The following is a summary of “Reverse triggering neural network and rules-based automated detection in acute respiratory distress syndrome,” published in the January 2023 issue of  Critical Care by Kassis et al.

Dyssynchrony, a condition characterized by a lack of coordination between the patient’s respiratory efforts and mechanical ventilation, has been identified as a potential cause of lung injury and is linked to unfavorable outcomes in patients receiving mechanical ventilation. Reverse triggering (RT) is a prevalent form of dyssynchrony that manifests in various phenotypes. These phenotypes can lead to direct pulmonary damage and pose a challenge in identification. Given the aforementioned difficulties, there is a necessity for automated software to aid in the process of identification. The present investigation was a prospective observational study that employed a training cohort of 15 patients and a validation cohort of 13 patients. Respiratory Triggered (RT) occurrences were manually detected and juxtaposed with “rules-based” algorithms with and without esophageal manometry and reverse triggering with breath stacking. 

These were then employed to educate a neural network artificial intelligence (AI) system. The identification of RT phenotypes was carried out following established criteria. Program performance was evaluated by analyzing sensitivity, specificity, positive predictive value (PPV), and F1 score. About 33,244 respiratory cycles were manually scrutinized, of which 8,718 were manually recognized as exhibiting reverse triggers. The programs based on rules and artificial intelligence demonstrated exceptional specificity, with a value exceeding 95% in all programs.

Additionally, the F1 score was greater than 75% in all programs. The most frequently observed phenotypes were RT, with breath stacking at 24.4% and mid-cycle RT at 37.8%. The automated identification of the respiratory tract has exhibited commendable efficacy, thereby indicating the possibility of utilizing such programs for both clinical and research purposes.

Source: sciencedirect.com/science/article/abs/pii/S0883944123000059