Most Anterior Cruciate Ligament Injuries in Professional Athletes Occur Without Contact to the Injured Knee: A Systematic Review of Video Analysis Studies.

PURPOSE: To systematically review studies utilizing video analyses to evaluate anterior cruciate ligament (ACL) injury mechanisms in athletes during sport to better understand risk factors and the potential for injury prevention. METHODS: A literature search was conducted in accordance with the 2020 PRISMA guidelines statement using SCOPUS, PubMed, Medline, and the Cochrane Central Register for Controlled Trials from database inception through June 2023. Inclusion criteria including studies reporting on ACL injury mechanisms occurring in athletes based on video analysis. Athlete demographics, injury mechanisms, position of the lower extremity, and activity at the time of injury were recorded. RESULTS: A total of 13 studies, consisting of 542 athletes, met inclusion criteria. The majority of athletes competed at the professional level (91%, n=495/542) with 79% (n=422/536) of athletes being male. The most common sports were soccer (33%, n=178/542) and American football (26%, n=140/542). The most common injury mechanism was non-contact in 42.9% (n=230/536) of athletes, followed by indirect contact (32.6%, n=175/536) and direct contact (22.4%, n=120/536). The most common position of injury was with a planted foot ( 91.7%, n=110/120), full or near full knee extension ( 84.4%, n=49/58), and axial loading ( 81.3%, n=87/107). Injuries commonly involved a deceleration/shift in momentum ( 50.4%, n=123/244) or pivoting maneuver ( 36.1%, n=77/213). At the time of injury, the knee commonly fell into valgus ( 76.8%, n=225/293) with associated internal (53.5%, n=46/86) or external tibiofemoral rotation ( 57.7%, n=101/175). CONCLUSIONS: The majority of ACL injuries, when evaluated by video analysis, involve professional athletes participating in soccer and American football. The most common injury mechanism occurred without contact with the knee in extension during a deceleration or momentum shift, with resultant valgus and rotational force across the knee. LEVEL OF EVIDENCE: IV; Systematic review of level IV studies.

Characteristics and Consequences of Falls in People Who Use Wheelchairs in Long-Term Care Settings.

OBJECTIVES: The purpose of this study was to understand the characteristics and consequences of falls in individuals using wheelchairs in long-term care settings. DESIGN: Observational analysis of real-world falls in long-term care. SETTING AND PARTICIPANTS: Residents using wheelchairs from 2 long-term care facilities in British Columbia, Canada (n = 32 participants, mean age = 84.7 years, 12 women). METHODS: Two raters used the validated Falls Video Analysis Questionnaire, adapted from the original version, to assess the causal, behavioral, and environmental aspects of falls from wheelchairs. RESULTS: A total of 58 wheelchair fall videos were identified out of 300 total videos that were collected from 2007 to 2014. Wheelchair falls were most often caused by incorrect transfer or shift of body weight (70.7%). Participants most often fell backward with 89.7% striking their pelvis. Individuals using wheelchairs had limited protective response, with only 10.3% demonstrating a step response. Improper brake position contributed to 67.2% of falls. No serious fall-related injuries were reported. CONCLUSIONS AND IMPLICATIONS: The findings highlight the unique nature of falls in older adults who use wheelchairs in long-term care settings. Overall, the results of this study support clinical practice and the critical need for developing specialized fall prevention and fall detection interventions for individuals who use wheelchairs in long-term care.

A 2D video-based assessment is associated with 3D biomechanical contributors to dynamic knee valgus in the coronal plane.

