Prospective Multi-Site Validation of AI to Detect Tuberculosis and Chest X-Ray Abnormalities
NEJM AI · 26.09.2024 · 1 (10)
DOI: 10.1056/aioa2400018 ↗Kurz gefasst
Fachartikel in der Zeitschrift NEJM AI, erschienen am 26.09.2024. Verfasst von Sahar Kazemzadeh und 23 weiteren Autor:innen. Beteiligte Einrichtung: Google, Mountain View, CA, USA. Die Arbeit wurde bisher 24-mal zitiert (Crossref) und nennt 26 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar.
Abstract (Originalsprache)
BACKGROUND: Using artificial intelligence (AI) to interpret chest X-rays (CXRs) could support accessible triage tests for active pulmonary tuberculosis (TB) in resource-constrained settings. METHODS: The performance of two cloud-based CXR AI systems - one to detect TB and the other to detect CXR abnormalities - in a population with a high TB and human immunodeficiency virus (HIV) burden was evaluated. We recruited 1978 adults who had TB symptoms, were close contacts of known TB patients, or were newly diagnosed with HIV at three clinical sites. The TB-detecting AI (TB AI) scores were converted to binary using two thresholds: a high-sensitivity threshold and an exploratory threshold designed to resemble radiologist performance. Ten radiologists reviewed images for signs of TB, blinded to the reference standard. Primary analysis measured AI detection noninferiority to radiologist performance. Secondary analysis evaluated AI detection as compared with the World Health Organization (WHO) targets (90% sensitivity, 70% specificity). Both used an absolute margin of 5%. The abnormality-detecting AI (abnormality AI) was evaluated for noninferiority to a high-sensitivity target suitable for triaging (90% sensitivity, 50% specificity). RESULTS: Of the 1910 patients analyzed, 1827 (96%) had conclusive TB status, of which 649 (36%) were HIV positive and 192 (11%) were TB positive. The TB AI's sensitivity and specificity were 87% and 70%, respectively, at the high-sensitivity threshold and 78% and 82%, respectively, at the balanced threshold. Radiologists' mean sensitivity was 76% and mean specificity was 82%. At the high-sensitivity threshold, the TB AI was noninferior to average radiologist sensitivity (P
Autor:innen und Institutionen
- Google, Mountain View, CA, USA.
- Centre for Infectious Disease Research in Zambia (CIDRZ), Lusaka, Zambia.
- Advanced Clinical, Deerfield, IL.
ORCID-Profile: Sahar Kazemzadeh, Atilla P Kiraly, Zaid Nabulsi, Nsala Sanjase, Minyoi Maimbolwa, Brian Shuma, Shahar Jamshy, Christina Chen, Arnav Agharwal, Charles T Lau, Andrew Sellergren, Daniel Golden, Jin Yu, Eric Wu, Yossi Matias, Katherine Chou, Greg S Corrado, Shravya Shetty, Daniel Tse, Krish Eswaran, Yun Liu, Rory Pilgrim, Monde Muyoyeta, Shruthi Prabhakara
Zitieren
APA
Kazemzadeh, S., Kiraly, A. P., Nabulsi, Z., Sanjase, N., Maimbolwa, M., Shuma, B., Jamshy, S., Chen, C., Agharwal, A., Lau, C. T., Sellergren, A., Golden, D., Yu, J., Wu, E., Matias, Y., Chou, K., Corrado, G. S., Shetty, S., Tse, D., …, & Prabhakara, S. (2024). Prospective Multi-Site Validation of AI to Detect Tuberculosis and Chest X-Ray Abnormalities. NEJM AI, 1(10). https://doi.org/10.1056/aioa2400018
Vancouver
Kazemzadeh S, Kiraly AP, Nabulsi Z, Sanjase N, Maimbolwa M, Shuma B, et al. Prospective Multi-Site Validation of AI to Detect Tuberculosis and Chest X-Ray Abnormalities. NEJM AI. 2024;1(10). doi:10.1056/aioa2400018
BibTeX
@article{Kazemzadeh2024,
author = {Kazemzadeh, Sahar and Kiraly, Atilla P and Nabulsi, Zaid and Sanjase, Nsala and Maimbolwa, Minyoi and Shuma, Brian and Jamshy, Shahar and Chen, Christina and Agharwal, Arnav and Lau, Charles T and Sellergren, Andrew and Golden, Daniel and Yu, Jin and Wu, Eric and Matias, Yossi and Chou, Katherine and Corrado, Greg S and Shetty, Shravya and Tse, Daniel and Eswaran, Krish and Liu, Yun and Pilgrim, Rory and Muyoyeta, Monde and Prabhakara, Shruthi},
title = {Prospective Multi-Site Validation of AI to Detect Tuberculosis and Chest X-Ray Abnormalities},
journal = {NEJM AI},
year = {2024},
volume = {1},
number = {10},
doi = {10.1056/aioa2400018}
}
EUVIMED