Table 3. Studies highlighting the use of artificial intelligence in acute respiratory distress syndrome (ARDS).

ANN - Artificial Neuronal Network; ARDS - Acute Respiratory Distress Syndrome; CR - Clinical Recognition; EHR - Electronic Health Records; ML - Machine Learning, n - Number of Subjects, SAP - Severe Acute Pancreatitis

Author	Year	Study type (n)	Aim	Main Conclusion
You J.Y. et al. [16]	2020	Retrospective (n=1,305)	Compare rate and time of recognition of ARDS by ML with bedside  CR	ML algorithm identified more cases of ARDS as compared to CR. No difference in the rate of identification
Le S. et al. [17]	2020	Retrospective (n=9,919)	Develop a model trained on patient data of health record to predict ARDS	Supervised ML can predict ARDS up to 48 hours before its actual onset
Sinha P. et al. [18]	2020	Retrospective (n=2,022 in training vs n=745 validation data)	Classify ARDS phenotypes by models trained on a clinical data set	ML models can accurately identify ARDS phenotypes
Zeiberg D. et al. [19]	2019	Retrospective (n=1,621 in training vs n=1,122 in test cohort)	Develop an ML approach to predict ARDS based on EHR	It is feasible to use the ML approach to risk-stratify patients for ARDS based on EHR
Fei Y. et al. [20]	2019	Retrospective(n=217)	Use ANN to predict and determine the severity of ARDS in SAP patients	Novel ANN can be used to predict ARDS in SAP