Introduction: Adolescent athletes involved in sports that involve cutting and landing maneuvers have an increased risk of anterior cruciate ligament (ACL) tears, highlighting the importance of identifying risky movement patterns such as dynamic knee valgus (DKV). Qualitative movement screenings have explored two-dimensional (2D) scoring criteria for DKV, however, there remains limited data on the validity of these screening tools. Determining a 2D scoring criterion for DKV that closely aligns with three-dimensional (3D) biomechanical measures will allow for the identification of poor knee position in adolescent athletes on a broad scale. The purpose of this study was to establish a 2D scoring criterion that corresponds to 3D biomechanical measures of DKV. Methods: A total of 41 adolescent female club volleyball athletes performed a three-task movement screen consisting of a single-leg squat (SLS), single-leg drop landing (SLDL), and double-leg vertical jump (DLVJ). A single rater scored 2D videos of each task using four criteria for poor knee position. A motion capture system was used to calculate 3D joint angles, including pelvic obliquity, hip adduction, knee abduction, ankle eversion, and foot progression angle. Receiver operating characteristic curves were created for each 2D scoring criterion to determine cut points for the presence of movement faults, and areas under the curve (AUC) were computed to describe the accuracy of each 2D criterion compared to 3D biomechanical data. Results: 3D measures indicated knee abduction angles between 2.4°-4.6° (SD 4.1°-4.3°) at the time point when the center of the knee joint was most medial during the three tasks. AUCs were between 0.62 and 0.93 across scoring items. The MEDIAL scoring item, defined as the knee joint positioned inside the medial border of the shoe, demonstrated the greatest association to components of DKV, with AUCs ranging from 0.67 to 0.93. Conclusion: The MEDIAL scoring criterion demonstrated the best performance in distinguishing components of DKV, specifically pelvic obliquity, hip adduction, ankle eversion, and foot progression. Along with the previously published scoring definitions for trunk-specific risk factors, the authors suggest that the MEDIAL criterion may be the most indicative of DKV, given an association with 3D biomechanical risk factors.

EndoViT: pretraining vision transformers on a large collection of endoscopic images.

PURPOSE: Automated endoscopy video analysis is essential for assisting surgeons during medical procedures, but it faces challenges due to complex surgical scenes and limited annotated data. Large-scale pretraining has shown great success in natural language processing and computer vision communities in recent years. These approaches reduce the need for annotated data, which is of great interest in the medical domain. In this work, we investigate endoscopy domain-specific self-supervised pretraining on large collections of data. METHODS: To this end, we first collect Endo700k, the largest publicly available corpus of endoscopic images, extracted from nine public Minimally Invasive Surgery (MIS) datasets. Endo700k comprises more than 700,000 images. Next, we introduce EndoViT, an endoscopy-pretrained Vision Transformer (ViT), and evaluate it on a diverse set of surgical downstream tasks. RESULTS: Our findings indicate that domain-specific pretraining with EndoViT yields notable advantages in complex downstream tasks. In the case of action triplet recognition, our approach outperforms ImageNet pretraining. In semantic segmentation, we surpass the state-of-the-art (SOTA) performance. These results demonstrate the effectiveness of our domain-specific pretraining approach in addressing the challenges of automated endoscopy video analysis. CONCLUSION: Our study contributes to the field of medical computer vision by showcasing the benefits of domain-specific large-scale self-supervised pretraining for vision transformers. We release both our code and pretrained models to facilitate further research in this direction: https://github.com/DominikBatic/EndoViT .

Inter-observer variation of head and foot point selection for subject height determination.

The purpose of this study was to examine factors affecting video analysts' decisions in marking the vertex of the head and foot point and corresponding inter-observer marking variances when conducting height analysis on individuals seen in video. Nineteen video analysts participated in an exercise at the 2022 Ontario Forensic Video Analysts' Association (OFVAA) conference where they were asked to mark the vertex of the head and a corresponding foot point of a "suspect" on extracted video frames in a variety of positions and with different headwear (no headwear, baseball cap, and hoodie). A height scale with discrete marking points located at the same positions as where the suspect was positioned was also included in a separate image set, offering a comparison to the suspect. Marked points for all analysts were overlayed onto the respective image frame for visual observations. Summary statistics were used for data interpretation. This study demonstrated that factors such as the suspect's proximity to the camera and suspect's headwear affected the variability and range of marking, which has a direct correlation to the estimated height of the suspect. In general, when the region to be marked was larger, the variability was also larger. This study also demonstrates that marking errors were significantly reduced when discrete marking locations were present such as on a height scale. The average percentage difference of height was most notable, approximately 3%, when the suspect was wearing a hoodie and was at a position closest to the camera. The range of the percentage difference was also the highest at this position, which was 10.6%. In comparison, the height scale had a maximum percent height difference of 0.6% at position D-5, the furthest position from the camera. The range at this location was approximately 2%, which was also the highest range value for the height scale. Future studies should consider suspect posture and look at how these errors may be minimized by examining the best locations to mark the head and foot points under different scenarios.

