Purpose of review Artificial intelligence has been integrated in nearly all aspects of clinical care to improve patient outcomes, augment human capabilities and improve efficiency. Particularly in the field of radiology and medical imaging, artificial intelligence could revolutionize how care is delivered. In this review, we review the recent literature and provide an assessment of the advantages (pros) and limitations (cons) of artificial intelligence in point-of-care ultrasonography (POCUS). Recent findings Emerging literature suggests that artificial intelligence assisted diagnostic models offer a performance advantage over standard imaging modalities with regards to image acquisition and diagnostic accuracy particularly with less experienced users. In pediatric POCUS, artificial intelligence has been shown to improve image acquisition and augment education, which is remarkably helpful in areas where ultrasound experts are limited. Nevertheless, integration of artificial intelligence in the growing field of POCUS requires careful assessment of its drawbacks, biases and limitations. Models that are trained primarily on adult populations should be assessed and validated before utilization in the pediatric population to ensure generalizability. Furthermore, the use of artificial intelligence should integrate with and not replace existing educational models and credentialing processes to preserve ultrasound skills. Finally, at the institutional and global levels, hospitals and organizations will need to weigh in on policies, data governance and oversight in this vulnerable population. Summary The use of artificial intelligence in POCUS in the fields of emergency medicine and critical care is promising but should be viewed with a lens of caution. It holds promise for improving accessibility, reducing variability in care and transforming care in resource-limited settings, but integration of this evolving technology should be thoughtful to address its potential limitations.
Neal et al. (Mon,) studied this question.