Psychophysiological models of hypovigilance detection: A scoping review

Abstract

Hypovigilance represents a major contributor to accidents. In operational contexts, the burden of monitoring/managing vigilance often rests on operators. Recent advances in sensing technologies allow for the development of psychophysiology‐based (hypo)vigilance prediction models. Still, these models remain scarcely applied to operational situations and need better understanding. The current scoping review provides a state of knowledge regarding psychophysiological models of hypovigilance detection. Records evaluating vigilance measuring tools with gold standard comparisons and hypovigilance prediction performances were extracted from MEDLINE, PsychInfo, and Inspec. Exclusion criteria comprised aspects related to language, non‐empirical papers, and sleep studies. The Quality Assessment tool for Diagnostic Accuracy Studies (QUADAS) and the Prediction model Risk Of Bias ASsessment Tool (PROBAST) were used for bias evaluation. Twenty‐one records were reviewed. They were mainly characterized by participant selection and analysis biases. Papers predominantly focused on driving and employed several common psychophysiological techniques. Yet, prediction methods and gold standards varied widely. Overall, we outline the main strategies used to assess hypovigilance, their principal limitations, and we discuss applications of these models.

Identifying Episodes of Hypovigilance in Intensive Care Units Using Routine Physiological Parameters and Artificial Intelligence: a Derivation Study (Preprint)

“…We employed three distinct AI models for detecting hypovigilance events: Random Forest (RF) from the Python Scikitlearn library [48]: eXtreme Gradient Boosting (XGBoost) from the XGBoost library [49], and the Light Gradient-Boosting Machine classifier (LGBM) from the LightGBM library [50]. We chose these three particular classifiers because RF and XGBoost were used in prior studies to identify delirium and hypovigilance [28]. LightGBM was also used because of its previous application in other ICU databases such as the Medical Information Mart for Intensive Care III database [51].…”

“…The majority of research on hypovigilance has previously been conducted in laboratory settings, which offer a highly regulated setting but might not accurately reflect real-world ICU situations [28]. The field of vigilance research has been hindered by inconsistent and poorly defined terminology, making it hard to compare our results to all of the existing literature on the subject [27].…”

“…Prediction models using artificial intelligence (AI) have been developed to quantify hypovigilance or related concepts (e.g., drowsiness and fatigue) using psychobehavioural correlates of vigilance in laboratory conditions [27,28], but realworld examples are few. A recent scoping review identified 21 psychophysiological models of hypovigilance detection, in which almost all relied on at least one of the following signals, targeting both central and autonomic nervous systems: electrocardiography, photoplethysmography, electroencephalography, electro-oculography, and eye tracking [28]. While sensitive, these systems are resource-intensive in dynamic environments such as the ICU.…”

“…Despite existing research on hypovigilance, and the potential applications and clinical implications of developing models for hypovigilance diagnosis, there is no consensus in the literature regarding the most accurate method. There are also significant barriers to the widespread use of more sophisticated diagnostic modalities such as electroencephalography in clinical settings because of poor signal quality [28], specialized and costly equipment requirements [29], and patient discomfort due to extended wear [30]. These factors hinder the adoption of these sensors in routine healthcare settings.…”

“…Current delirium prediction and detection models rely on static data that fail to capture the rapid-cycle fluctuating nature of vigilance [2]. Moreover, vigilance detection models developed in lab environments lack real-world clinical validation [28]. Clinical real-world settings, such as ICUs, can provide a reliable data collection environment where patients often experience frequent episodes of hypovigilance.…”