Lateral Ankle Joint Injuries in Indoor and Court Sports: A Systematic Video Analysis of 445 Nonconsecutive Case Series.

BACKGROUND: Lateral ankle sprains are one of the most common injuries in indoor and court sports. Self-reports and case studies have indicated that these injuries occur via both contact and noncontact injury mechanisms typically because of excessive inversion in combination with plantarflexion and adduction of the foot. Video-based documentation of the injury mechanism exists, but the number of cases reported in the literature is limited. PURPOSE: To retrieve and systematically analyze a large number of video-recorded lateral ankle injuries from indoor and court sports, as well as describe the injury mechanism, injury motion, and injury pattern across different sports. STUDY DESIGN: Cross-sectional study; Level of evidence, 3. METHODS: A total of 445 unique video-recorded lateral ankle sprain injuries were retrieved from indoor and court sports of broadcasted levels of competition. The videos were independently analyzed by 2 different reviewers. Outcomes included classification of the injury mechanism according to the International Olympic Committee consensus guidelines, primary and secondary motions of ankle joint distortion, and documentation of the fixation point (fulcrum) around which the foot rotates. RESULTS: Overall, 298 (67%) injuries were direct contact, 113 (25%) were noncontact, and 32 (7%) were indirect contact incidents. Direct contact injuries were especially prevalent in basketball (76%), handball (80%), and volleyball (85%), while noncontact injuries dominated in tennis and badminton (96% vs 95% across both). Inversion (65%) and internal rotation (33%) were the primary distortion motions, with the lateral forefoot (53%) and lateral midfoot (40%) serving as the main fulcrums. Landing on another player's foot was the leading cause of injury (n = 246; 55%), primarily characterized by inversion (79%) around a midfoot fulcrum (54%). The noncontact and indirect landings on floor (n = 144; 33%) were primarily characterized by a distortion around a forefoot fulcrum (69%). CONCLUSION: Two of 3 ankle sprains from online video platforms were direct contact injuries, with most involving landing on another player's foot. The distortion motion seems to be related to the injury mechanism and the fixation point between the foot and the floor. The injury mechanisms varied greatly between sports, and future studies should clearly differentiate and investigate the specific injury mechanisms.

Performance evaluation in cataract surgery with an ensemble of 2D-3D convolutional neural networks.

An important part of surgical training in ophthalmology is understanding how to proficiently perform cataract surgery. Operating skill in cataract surgery is typically assessed by real-time or video-based expert review using a rating scale. This is time-consuming, subjective and labour-intensive. A typical trainee graduates with over 100 complete surgeries, each of which requires review by the surgical educators. Due to the consistently repetitive nature of this task, it lends itself well to machine learning-based evaluation. Recent studies utilize deep learning models trained on tool motion trajectories obtained using additional equipment or robotic systems. However, the process of tool recognition by extracting frames from the videos to perform phase recognition followed by skill assessment is exhaustive. This project proposes a deep learning model for skill evaluation using raw surgery videos that is cost-effective and end-to-end trainable. An advanced ensemble of convolutional neural network models is leveraged to model technical skills in cataract surgeries and is evaluated using a large dataset comprising almost 200 surgical trials. The highest accuracy of 0.8494 is observed on the phacoemulsification step data. Our model yielded an average accuracy of 0.8200 and an average AUC score of 0.8800 for all four phase datasets of cataract surgery proving its robustness against different data. The proposed ensemble model with 2D and 3D convolutional neural networks demonstrated a promising result without using tool motion trajectories to evaluate surgery expertise.

