Video-based System Research Articles

Rapid and low-cost detection of saffron (Crocus sativus L.) adulteration using smartphone videos and spectral data fusion strategy

Saffron (Crocus sativus L.) is the most expensive spice in the world and is highly susceptible to economically motivated adulteration. Hyperspectral imaging (HSI) and near-infrared (NIR) spectroscopy are recently popular techniques for rapid, non-destructive detection of saffron adulteration. However, the high cost and measurement complexity of these techniques make them less suitable for consumer-level applications. In this work, a smartphone video-based system combined with chemometrics is used to detect saffron adulteration. A smartphone is utilised to capture short videos of pure and adulterated saffron powder samples illuminated by its colour-changing screen. The video frames are decomposed into RGB colour images and converted into spectral-like data. Partial least squares regression is used to model the relationship between data and saffron purity. The smartphone video-based system obtains a coefficient of determination of 0.9774 for prediction, which is comparable to reference techniques such as NIR spectroscopy and HSI. Moreover, it is efficient in presenting distribution maps related to saffron purity. In addition, when used in conjunction with the smartphone video-based system, HSI and NIR spectroscopy achieve higher performance without significantly increasing measurement cost and complexity. Such results suggest that the smartphone video-based system has the potential to be a viable primary screening and auxiliary tool for detecting saffron adulteration.

Agreement between an automated video-based system and tethered system to measure instantaneous swimming velocity

ABSTRACT Successful performance in competitive swimming requires a swimmer to maximise propulsion and minimise drag, which can be assessed using instantaneous swimming velocity. Many systems exist to quantify velocity, and therefore, it is important to understand the agreement between systems. This study examined the agreement between an automated video-based system and a tethered system to measure instantaneous velocity. Twenty-two competitive swimmers (state level or higher) completed 25 m of each stroke at maximal intensity. The tethered speedometer was attached to the swimmer’s waist, while videos of each trial were recorded. The swimmer’s head was then automatically tracked using proprietary software, and instantaneous velocity was determined from each system. Bland-Altman plots showed good agreement between the two systems in backstroke (95% Limits of Agreement (LOA): −0.24–0.26 m.s−1) and freestyle (95% LOA: −0.36–0.38 m.s−1) but poorer agreement in butterfly (95% LOA: −0.51–0.53 m.s−1) and breaststroke (95% LOA: −0.88–0.92 m.s−1). The root mean square error was higher in butterfly (0.27 m.s−1) and breaststroke (0.46 m.s−1) compared to backstroke (0.13 m.s−1) and freestyle (0.19 m.s−1). Results demonstrated that the two systems are comparable for measuring instantaneous swimming velocity; however, larger discrepancies are evident for butterfly and breaststroke.

AI Trainer : Video-Based Squat Analysis

This research proposes a video-based system for analyzing human squats and providing real-time feedback to improve posture. The system leverages MediaPipe, an open-source pose estimation library, to identify key body joints during squats. By calculating crucial joint angles (knee flexion, hip flexion, ankle dorsiflexion), the system assesses squat form against established biomechanical principles. Deviations from these principles trigger real-time feedback messages or visual cues to guide users towards optimal squat posture. The paper details the system architecture, with a client-side application performing pose estimation and feedback generation. The methodology outlines data collection with various squat variations, system development integrating MediaPipe, and evaluation through user testing with comparison to expert evaluations. Key features include real-time feedback and customizable thresholds for user adaptation. Potential applications encompass fitness training, physical therapy, and sports training. Finally, the paper explores future work possibilities like mobile integration, advanced feedback mechanisms, and machine learning for automatic threshold adjustments. This research offers a valuable tool for squat analysis, empowering users to achieve their fitness goals with proper form and reduced injury risk.

FastEval Parkinsonism: an instant deep learning–assisted video-based online system for Parkinsonian motor symptom evaluation

The Motor Disorder Society’s Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) is designed to assess bradykinesia, the cardinal symptoms of Parkinson’s disease (PD). However, it cannot capture the all-day variability of bradykinesia outside the clinical environment. Here, we introduce FastEval Parkinsonism (https://fastevalp.cmdm.tw/), a deep learning-driven video-based system, providing users to capture keypoints, estimate the severity, and summarize in a report. Leveraging 840 finger-tapping videos from 186 individuals (103 patients with Parkinson’s disease (PD), 24 participants with atypical parkinsonism (APD), 12 elderly with mild parkinsonism signs (MPS), and 47 healthy controls (HCs)), we employ a dilated convolution neural network with two data augmentation techniques. Our model achieves acceptable accuracies (AAC) of 88.0% and 81.5%. The frequency-intensity (FI) value of thumb-index finger distance was indicated as a pivotal hand parameter to quantify the performance. Our model also shows the usability for multi-angle videos, tested in an external database enrolling over 300 PD patients.

