly identifying cases with fractures (Positive Predictive Value) and non-fractures (Negative Predictive Value).
95%
Positive predictive value
98%
Negative predictive value
Negative Predictive Value (NPV) and Positive Predictive Value (PPV) are statistical measures used in diagnostic testing to assess the effectiveness of a test in identifying the presence or absence of a condition. They provide insight into the reliability of test results in real-world populations.
PPV is the proportion of individuals with a positive test result who actually have the condition.
NPV is the proportion of individuals with a negative test result who do not have the condition.
RBfracture’s PPV and SPV scores from our One Laudos audit demonstrate the potential for using RBfracture as a clinical decision support tool for trauma X-rays.