Video-based analysis of the blink reflex in Parkinson's disease patients.

We developed a video-based tool to quantitatively assess the Glabellar Tap Reflex (GTR) in patients with idiopathic Parkinson's disease (iPD) as well as healthy age-matched participants. We also video-graphically assessed the effect of dopaminergic medication on the GTR in iPD patients, as well as the frequency and blinking duration of reflex and non-reflex blinks. The Glabellar Tap Reflex is a clinical sign seen in patients e.g. suffering from iPD. Reliable tools to quantify this sign are lacking. METHODS: We recorded the GTR in 11 iPD patients and 12 healthy controls (HC) with a consumer-grade camera at a framerate of at least 180 images/s. In these videos, reflex and non-reflex blinks were analyzed for blink count and blinking duration in an automated fashion. RESULTS: With our setup, the GTR can be extracted from high-framerate cameras using landmarks of the MediaPipe face algorithm. iPD patients did not habituate to the GTR; dopaminergic medication did not alter that response. iPD patients' non-reflex blinks were higher in frequency and higher in blinking duration (width at half prominence); dopaminergic medication decreased the median frequency (Before medication-HC: p < 0.001, After medication-HC: p = 0.0026) and decreased the median blinking duration (Before medication-HC: p = 0.8594, After medication-HC: p = 0.6943)-both in the direction of HC. CONCLUSION: We developed a quantitative, video-based tool to assess the GTR and other blinking-specific parameters in HC and iPD patients. Further studies could compare the video data to electromyogram (EMG) data for accuracy and comparability, as well as evaluate the specificity of the GTR in patients with other neurodegenerative disorders, in whom the GTR can also be present. SIGNIFICANCE: The video-based detection of the blinking parameters allows for unobtrusive measurement in patients, a safer and more comfortable option.

Comparing region of interest selection and whole-field analysis for measurement of ciliary beat frequency in high-speed video analysis.

BACKGROUND: Ciliary beat frequency (CBF) is crucial in mucociliary clearance. High-speed video analysis (HSVA) is commonly used to measure CBF but lacks standardization. We compared visual observation and computer-assisted calculation using fast Fourier transformation (FFT) in freshly collected bronchial ciliary epithelial cells and cultured cells. METHODS: Bronchial epithelial cells were obtained from 12 patients who required bronchoscopic examination. Eighty-five videos of ciliary movement of freshly collected and cultured cells were recorded and used to calculate CBF using manual observation, region of interest (ROI) selection, and whole-field analysis. RESULTS: CBF measured by the ROI selection method strongly correlated with that measured using manual observation, especially in freshly collected cells. However, 27.8% of the manual observation method values were doubled in the ROI selection method, probably because a round trip of cilia was calculated as two cycles and needed to be corrected to 1/2 value. Upon increasing the number of ROIs, the results of the ROI selection method came closer to that of WFA. CONCLUSIONS: Computer-assisted calculation using FFT can aid in measuring CBF; however, current methods require visual confirmation. Further automated evaluation techniques are needed to establish more standardized and generalized CBF measurement methods using HSVA.

A comprehensive two-dimensional scoring system to assess the single-leg squat task in football players.

BACKGROUND: The single-leg squat (SLS) is a safe and widespread functional test commonly performed in the mid-stages of rehabilitation after severe knee injuries. The use of reliable objective measures has been advocated to improve the quality of SLS assessment. The aim of this study was to describe a qualitative whole-body scoring system based on two-dimensional (2D) video analysis during SLS test and validate it against three-dimensional (3D) kinetics and kinematics. METHODS: Thirty-four competitive football (soccer) players performed a series of SLS tasks. 3D kinematics and kinetics were collected through infrared cameras, and 2D video analysis was performed through a scoring system with sub-scores ranging from 0/2 (non-adequate movement) to 2/2 (adequate movement) based on frontal and lateral planes objective measurements. 3D kinematics and kinetics were grouped according to the results of the 2D evaluation and compared through the analysis of variance (P < 0.05). RESULTS: Higher hip adduction, hip internalrotation, and knee valgus collapse were found in trials rated 0/2 or 1/2 compared with theone rated 2/2 in the limb stability score. Hip flexion and hip/knee moment ratio were lower in those scoring 0/2 comparedwith those scoring 2/2 in the movement strategy criterion. A low total score was associated with higherknee valgus collapse and lower hip/knee extensor moment ratio. Compensatory strategieswere found in frontal plane scores. CONCLUSIONS: The 2D scoring system described was strongly associated with kinematics and kinetics from gold-standard 3D motion capture and might represent a valid tool to describe the movement quality of an SLS task.

