PPubMed4 Oct 2024Cited 10×

Applications of artificial intelligence in emergency and critical care diagnostics: a systematic review and meta-analysis

Jithin K. Sreedharan, Fred Saleh, Abdullah Alqahtani, Ibrahim Ahmed Albalawi, Gokul Krishna Gopalakrishnan, Hadi Abdullah Alahmed, Basem Ahmed Alsultan, Dhafer Mana Alalharith, Musallam Alnasser, Ayedh Dafer Alahmari, Manjush Karthika

Abstract

The search strategy yielded 12,173 records from PubMed, ScienceDirect and Medline. After sequential and systematic search and filtration, 17 studies in total were included for this review. Although only 17 studies were included, it is important to note that the methodology followed strict inclusion criteria and systematic review guidelines, to ensure the validity of the objectives. A detailed flow of study selection is depicted in [fig3]. The 17 studies included in this review, focused on the accuracy and ability of AI to detect and diagnose various disease conditions like Barrette’s neoplasia

A figure from Applications of artificial intelligence in emergency and critical care diagnostics: a systematic review and meta-analysis — fig. from the paper

§ The Valyu brief

The paper, in five minutes

Read from the full text via Valyu Contents

AI improves diagnostic accuracy and early detection in emergency and critical care, enhancing patient outcomes.

What’s new

i.Systematic review and meta‑analysis of AI applications in emergency and critical care diagnostics
ii.Extraction and synthesis of diagnostic performance metrics (sensitivity, specificity, PPV, NPV) from 17 studies
iii.Assessment of study quality and bias using the QUADAS‑2 tool
iv.Identification of significant heterogeneity across studies (p‑value < 0.05)
v.Highlighting gaps and proposing future research directions for AI in critical care

How they did it

The authors searched PubMed, ScienceDirect and Medline for studies published between 1 Jan 2013 and 1 Feb 2023, screened records with the Rayyan tool, and selected 17 eligible studies. Risk of bias was evaluated using QUADAS‑2, and data were extracted by three independent reviewers. Meta‑analysis employed chi‑square tests, regression analysis, and calculation of sensitivity, specificity, likelihood ratios, and AUC with 95 % confidence intervals.

What they found

17 studies were included out of 12,173 screened records
All studies showed significant heterogeneity (p < 0.05)
Barrett’s neoplasia AI models achieved up to 95 % sensitivity and 85 % specificity
Gastrointestinal stromal tumor AI models reached 100 % sensitivity and up to 100 % specificity
Sepsis and cardiac arrest AI models reported 87 %–98 % sensitivity and 89 %–97 % specificity

Caveats

✕Limited number of studies focusing on a narrow set of conditions restricts generalizability
✕Absence of long‑term outcome data prevents assessment of sustained clinical impact
✕Variability in reported metrics and missing PPV/NPV data hinder full accuracy evaluation
✕Lack of standardized protocols and external validation across diverse clinical settings

§ The Valyu brief

Reading the full paper and taking notes. This takes a few seconds…

§ Ask this paper

Ask a question about this paper

Valyu reads the full text and answers from what the paper actually says.

§ Related work in other archives

Similar papers in all four archives

The same idea searched in arXiv, PubMed, bioRxiv and chemRxiv at once, so a method can lead you to where it is being applied.

AarXiv4 found

PPubMed4 found

BbioRxiv4 found

CchemRxiv3 found

Searching the other archives…