AI medical trial: "Carebot AI CXR" reads chest X-rays vs. radiologists (completed)

1Testing chest X-ray reading in practice

A clinical study (NCT05963945) evaluated Carebot's AI chest X-ray (CXR) reading software against radiologists in real-world radiology practice.

2A multi-reader, retrospective design

Per the registry, it is a multi-reader, retrospective study whose primary objective is to evaluate the performance parameters of the proposed DLAD (Deep Learning Automated Detection) "Carebot AI CXR" compared with individual radiologists. The chest X-ray findings span pneumothorax, solitary pulmonary nodule, atelectasis, subcutaneous emphysema, cardiomegaly, consolidation, and pleural effusion.

Enrollment was 956; the period was October 18, 2022 – March 21, 2023; status as of the check date is completed.

3Because it is the most-ordered imaging test

Chest X-ray is among the most basic and high-volume imaging exams, and reading it requires expertise and time. AI automated detection is expected to help reduce misses, triage reads, and offset radiologist shortages. Comparing AI detection directly with multiple radiologists, as here, is a standard design for demonstrating the clinical validity of imaging AI.

4What you compare against changes what you learn

The meaning of an evaluation depends on the standard chosen. This study chose to compare directly against several radiologists.

The chest X-ray findings covered range widely: pneumothorax, solitary pulmonary nodule, atelectasis, subcutaneous emphysema, cardiomegaly, consolidation and pleural effusion. There are 956 cases, running from October 2022 to March 2023, and the status at the time of checking is complete. Chest radiography is the most basic and most frequently performed imaging study, and reading it takes expertise and time.