Assessment of reliability and quality of YouTube® exercise videos in people with rheumatoid arthritis.

BACKGROUND: YouTube®, one of the busiest video-sharing platforms, may lack sufficient or accurate information on health information and practices given the absence of a rigorous evaluation process. PURPOSE: This study aims to evaluate the reliability of information and quality of exercise videos on YouTube® for rheumatoid arthritis. METHODS: Video features and the source of the upload were sorted. Reliability of the information shared in the videos was evaluated using the mDISCERN too, and quality was assessed using the Global Quality Scale and JAMA scoring system. Videos were independently analyzed by two physiotherapists specialized in rheumatologic rehabilitation. Any bias was resolved by an independent third assessor. RESULTS: One hundred thirty-four exercise videos were included. A hundred and twenty-six (94%) of the 134 exercise videos were found to be useful, and 8 (6%) were misleading. The useful and misleading videos had similar numbers of views, likes, comments, and subscribers (p > .05), while video duration and time since upload were higher for useful videos (p < .05). Cohen's Kappa scores demonstrated that the level of agreement between the assessors were moderate (mDISCERN = 0.417, Global Quality Scale = 0.582, and JAMA = 0.555). There was a significant difference in JAMA scores (p = .013) between the sources of the videos. However, no significant difference was found in mDISCERN (p = .104) and Global Quality Scale (p = .128) scores. CONCLUSION: YouTube® exercise videos provide partially reliable and moderate-quality information for people with rheumatoid arthritis. However, patients should be cautious and not rely directly on YouTube® exercise videos. Instead, they should consult a physician or physical therapist for exercises.

Identifying Anterior Cruciate Ligament Injuries Through Automated Video Analysis of In-Game Motion Patterns.

Background: Failure to diagnose anterior cruciate ligament (ACL) injury during a game can delay adequate treatment and increase the risk of further injuries. Artificial intelligence (AI) has the potential to be an accurate, cost-efficient, and readily available diagnostic tool for ACL injury in in-game situations. Purpose: To develop an automated video analysis system that uses AI to identify biomechanical patterns associated with ACL injury and to evaluate whether the system can enhance the ability of orthopaedic and sports medicine specialists to identify ACL injuries on video. Study Design: Descriptive laboratory study. Methods: A total of 91 ACL injury and 38 control movement scenes from online available match recordings were analyzed. The videos were processed to identify and track athletes and to estimate their 3-dimensional (3D) poses. Geometric features, including knee flexion, knee and hip abduction, and foot and hip rotation, were extracted from the athletes' 3D poses. A recurrent neural network algorithm was trained to classify ACL injury, using these engineered features as its input. Analysis by 2 orthopaedic surgeons examined whether providing clinical experts with the reconstructed 3D poses and their derived signals could increase their diagnostic accuracy. Results: All AI models performed significantly better than chance. The best model, which used the long short-term memory network with engineered features, demonstrated decision interpretability and good performance (F1 score = 0.63 ± 0.01, area under the receiver operating characteristic curve = 0.88 ± 0.01). The analysis by the 2 orthopaedic surgeons demonstrated improved diagnostic accuracy for ACL injury recognition when provided with system data, resulting in a 0.08 increase in combined F1 scores. Conclusion: Our approach successfully reconstructed the 3D motion of athletes from a single-camera view and derived geometry-based biomechanical features from pose sequences. Our trained AI model was able to automatically detect ACL injuries with relatively good performance and prelabel and highlight regions of interest in video footage. Clinical Relevance: This study demonstrated the feasibility of using AI to automatically evaluate in-game video footage and identify dangerous motion patterns. Further research can explore the full potential of the biomechanical markers and use of the system by nonspecialists, potentially diminishing the rate of missed diagnosis and the detrimental outcomes that follow.

