Obstructive sleep apnea (OSA) is a respiratory disorder characterized by interruption to breathing during sleep. Usually, the OSA is more severe in the supine sleeping position. Recent studies also demonstrated that the head position may play an important role in the pathophysiology of the OSA. Therefore, monitoring the sleeping body and the head position has high clinical importance to optimize the treatment of the OSA. In this paper, three machine learning approaches were used to detect the head position during sleep in infrared images. In the first two methods, supervised classifiers were trained to estimate the head position based on different feature sets extracted from infrared images. In the third method, three different convolutional neural network (CNN) structures (ResNet50, MobileNet, and Darknet19) were trained to detect the head position during sleep. To detect the body position, the same CNN architectures were trained on infrared images. Overnight sleeping data (sleep duration = 5±1 h) from 50 participants (age: 53 ± 15 years, BMI: 29 ± 6 kg/m2, and 30 men/20 women) with various levels of OSA severity as measured by the apnea-hypopnea index (AHI = 25 ± 29 events/h and OSA severity: 12 normal, 13 mild, 11 moderate, and 14 severe) were collected for this paper. The models were trained on the data collected in one laboratory room from half of the participants and tested on the data from the other half collected in a different laboratory room. The best performing model (Darknet19) correctly estimated the lateral versus supine head position with 92% accuracy and 94% F1-Score and the lateral versus supine body position with 87% accuracy and 87% F1-Score. INDEX TERMS Computer vision, machine learning, position detection, sleep apnea, non-contact monitoring. The associate editor coordinating the review of this manuscript and approving it for publication was Thomas Penzel.

Background: Falls are a major health concern, with one in three adults over the age of 65 falling each year. A key gait parameter that is indicative of tripping is minimum foot clearance (MFC), which occurs during the mid-swing phase of gait. This is the second of a two-part scoping review on MFC literature. The aim of this paper is to identify vulnerable populations and conditions that impact MFC mean or median relative to controls. This information will inform future design/maintenance standards and outdoor built environment guidelines. Methods: Four electronic databases were searched to identify journal articles and conference papers that report level-ground MFC characteristics. Two independent reviewers screened papers for inclusion. Results: Out of 1571 papers, 43 relevant papers were included in this review. Twenty-eight conditions have been studied for effects on MFC. Eleven of the 28 conditions led to a decrease in mean or median MFC including dual-task walking in older adults, fallers with multiple sclerosis, and treadmill walking. All studies were conducted indoors. Conclusions: The lack of standardized research methods and covariates such as gait speed made it difficult to compare MFC values between studies for the purpose of defining design and maintenance standards for the outdoor built environment. Standardized methods for defining MFC and an emphasis on outdoor trials are needed in future studies.

Background: Falls are a major public health issue and tripping is the most common self-reported cause of outdoor falls. Minimum foot clearance (MFC) is a key parameter for identifying the probability of tripping. Optical motion capture systems are commonly used to measure MFC values; however, there is a need to identify alternative modalities that are better suited to collecting data in real-world settings. Objective: This is the first of a two-part scoping review. The objective of this paper is to identify and evaluate alternative measurement modalities to optical motion capture systems for measuring level-ground MFC. A companion paper identifies conditions that impact MFC and the range of MFC values individuals that these conditions exhibit. Methods: We searched four electronic databases, where peer-reviewed journals and conference papers reporting level-ground MFC characteristics were identified. The papers were screened by two independent reviewers for inclusion. The reporting was done in keeping with the PRISMA-ScR reporting guidelines. Results: From an initial search of 1571 papers, 17 papers were included in this paper. The identified technologies were inertial measurement units (IMUs) (n = 10), ultrasonic sensors (n = 2), infrared sensors (IR) (n = 2), optical proximity sensors (OPS) (n = 1), laser ranging sensors (n = 1), and ultra-wideband sensors (n = 1). From the papers, we extracted the sensor type, the analysis methods, the properties of the proposed system, and its accuracy and validation methods. Conclusions: The two most commonly used alternative modalities were IMUs and OPS. There was a lack of standardization among studies utilizing the same measurement modalities, as well as discrepancies in the methods used to assess performance. We provide a list of recommendations for future work to allow for more meaningful comparison between modalities as well as future research directions.