Investigation of Camera-Free Eye-Tracking Glasses Compared to a Video-Based System.

Technological advances in eye-tracking have resulted in lightweight, portable solutions that are capable of capturing eye movements beyond laboratory settings. Eye-tracking devices have typically relied on heavier, video-based systems to detect pupil and corneal reflections. Advances in mobile eye-tracking technology could facilitate research and its application in ecological settings; more traditional laboratory research methods are able to be modified and transferred to real-world scenarios. One recent technology, the AdHawk MindLink, introduced a novel camera-free system embedded in typical eyeglass frames. This paper evaluates the AdHawk MindLink by comparing the eye-tracking recordings with a research "gold standard", the EyeLink II. By concurrently capturing data from both eyes, we compare the capability of each eye tracker to quantify metrics from fixation, saccade, and smooth pursuit tasks-typical elements in eye movement research-across a sample of 13 adults. The MindLink system was capable of capturing fixation stability within a radius of less than 0.5∘, estimating horizontal saccade amplitudes with an accuracy of 0.04∘± 2.3∘, vertical saccade amplitudes with an accuracy of 0.32∘± 2.3∘, and smooth pursuit speeds with an accuracy of 0.5 to 3∘s, depending on the pursuit speed. While the performance of the MindLink system in measuring fixation stability, saccade amplitude, and smooth pursuit eye movements were slightly inferior to the video-based system, MindLink provides sufficient gaze-tracking capabilities for dynamic settings and experiments.

On Validating a Generic Video-Based Blink Detection System for Cognitive Load Detection

This work aims to validate using a video-based blink detection system for detecting changes in cognitive load. Participants completed a cognitive task with increasing levels of difficulty. Blink rate was recorded via both our video-based system and a scientific-grade eye-tracker. Results showed no differences in the blink rates recorded by the two systems. However, while strong evidence was found that the blink rate recorded through the eye-tracker increased under greater cognitive task demand, no such pattern was observed for the output from the video-based system. Our findings show that the overall performance of the video-based system was comparable with that of the eye-tracker's, but its sensitivity in detecting changes in cognitive load was inferior to that of its counterpart's.

Investigation of camera-free eye tracking glasses compared to a video-based system

Investigation of camera-free eye tracking glasses compared to a video-based system

Paper 41: Increased Knee Valgus and Hip Adduction Moments after Hamstring Autograft compared to Quadriceps Autograft in Adolescents 6 Months after Anterior Cruciate Ligament Reconstruction