YouTube as a source of information for pressure injury: Quality assessment and content analysis.

We aimed to assess the quality and reliability of pressure injury-related videos uploaded on YouTube, analyse the sources and contents, and examine the correlation between video parameters. We searched YouTube using two keywords, "pressure ulcer" and "pressure sore", on August 20, 2022. We sorted the videos according to their number of views and included the top 100 videos for each keyword. The quality of videos was assessed using the Global Quality Scale (GQS), while their reliability was evaluated by the modified DISCERN (mDISCERN) tool. In addition, we evaluated the videos in which content was included, analysed the correlations and differences between GQS, mDISCERN, and video parameters. We initially found a total of 100 videos for each keyword and finally included and analysed 77 videos. The mean scores for the mDISCERN and GQS were 2.35 ± 0.98 and 3.09 ± 0.99, respectively. Both GQS and mDISCERN showed statistically significant correlations with each other (rho = 0.54, p < 0.0001*) and with the length of the videos, respectively (rho = 0.36, p = 0.001*), (rho = 0.29, p = 0.01*). Of the videos created by physicians, 8 (57.1%) included content related to treatment, while of the videos created by nonphysician health personnel, 22 (57.9%) included content related to prevention. Analysing whether there were differences in video parameters based on the sources, we observed significant differences between sources in GQS (p < 0.0001*), mDISCERN (p < 0.0001*), and video length (p = 0.001*). In the post-hoc analysis, videos uploaded by physicians or nonphysician health personnel showed higher quality and reliability than videos uploaded by other sources. Therefore, the results of this study could be useful for healthcare providers, as well as patients and caregivers, to search for high-quality and reliable YouTube videos related to pressure injury.

The influence of sex-division, experience, and pacing strategy on performance in the 2020 CrossFit® Open.

To observe workout pacing strategies and determine which best predicted performance, this retrospective study analyzed recorded efforts from a random selection of 160 high-ranking (top 10,000) men and women (n = 80 each) in the 2020 CrossFit® Open (CFO). Video recordings submitted to the official competition leaderboard for all five tests were analyzed to quantify overall test completion rates (and tie-break time for test 5 only) and within-test repetition completion rate (repetitions × sec-1) for each exercise, as well as the quantity of failed repetitions, break strategy (count and duration), and transition times. Each variable was aggregated into first-half, last-half, and total-test averages, slopes, and coefficient of variation; except on test 5 (total-test only). Spearman's rank correlation coefficients were calculated between test completion rates, each test's respective pacing variables, competitor demographics (height and body mass) and CFO experience (i.e., past participation, consecutive competitions, and ranks). Stepwise regression using significantly (p < 0.05) correlated variables produced two prediction models for test performance (best predictor only and best overall model within 8 variables) in a validation group (50% of valid efforts) and then cross-validated against remaining athletes. When no between-group differences were seen, data were combined and used to create the final prediction models for test 1 (r2adj = 0.64-0.96, SEE = 0.4-1.2 repetitions × sec-1), test 2 (r2adj = 0.28-0.85, SEE = 2.0-4.5 repetitions × sec-1), test 3 (r2adj = 0.49-0.81, SEE = 1.1-1.7 repetitions × sec-1), test 4 (r2adj = 0.63-0.78, SEE = 0.6-0.9 repetitions × sec-1), and test 5 (rate: r2adj = 0.71-0.84, SEE = 1.2-1.6 repetitions × sec-1; tie-break time: r2adj = 0.06-0.62, SEE = 1.4-2.3 min). Across the five 2020 CFO tests, the data suggested that repetition pace, breaking strategy, and/or consistency in completing calisthenic-gymnastics components (when prescribed) was most predictive of performance. However, their influence was affected by the complexity of prescribed resistance training exercises and their relative loads. Athletes should prioritize calisthenic-gymnastics components but divert attention to more complex resistance training exercises when prescribed at higher relative intensity loads. Neither previous competition experience nor sex-division altered the hierarchal importance of these considerations.

