A systematic review of machine learning models for management, prediction and classification of ARDS.

AIM: Acute respiratory distress syndrome or ARDS is an acute, severe form of respiratory failure characterised by poor oxygenation and bilateral pulmonary infiltrates. Advancements in signal processing and machine learning have led to promising solutions for classification, event detection and predictive models in the management of ARDS. METHOD: In this review, we provide systematic description of different studies in the application of Machine Learning (ML) and artificial intelligence for management, prediction, and classification of ARDS. We searched the following databases: Google Scholar, PubMed, and EBSCO from 2009 to 2023. A total of 243 studies was screened, in which, 52 studies were included for review and analysis. We integrated knowledge of previous work providing the state of art and overview of explainable decision models in machine learning and have identified areas for future research. RESULTS: Gradient boosting is the most common and successful method utilised in 12 (23.1%) of the studies. Due to limitation of data size available, neural network and its variation is used by only 8 (15.4%) studies. Whilst all studies used cross validating technique or separated database for validation, only 1 study validated the model with clinician input. Explainability methods were presented in 15 (28.8%) of studies with the most common method is feature importance which used 14 times. CONCLUSION: For databases of 5000 or fewer samples, extreme gradient boosting has the highest probability of success. A large, multi-region, multi centre database is required to reduce bias and take advantage of neural network method. A framework for validating with and explaining ML model to clinicians involved in the management of ARDS would be very helpful for development and deployment of the ML model.

Quantifying heterogeneity in an animal model of acute respiratory distress syndrome, a comparison of inspired sinewave technique to computed tomography.

The inspired sinewave technique (IST) is a non-invasive method to measure lung heterogeneity indices (including both uneven ventilation and perfusion or heterogeneity), which reveal multiple conditions of the lung and lung injury. To evaluate the reproducibility and predicted clinical outcomes of IST heterogeneity values, a comparison with a quantitative lung computed tomography (CT) scan is performed. Six anaesthetised pigs were studied after surfactant depletion by saline-lavage. Paired measurements of lung heterogeneity were then taken with both the IST and CT. Lung heterogeneity measured by the IST was calculated by (a) the ratio of tracer gas outputs measured at oscillation periods of 180 s and 60 s, and (b) by the standard deviation of the modelled log-normal distribution of ventilations and perfusions in the simulation lung. In the CT images, lungs were manually segmented and divided into different regions according to voxel density. A quantitative CT method to calculate the heterogeneity (the Cressoni method) was applied. The IST and CT show good Pearson correlation coefficients in lung heterogeneity measurements (ventilation: 0.71, and perfusion, 0.60, p < 0.001). Within individual animals, the coefficients of determination average ventilation (R2 = 0.53) and perfusion (R2 = 0.68) heterogeneity. Strong concordance rates of 98% in ventilation and 89% when the heterogeneity changes were reported in pairs measured by CT scanning and IST methods. This quantitative method to identify heterogeneity has the potential to replicate CT lung heterogeneity, and to aid individualised care in ARDS.