Objectives: Evidence is limited related to the effects of autograft type on functional performance during gait lab-based functional testing after ACL reconstruction (ACLR), especially the quadriceps autograft. The goal of this study was to compare biomechanical outcomes as measured by 3D motion analysis during the drop vertical test between quadriceps autograft with bone block (QB), quadriceps autograft without bone block (Q), hamstring autograft (HS), and bone-patellar tendon-bone autograft (BTB) six-months postoperatively in an adolescent population. We hypothesized that the quadriceps groups would demonstrate decreased knee extension moments as compared to the hamstring autograft group without other significant differences. Methods: Patients aged 8-18 years who underwent primary ACLR were included. Patients with previous history of ipsilateral knee surgery were excluded. Kinematic and kinetic data collected during a drop vertical jump test in a formal gait lab using a computerized video-based system (Motion Analysis Corp. CORTEX software) was collected six months after ACLR and compared to the uninjured contralateral limb. The Kruskal Wallis test and subsequent pairwise comparisons were carried out for each studied parameter. Primary outcomes included transverse plane hip power, hip internal rotation moment, knee valgus angles, and knee extensor moments between the graft types. Secondary outcomes included hip abduction moment, knee valgus moment, and knee external rotation moment. Results: A total of 155 subjects’ charts were included for review, with 54, 40, 35, and 26 subjects in the HS, QB, Q, and BTB groups respectively. There were no significant differences in terms of age, sex, or affected leg (p > 0.1973). Athletic activity for each subject was also noted when appropriate, including soccer (37), basketball (34), and American football (30). The QB group had smaller knee extension moments maximums (-0.97 N*m/kg vs. -0.35 N*m/kg, p = 0.0323) and smaller knee extension moment averages (-0.12 N*m/kg vs. -0.03 N*m/kg, p = 0.0265) as compared to the HS group. The Q group demonstrated smaller knee extension moment averages (-0.13 N*m/kg vs. -0.03 N*m/kg, p = 0.459) as compared to the HS group. Analysis of secondary outcomes revealed that the HS group demonstrated larger knee valgus moments at initial contact (0.28 N*m/kg vs -0.35 N*m/kg, p = 0.0254) as compared to the Q group. The HS group demonstrated significantly larger hip adduction moment maximums (0.30 N*m/kg vs -0.04 N*m/kg, p = 0.0426) and smaller knee external rotation moment averages (-0.02 N*m/kg vs 0.02 N*m/kg, p = 0.0206) as compared to the QB group. No other secondary outcomes were statistically significant between any of the groups. Conclusions: The findings of this study suggest that the quadriceps autograft may confer improved knee coronal plane biomechanics in regard to ACL reinjury rates when compared to a hamstring autograft at 6 months after ACLR in adolescent patients performing a drop vertical jump. The hamstring autograft group was significantly associated with larger knee valgus moments at initial contact as compared to the quadriceps autograft without bone block and significantly larger hip adduction moments compared to the quadriceps autograft with bone block. Both quadriceps grafts were shown to demonstrate significantly decreased knee extension moments as compared to a hamstring autograft. Further long-term clinical studies may be useful in determining whether the apparent biomechanical differences seen between autograft types are correlated with reinjury rates.

Conventional video-based system for measuring the subtask speed of the Timed Up and Go Test in older adults: Validity and reliability study.

The Timed Up and Go Test (TUG) is a simple fall risk screening test that covers basic functional movement; thus, quantifying the subtask movement ability may provide a clinical utility. The video-based system allows individual's movement characteristics assessment. This study aimed to investigate the concurrent validity and test-retest reliability of the video-based system for assessing the movement speed of TUG subtasks among older adults. Twenty older adults participated in the validity study, whilst ten older adults participated in the reliability study. Participant's movement speed in each subtask of the TUG under comfortable and fast speed conditions over two sessions was measured. Pearson correlation coefficient was used to identify the validity of the video-based system compared to the motion analysis system. Intraclass correlation coefficient (ICC3,2) was used to determine the reliability of the video-based system. The Bland-Altman plots were used to quantify the agreement between the two measurement systems and two repeatable sessions. The validity analysis demonstrated a moderate to very high relationship in all TUG subtask movement speeds between the two systems under the comfortable speed (r = 0.672-0.906, p < 0.05) and a moderate to high relationship under the fast speed (r = 0.681-0.876, p < 0.05). The reliability of the video-based system was good to excellent for all subtask movement speeds in both the comfortable speed (ICCs = 0.851-0.967, p < 0.05) and fast speed (ICCs = 0.720-0.979, p < 0.05). The Bland-Altman analyses showed that almost all mean differences of the subtask speed of the TUG were close to zero, within 95% limits of agreement, and symmetrical distribution of scatter plots. The video-based system was a valid and reliable tool that may be useful in measuring the subtask movement speed of TUG among healthy older adults.

Emergence of Treadmill Running Ability and Quantitative Assessment of Gait Dynamics in Young Ts65Dn Mice: A Mouse Model for Down Syndrome.

Down syndrome (DS), which results from the complete or partial trisomy of chromosome 21 (trisomy-21), is the most common genetically defined cause of intellectual disability. Trisomy-21 also produces, or is associated with, many neurodevelopmental phenotypes and neurological comorbidities, including delays and deficits in fine and gross motor development. The Ts65Dn mouse is the most studied animal model for DS and displays the largest known subset of DS-like phenotypes. To date, however, only a small number of developmental phenotypes have been quantitatively defined in these animals. Here, we used a commercially available high-speed, video-based system to record and analyze the gait of Ts65Dn and euploid control mice. Longitudinal treadmill recordings were performed from p17 to p35. One of the main findings was the detection of genotype- and sex-dependent developmental delays in the emergence of consistent, progressive-intensity gait in Ts65Dn mice when compared to control mice. Gait dynamic analysis showed wider normalized front and hind stances in Ts65Dn mice compared to control mice, which may reflect deficits in dynamic postural balance. Ts65Dn mice also displayed statistically significant differences in the variability in several normalized gait measures, which were indicative of deficits in precise motor control in generating gait.

