Page 64 - JSOM Spring 2026

of novice users. With AI assistance, military medics with lim- 4. Monti JD, Perreault MD. Impact of a 4-hour introductory eFAST
ited ultrasound training achieved substantial improvements in training intervention among ultrasound-naïve U.S. military med-
sensitivity, specificity, accuracy, and AUROC when identifying ics. Mil Med. 2020;185(5–6):e601–e608. doi:10.1093/milmed/
usaa014
absent lung sliding. In addition to improving diagnostic accu- 5. Arntfield R, Wu D, Tschirhart J, et al. Automation of lung ultra-
racy, AI support increased user confidence, resulting in a greater sound interpretation via deep learning for the classification of
proportion of high-certainty assessments and a reduction in normal versus abnormal lung parenchyma: a multicenter study.
uncertain assessments. These findings suggest that AI-enabled Diagnostics (Basel). 2021;11(11):2049. doi:10.3390/diagnostics
decision support may help bridge experience gaps in prehos- 11112049
pital care and serve as a valuable asset for frontline providers 6. Fiedler HC, Prager R, Smith D, et al. Automated real-time detec-
working in resource-limited or austere environments. tion of lung sliding using artificial intelligence: a prospective diag-
nostic accuracy study. Chest. 2024;166(2):362–370. doi:10.1016/
j.chest.2024.02.011
Author Contributions 7. Wu D, Smith D, VanBerlo B, et al. Improving the generalizabil-
All authors conceived the study and designed the method- ity and performance of an ultrasound deep learning model using
ology. RA obtained funding and Institutional Review Board limited multicenter data for lung sliding artifact identification.
approval. All authors coordinated data collection and carried Diagnostics (Basel). 2024;14(11):1081. doi:10.3390/diagnostics
out the experiment sessions. NO and DS performed the data 14111081
analysis. All authors drafted the manuscript and provided crit- 8. Tenajas R, Miraut D, Illana CI, Alonso-Gonzalez R, Arias-Valcayo
F, Herraiz JL. Recent advances in artificial intelligence-assisted
ical revisions. RA provided oversight throughout the project. ultrasound scanning. Appl Sci. 2023;13(6):3693. doi:10.3390/
All authors reviewed and approved the final manuscript. The app13063693
authors would like to thank Richard Neading for his support 9. Mongodi S, Arioli R, Quaini A, Grugnetti G, Grugnetti AM,
on this project. Mojoli F. Lung ultrasound training: how short is too short? Obser-
vational study on the effects of a focused theoretical training for
novice learners. BMC Med Educ. 2024;24(1):166. doi:10.1186/
Disclaimer s12909-024-05148-0
This study was conducted under institutional ethics approval 10. Kim S, Fischetti C, Guy M, Hsu E, Fox J, Young SD. Artificial in-
(Advara Research Ethics Board). The authors adhered to stan- telligence (AI) applications for point of care ultrasound (POCUS)
dard military Public Affairs and Operational Security guide- in low-resource settings: a scoping review. Diagnostics (Basel).
lines; no sensitive unit information or tactics are disclosed. The 2024;14(15):1669. doi:10.3390/diagnostics14151669
views expressed in this article are those of the authors and do
not reflect the official policy or position of the U.S. Marine PMID: 41774835;
Corps, the U.S. Department of Defense, the Canadian Depart- DOI: 10.55460/J.Spec.Oper.Med.2026.1SDN-NWTW
ment of National Defence, or any other agency.
Disclosures
KT, DS, and RA helped with data collection, data analysis, and APPENDIX 1 LUS Findings Definitions.
oversight, respectively, on this project and were paid through Pleural Line: The bright horizontal artifact seen at the interface
Deep Breathe Inc. All other authors report no financial disclo- between the lung tissue and chest wall.
sures. The AI model used was developed by Deep Breathe Inc. Lung Sliding: A subtle shimmering or sliding motion of the pleural
and was not modified for this study. KT, DS, and RA disclose line observed during breathing. Presence of this artifact rules out
a relationship with the company Deep Breathe Inc. RP is a pneumothorax at that probe position.
consultant with Deep Breathe.
APPENDIX 2 Confidence Rating Scale.
Funding
This research was supported by the Office of Naval Research Rating scale Definition
Global (ONRG). 1 – Not at all confident You are highly uncertain about your
assessment. You are guessing.
References 2 – Slightly confident You are somewhat unsure but leaning
1. Meadows RM, Monti JD, Umar MA, et al. US Army Combat medic toward a label.
performance with portable ultrasound to detect sonographic find- 3 – Moderately confident You are reasonably confident, though
ings of pneumothorax in a cadaveric model. J Spec Oper Med. not completely certain.
2020;20(3):71–75. doi:10.55460/SOPZ-STAP 4 – Confident You are confident in your decision,
2. Gentry Wilkerson R, Stone MB. Sensitivity of bedside ultrasound with minimal hesitation.
and supine anteroposterior chest radiographs for the identification 5 – Very confident You are certain about your label with
of pneumothorax after blunt trauma. Acad Emerg Med. 2010;17 no hesitation.
(1):11–17. doi:10.1111/j.1553-2712.2009.00628.x
3. Nhat PTH, Van Hao N, Tho PV, et al. Clinical benefit of AI-
assisted lung ultrasound in a resource-limited intensive care unit.
Crit Care. 2023;27(1):257. doi:10.1186/s13054-023-04548-w

62 | JSOM Volume 26, Edition 1 / Spring 2026