Remote Heart Rate Estimation Based on Transformer with Multi-Skip Connection Decoder: Method and Evaluation in the Wild.

Heart rate is an essential vital sign to evaluate human health. Remote heart monitoring using cheaply available devices has become a necessity in the twenty-first century to prevent any unfortunate situation caused by the hectic pace of life. In this paper, we propose a new method based on the transformer architecture with a multi-skip connection biLSTM decoder to estimate heart rate remotely from videos. Our method is based on the skin color variation caused by the change in blood volume in its surface. The presented heart rate estimation framework consists of three main steps: (1) the segmentation of the facial region of interest (ROI) based on the landmarks obtained by 3DDFA; (2) the extraction of the spatial and global features; and (3) the estimation of the heart rate value from the obtained features based on the proposed method. This paper investigates which feature extractor performs better by captioning the change in skin color related to the heart rate as well as the optimal number of frames needed to achieve better accuracy. Experiments were conducted using two publicly available datasets (LGI-PPGI and Vision for Vitals) and our own in-the-wild dataset (12 videos collected by four drivers). The experiments showed that our approach achieved better results than the previously published methods, making it the new state of the art on these datasets.

Identifying stress scores from gait biometrics captured using a camera: A cross-sectional study.

BACKGROUND: Stress is a critical risk factor for various health issues, but an objective, non-intrusive and effective measurement approach for stress has not yet been established. Gait, the pattern of movements in human locomotion, has been proven to be a valid behavioral indicator for recognizing various mental states in a convenient manner. RESEARCH QUESTION: This study aims to identify the severity of stress by assessing human gait recorded through an objective, non-intrusive measurement approach. METHODS: One hundred and fifty-two participants with an average age of 23 years old (SD = 1.07) were recruited. The Chinese version of the Perceived Stress Scale with 10 items (PSS-10) was used to assess participants' stress levels. The participants were then required to walk naturally while being recorded with a regular camera. A total of 1320 time-domain and 1152 frequency-domain gait features were extracted from the videos. The top 40 contributing features, confirmed by dimensionality reduction, were input into models consisting of four machine-learning regression algorithms (i.e., Gaussian Process Regressor, Linear Regression, Random Forest Regressor, and Support Vector regression), to assess stress levels. RESULTS: The models that combined time- and frequency-domain features performed best, with the lowest RMSE (4.972) and highest validation (r = 0.533). The Gaussian Process Regressor and Linear Regression outperformed the others. The greatest contribution to model performance was derived from gait features of the waist, hands, and legs. SIGNIFICANCE: The severity of stress can be accurately detected by machine learning models using two-dimensional (2D) video-based gait data. The machine learning models used for assessing perceived stress were reliable. Waist, hand, and leg movements were found to be critical indicator in detecting stress.

Forward dynamics computational modelling of a cyclist fall with the inclusion of protective response using deep learning-based human pose estimation.

Single bicycle crashes, i.e., falls and impacts not involving a collision with another road user, are a significantly underestimated road safety problem. The motions and behaviours of falling people, or fall kinematics, are often investigated in the injury biomechanics research field. Understanding the mechanics of a fall can help researchers develop better protective gear and safety measures to reduce the risk of injury. However, little is known about cyclist fall kinematics or dynamics. Therefore, in this study, a video analysis of cyclist falls is performed to investigate common kinematic forms and impact patterns. Furthermore, a pipeline involving deep learning-based human pose estimation and inverse kinematics optimisation is created for extracting human motion from real-world footage of falls to initialise forward dynamics computational human body models. A bracing active response is then optimised for using a genetic algorithm. This is then applied to a case study of a cyclist fall. The kinematic forms characterised in this study can be used to inform initial conditions for computational modelling and injury estimation in cyclist falls. Findings indicate that protective response is an important consideration in fall kinematics and dynamics, and should be included in computational modelling. Furthermore, the novel reconstruction pipeline proposed here can be applied more broadly for traumatic injury biomechanics tasks. The tool developed in this study is available at https://kevgildea.github.io/KinePose/.