Static Laser Feedback Interferometry-Based Gaze Estimation for Wearable Glasses

Fast and robust gaze estimation is a key technology for wearable glasses, as it enables novel methods of user interaction as well as display enhancement applications, such as foveated rendering. State-of-the-art video-based systems lack a high update rate, integrateability, slippage robustness, and low power consumption. To overcome these limitations, we propose a model-based fusion algorithm to estimate gaze from multiple static laser feedback interferometry (LFI) sensors, which are capable of measuring distance toward the eye and the eye’s rotational velocity. The proposed system is ambient light robust and robust to glasses slippage. During evaluation, a gaze accuracy of 1.79° at an outstanding update rate of 1 kHz is achieved, while the sensors consume only a fraction of the power compared with the state-of-the-art video-based system.

A Low-Cost Video-Based System for Neurodegenerative Disease Detection by Mobility Test Analysis

The observation of mobility tests can greatly help neurodegenerative disease diagnosis. In particular, among the different mobility protocols, the sit-to-stand (StS) test has been recognized as very significant as its execution, both in terms of duration and postural evaluation, can indicate the presence of neurodegenerative diseases and their advancement level. The assessment of an StS test is usually done by physicians or specialized physiotherapists who observe the test and evaluate the execution. Thus, it mainly depends on the experience and expertise of the medical staff. In this paper, we propose an automatic visual system, based on a low-cost camera, that can be used to support medical staff for neurodegenerative disease diagnosis and also to support mobility evaluation processes in telehealthcare contexts. The visual system observes people while performing an StS test, then the recorded videos are processed to extract relevant features based on skeleton joints. Several machine learning approaches were applied and compared in order to distinguish people with neurodegenerative diseases from healthy subjects. Real experiments were carried out in two nursing homes. In light of these experiments, we propose the use of a quadratic SVM, which outperformed the other methods. The obtained results were promising. The designed system reached an accuracy of 95.2% demonstrating its effectiveness.

Relationship of Quantitative Measures of Jumping Performance with Gross Motor Development in Typically Developed Preschool Children.

Jumping is a key movement developing in the preschool period, but limited studies have reported the determinants of jumping performance and its relationship with gross motor development. This study aimed to determine the correlations among jumping performance, quantitative parameters of jumping, and gross motor development in preschool children. Twenty-one preschool children were recruited from one kindergarten, and fifteen of them with complete data were further analyzed. The quantitative parameters of standing long jump (SLJ) and standing vertical jump (SVJ) were measured using a video-based motion capture system. The gross motor development was measured using the Preschooler Gross Motor Quality Scale (PGMQ). The Spearman’s rho value and a linear regression model were used to determine the relationships among the jumping performance, the quantitative measures, and the total PGMQ scores. The results indicate that the jumping performances were significantly correlated with the takeoff velocity, which was predicted by trunk inclination before takeoff in SLJ and by the ranges of trunk inclination during jumping in SVJ. Regression analysis showed that the preschool children with higher normalized jump height had better gross motor development, and that the jump performance and the gross motor development were directly or indirectly predicted by the slope of the hip-to-ankle angle plot during pre-takeoff. In conclusion, this study identifies key components of jumping in jumping performance and gross motor development in preschool children for physical education.

Watch Your Back: Development and Usability of a Visual Feedback System for People with Low Back Pain – A Single Subject Usability Study

Abstract Background To find effective treatments for low back pain remains a major challenge in physical therapy. Numerous treatment methods currently exist, but the use of real-time visual feedback is a promising new therapeutic approach. The opportunities for its implementation and use require further research. Objective To develop a video-supported system for visual feedback and to test its usability. Method The prototype of a video-based system for visual feedback was developed and tested on a person with persistent unspecific lumbar back pain. The system was evaluated through an interview with the subject and the system usability scale (SUS). Results The subject was optimistic about the system’s functionality. He was astonished to see his own back while moving, which triggered positive emotions. He also found the apparatus comfortable to wear. The system achieved a value of 77.5 out of 100 on the SUS, which represents good usability. Conclusion By means of technical adjustments to avoid a delay in transmission and an exact positioning of the monitor, further improvements in the usability can be achieved. More research will be required to evaluate the efficacy of the developed system in the treatment of lumbar back pain.

