Collaborative strategies for deploying artificial intelligence to complement physician diagnoses of acute respiratory distress syndrome

Abstract

There is a growing gap between studies describing the capabilities of artificial intelligence (AI) diagnostic systems using deep learning versus efforts to investigate how or when to integrate AI systems into a real-world clinical practice to support physicians and improve diagnosis. To address this gap, we investigate four potential strategies for AI model deployment and physician collaboration to determine their potential impact on diagnostic accuracy. As a case study, we examine an AI model trained to identify findings of the acute respiratory distress syndrome (ARDS) on chest X-ray images. While this model outperforms physicians at identifying findings of ARDS, there are several reasons why fully automated ARDS detection may not be optimal nor feasible in practice. Among several collaboration strategies tested, we find that if the AI model first reviews the chest X-ray and defers to a physician if it is uncertain, this strategy achieves a higher diagnostic accuracy (0.869, 95% CI 0.835–0.903) compared to a strategy where a physician reviews a chest X-ray first and defers to an AI model if uncertain (0.824, 95% CI 0.781–0.862), or strategies where the physician reviews the chest X-ray alone (0.808, 95% CI 0.767–0.85) or the AI model reviews the chest X-ray alone (0.847, 95% CI 0.806–0.887). If the AI model reviews a chest X-ray first, this allows the AI system to make decisions for up to 79% of cases, letting physicians focus on the most challenging subsets of chest X-rays.

Artificial intelligence for the optimal management of community-acquired pneumonia

“…Other studies focused on ARDS complications and survival predictions employing several methodologies, including gas chromatography devices, predictive models based on EHR, NLP, and ML algorithms using radiological data [27–29]. The prediction of ARDS in patients with CAP was based on two predictive models, the Artificial Neural Network (ANN) and LR.…”

“…Other studies focused on ARDS complications and survival predictions employing several methodologies, including gas chromatography devices, predictive models based on EHR, NLP, and ML algorithms using radiological data [27][28][29].…”

Artificial intelligence and the impact on medical genetics

“…Thus, even if much of manual analysis and other human-centric work is replaced by AI and other forms of automation, the increasing volume of testing will mean that many edge cases will remain that require human resolution. The laboratory geneticist of the future-like the pathologist and radiologist-will work increasingly hand-in-hand with assistive AI tools but will still be a key part of the workflow (Farzaneh et al, 2023;He et al, 2023;Kann et al, 2023;Topol, 2019). Second, and related to the previous point, the availability of AI-based tools will continue to decrease the activation energy needed to order genetic testing.…”

Uncertainty-Aware Convolutional Neural Network for Identifying Bilateral Opacities on Chest X-rays: A Tool to Aid Diagnosis of Acute Respiratory Distress Syndrome

“…Machine Learning (ML) methods can reliably and robustly learn complex relationships between clinical data, and several research efforts towards using ML for the predictive modeling of medical diseases are underway [ 5 , 6 , 7 , 8 ]. Similar research efforts for predicting the early onset of ARDS are ongoing to improve clinical recognition of the syndrome [ 9 , 10 , 11 , 12 ]. Although there is no single diagnostic test that can rule in or rule out ARDS, one significant limitation both clinically and with ML models is with reading chest radiographs.…”

“…Interpretation of chest radiographs per the Berlin criteria can be unreliable and subject to inter-rater variability, often leading to missed or delayed ARDS diagnosis [ 13 ]. Therefore, we propose a robust and reliable ML model that can be trained to identify bilateral pulmonary opacities on chest X-ray (CXR) images, which could be an invaluable inclusion in the clinical workflow [ 9 ]. An automatic system that can raise alerts on detecting lung airspace opacities in chest X-rays of critically ill patients can identify possible ARDS cases, which a physician can then review.…”