Thermal preference of Culicoides biting midges in laboratory and semi-field settings.

Biting midges of the genus Culicoides (Diptera: Ceratopogonidae) are hematophagous insects, and some species can transmit a plethora of pathogens, e.g., bluetongue virus and African horse sickness virus, that mainly affect animals. The transmission of vector-borne pathogens is strongly temperature dependent, and recent studies pointed to the importance of including microclimatic data when modelling disease spread. However, little is known about the preferred temperature of biting midges. The present study addressed the thermal selection of field-caught Culicoides with two experiments. In a laboratory setup, sugar-fed or blood-fed midges were video tracked for 15 min while moving inside a 60 × 30 × 4 cm setup with a 15-25 °C temperature gradient. Culicoides spent over double the time in the coldest zone of the setup compared to the warmest one. This cold selection was significantly stronger for sugar-fed individuals. Calculated preferred temperatures were 18.3 °C and 18.9 °C for sugar-fed and blood-fed Culicoides, respectively. The effect of temperature on walking speed was significant but weak, indicating that their skewed distribution results from preference and not cold trapping. A second experiment consisted of a two-way-choice-setup, performed in a 90 × 45 × 45 cm net cage, placed outdoors in a sheltered environment. Two UV LED CDC traps were placed inside the setup, and a mean temperature difference of 2.2 °C was created between the two traps. Hundred-fifty Culicoides were released per experiment. Recapture rates were negatively correlated with ambient temperature and were on average three times higher in the cooled trap. The higher prevalence of biting midges in cooler environments influences fitness and ability to transmit pathogens and should be considered in models that predict Culicoides disease transmission.

The incidence and characteristics of heading in the 2019 FIFA Women's World Cup™.

INTRODUCTION: To quantify the incidence and characteristics of purposeful heading and other head impacts in professional women's football at the 2019 FIFA Women's World Cup™. METHODS: This cross-sectional cohort study analysed purposeful headers (uncontested and contested) and their characteristics (e.g. playing position, match situation, field location, and distance ball travelled), and other head impact events using video analysis. Total headers and head impact events, and incidence rate (IR) per 1000 match-hours were calculated for countries, positions, and other characteristics, such as location on the pitch. RESULTS: Purposeful headers accounted for 76% of all coded events (uncontested: 71%; contested: 29%), followed by attempted headers (21%), unintentional ball-head impacts (2%), and other head impacts (1%). Headers ranged from 0 to 22 per player, per match with a mean of 4.8 [±1.2]. Of all field positions, centrebacks had the highest heading rates and wingers the lowest. Strikers performed significantly more contested headers than any other position, and significantly less uncontested headers. Most headers occurred in the middle third (48%), from free game play (72%) and from long balls (>20 m) (68%). CONCLUSION: The findings of this study could assist the development of player heading risk profiles, sex-specific heading guidelines, and coaching practices.

Individual Activity Anomaly Estimation in Operating Rooms Based on Time-Sequential Prediction.

Surveillance videos of operating rooms have potential to benefit post-operative analysis and study. However, there is currently no effective method to extract useful information from the long and massive videos. As a step towards tackling this issue, we propose a novel method to recognize and evaluate individual activities using an anomaly estimation model based on time-sequential prediction. We verified the effectiveness of our method by comparing two time-sequential features: individual bounding boxes and body key points. Experiment results using actual surgery videos show that the bounding boxes are suitable for predicting and detecting regional movements, while the anomaly scores using key points can hardly be used to detect activities. As future work, we will be proceeding with extending our activity prediction for detecting unexpected and urgent events.