Digital human factor management

Now culture of manage safety, health and wellbeing in the mining industry needs a scientifically sound new approach to the training process for professionals at all levels. Insufficient qualification of employees and managers in occupational safety, health and wellbeing has a negative impact on the occupational injury rate, and the current occupational safety, health and wellbeing management system does not allow to instill in employees compliance with the requirements for their own safety. To improve the quality of employees training in safe work practices and to ensure their health, safety and wellbeing in processes, a new training process has been developed and implemented based on a video-based continuous learning system. The innovation will make it possible to manage the human factor by improving employees competence, increasing the efficiency of the system OSH management, taking prompt corrective action, reducing injury rates and retaining human resources.

Benefits of Residual Aluminum Oxide for Sand Blasting Titanium Dental Implants: Osseointegration and Bactericidal Effects.

Objectives. The purpose of this work was to determine the influence of residual alumina after sand blasting treatment in titanium dental implants. This paper studied the effect of alumina on physico-chemical surface properties, such as: surface wettability, surface energy. Osseointegration and bacteria adhesion were determined in order to determine the effect of the abrasive particles. Materials and Methods. Three surfaces were studied: (1) as-received, (2) rough surface with residual alumina from sand blasting on the surface and (3) with the same roughness but without residual alumina. Roughness was determined by white light interferometer microscopy. Surface wettability was evaluated with a contact angle video-based system and the surface free energy by means of Owens and Wendt equation. Scanning electron microscopy equipped with microanalysis was used to study the morphology and determine the chemical composition of the surfaces. Bacteria (Lactobacillus salivarius and Streptococcus sanguinis) were cultured in each surface. In total, 110 dental implants were placed into the bone of eight minipigs in order to compare the osseointegration. The percentage of bone-to-implant contact was determined after 4 and 6 weeks of implantation with histometric analysis. Results. The surfaces with residual alumina presented a lower surface free energy than clean surfaces. The in vivo studies demonstrated that the residual alumina accelerated bone tissue growth at different implantation times, in relation to clean dental implants. In addition, residual alumina showed a bactericidal effect by decreasing the quantity of bacteria adhering to the titanium. Conclusions. It is possible to verify the benefits that the alumina (percentages around 8% in weight) produces on the surface of titanium dental implants. Clinical relevance. Clinicians should be aware of the benefits of sand-blasted alumina due to the physico-chemical surface changes demonstrated in in vivo tests.

Remote Evaluation of Parkinson's Disease Using a Conventional Webcam and Artificial Intelligence.

Objective: This study aimed to prove the concept of a new optical video-based system to measure Parkinson's disease (PD) remotely using an accessible standard webcam.Methods: We consecutively enrolled a cohort of 42 patients with PD and healthy subjects (HSs). The participants were recorded performing MDS-UPDRS III bradykinesia upper limb tasks with a computer webcam. The video frames were processed using the artificial intelligence algorithms tracking the movements of the hands. The video extracted features were correlated with clinical rating using the Movement Disorder Society revision of the Unified Parkinson's Disease Rating Scale and inertial measurement units (IMUs). The developed classifiers were validated on an independent dataset.Results: We found significant differences in the motor performance of the patients with PD and HSs in all the bradykinesia upper limb motor tasks. The best performing classifiers were unilateral finger tapping and hand movement speed. The model correlated both with the IMUs for quantitative assessment of motor function and the clinical scales, hence demonstrating concurrent validity with the existing methods.Conclusions: We present here the proof-of-concept of a novel webcam-based technology to remotely detect the parkinsonian features using artificial intelligence. This method has preliminarily achieved a very high diagnostic accuracy and could be easily expanded to other disease manifestations to support PD management.

