Annalise Enterprise CXR Triage Trauma

Annalise Enterprise CXR Triage Trauma is a software tool utilizing a deep-learning artificial intelligence algorithm to analyze adult chest X-rays for features suggestive of vertebral compression fractures, aiming to prioritize these cases in clinical worklists. The system integrates with PACS/RIS and demonstrated high diagnostic accuracy (AUC 0.954) and specific sensitivity/specificity performance in identifying vertebral compression fractures from erect CXRs, thereby facilitating expedited review for clinicians in specialized bone health programs.

Annalise Enterprise CXR Triage Trauma is an AI-powered software workflow tool designed to assist clinicians in the medical care environment. Its primary function is to analyze adult chest X-ray (CXR) studies to identify features suggestive of vertebral compression fractures. The system employs a convolutional neural network, trained using deep-learning techniques on diverse datasets, to detect these findings. Upon identification, the device integrates with existing Picture Archiving and Communication Systems (PACS) or Radiology Information Systems (RIS) to provide case-level output, enabling worklist prioritization or triage, specifically for clinicians in Bone Health and Fracture Liaison Service programs. It operates in parallel with standard clinical workflows and is not intended for standalone diagnostic decision-making, to direct attention to specific image portions other than vertebral compression fractures, or to rule out other critical findings. The device's performance was validated through a standalone, retrospective study using 589 CXR cases (272 positive for vertebral compression fracture, 317 negative) independent of the training data, sourced from four US hospital networks, with ground truth established by board-certified radiologists. For vertebral compression fractures, the AI algorithm demonstrated an Area Under the Curve (AUC) of 0.954 (95% CI: 0.939-0.968). At specific operating points, it achieved sensitivities of 89.3% (95% CI: 85.7-93.0) and 85.3% (95% CI: 80.9-89.3) with corresponding specificities of 89.0% (95% CI: 85.8-92.1) and 90.9% (95% CI: 87.7-94.0) respectively. The system also showed an average triage turn-around time of 30.0 seconds, though a noted limitation is that the performance evaluation was conducted using only erect positioning CXRs, and performance with supine positioning may differ.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the device's performance change when analyzing CXRs taken with supine positioning, given the current evaluation was limited to erect positioning?; What specific impact does the 30-second average triage turn-around time have on overall patient management and outcomes in Bone Health and Fracture Liaison Service programs?; Beyond vertebral compression fractures, are there plans to expand the AI algorithm's capability to detect other subtle or commonly missed traumatic findings on chest X-rays?; How does the algorithm differentiate between acute and chronic vertebral compression fractures, or does it flag all such fractures regardless of suspected acuity?; What were the characteristics of the false positive and false negative cases in the standalone performance study for vertebral compression fractures, and what strategies could further refine the algorithm's accuracy?