Effect of the Acid-Etching on Grit-Blasted Dental Implants to Improve Osseointegration: Histomorphometric Analysis of the Bone-Implant Contact in the Rabbit Tibia Model

Previous studies have shown that the most reliable way to evaluate the success of an implant is by bone-to-implant contact (BIC). Recent techniques allow modifications to the implant surface that improve mechanical and biological characteristics, and also upgrade osseointegration. Objective: The aim was to evaluate the osseointegration in rabbit tibia of two different titanium dental implant surfaces: shot-blasted with Al2O3 (SB) and the same treatment with an acid-etching by immersion for 15 s in HCl/H2SO4 (SB + AE). Material and methods: Roughness parameters (Ra, Rt, and Rz) were determined by white light interferometer microscopy. Surface wettability was evaluated with a contact angle video-based system using water, di-iodomethane, and formamide. Surface free energy was determined by means of Owens and Wendt equations. Scanning electron microscopy equipped with X-ray microanalysis was used to study the morphology and determine the chemical composition of the surfaces. Twenty-four grade 4 titanium dental implants (Essential Klockner®) were implanted in the rabbit’s tibia, 12 for each surface treatment, using six rabbits. Six weeks later the rabbits were sacrificed and the implants were sent for histologic analysis. Resonance frequency analysis (RFA) was recorded both at the time of surgery and the end of the research with each device (Osstell Mentor and Osstell ISQ). Results: The roughness measurements between the two treatments did not show statistically significant differences. However, the effect of the acid etching made the surface slightly more hydrophilic (decreasing contact angle from 74.7 for SB to 64.3 for SB + AE) and it presented a higher surface energy. The bone-to-implant contact ratio (BIC %) showed a similar tendency, with 55.18 ± 15.67 and 59.9 ± 13.15 for SB and SB + AE implants, respectively. After 6 weeks of healing, the SB + AE showed an implant stability quotient (ISQ) value of 76 ± 4.47 and the shot-blasted one an ISQ value of 75.83 ± 8.44 (no statistically significant difference). Implants with different surface properties had distinctive forms of behavior regarding osseointegration. Furthermore, the Osstell system was an invasive and reliable method to measure implant stability. Conclusion: Both surfaces of implants studied showed high osseointegration. The SB and SB + AE implants used in our study had similar behavior both in terms of BIC values and RFA. The RFA systems in Osstell Mentor and Osstell ISQ confirmed nearly perfect reproducibility and repeatability.

Vision-based human joint angular velocity estimation during squat and walking on a treadmill actions.

Elderly health monitoring, rehabilitation training, and sport supervision could benefit from continuous assessment of joint angle, and angular velocity to identify the joint movement patterns. However, most of the measurement systems are designed based on special kinematic sensors to estimate angular velocities. The study aims to measure the lower limb joint angular velocity based on a 2D vision camera system during squat and walking on treadmill action using deep convolution neural network (CNN) architecture. Experiments were conducted on 12 healthy adults, and six digital cameras were used to capture the videos of the participant actions in lateral and frontal view. The normalized cross-correlation (Ccnorm) analysis was performed to obtain a degree of symmetry of the ground truth and estimated angular velocity waveform patterns. Mean Ccnorm for angular velocity estimation by deep CNN model has higher than 0.90 in walking on the treadmill and 0.89 in squat action. Furthermore, joint-wise angular velocities at the hip, knee, and ankle joints were observed and compared. The proposed system gets higher estimation performance under the lateral view and the frontal view of the camera. This study potentially eliminates the requirement of wearable sensors and proves the applicability of using video-based system to measure joint angular velocities during squat and walking on a treadmill actions.

Video-Based Inpatient Fall Risk Assessment: A Case Study.

Inpatient falls are a serious safety issue in hospitals and healthcare facilities. Recent advances in video analytics for patient monitoring provide a non-intrusive avenue to reduce this risk through continuous activity monitoring. However, in- bed fall risk assessment systems have received less attention in the literature. The majority of prior studies have focused on fall event detection, and do not consider the circumstances that may indicate an imminent inpatient fall. Here, we propose a video-based system that can monitor the risk of a patient falling, and alert staff of unsafe behaviour to help prevent falls before they occur. We propose an approach that leverages recent advances in human localisation and skeleton pose estimation to extract spatial features from video frames recorded in a simulated environment. We demonstrate that body positions can be effectively recognised and provide useful evidence for fall risk assessment. This work highlights the benefits of video-based models for analysing behaviours of interest, and demonstrates how such a system could enable sufficient lead time for healthcare professionals to respond and address patient needs, which is necessary for the development of fall intervention programs.