Partitioning variability in animal behavioral videos using semi-supervised variational autoencoders.

Recent neuroscience studies demonstrate that a deeper understanding of brain function requires a deeper understanding of behavior. Detailed behavioral measurements are now often collected using video cameras, resulting in an increased need for computer vision algorithms that extract useful information from video data. Here we introduce a new video analysis tool that combines the output of supervised pose estimation algorithms (e.g. DeepLabCut) with unsupervised dimensionality reduction methods to produce interpretable, low-dimensional representations of behavioral videos that extract more information than pose estimates alone. We demonstrate this tool by extracting interpretable behavioral features from videos of three different head-fixed mouse preparations, as well as a freely moving mouse in an open field arena, and show how these interpretable features can facilitate downstream behavioral and neural analyses. We also show how the behavioral features produced by our model improve the precision and interpretation of these downstream analyses compared to using the outputs of either fully supervised or fully unsupervised methods alone.

A Needs Assessment of Video-based Education Resources Among General Surgery Residents.

BACKGROUND: Video-based education (VBE) is an effective tool for knowledge and skill acquisition for medical students, but its utility is less clear for resident physicians. We sought to determine how to incorporate VBE into a general surgery resident operative curriculum. METHODS: We conducted a single-institution, survey-based needs assessment of general surgery residents to determine desired content and format of an operative VBE module. RESULTS: The response rate was 84% (53/63), with 66% senior (postgraduate year ≥3) resident respondents. VBE was the most commonly cited resource that residents used to prepare for an operation (93%) compared with surgical textbooks (89%) and text-based website content (57%). Junior residents were more likely to utilize text-based website content than senior residents (P < 0.01). The three most important operative video components were accuracy, length, and cost. Senior residents significantly preferred videos that were peer-reviewed (P < 0.05) and featured attending surgeons whom they knew (P = 0.03). A majority of residents (59%) believed 5-10 min is the ideal length of an operative video. Across all postgraduate year levels, residents indicated that detailed instruction of each operative step was the most important content of a VBE module. Senior residents believed that the overall indications and details of each step of the operation were the most important contents of VBE for a junior resident. CONCLUSIONS: At this institution, general surgery residents preferentially use VBE resources for operative preparation. A centralized, standardized operative resource would likely improve resident studying efficiency, but would require personalized learning options to work for both junior and senior surgery residents.

The COVID-19 outbreak and approaches to performing EEG in Europe.

The coronavirus SARS-CoV-2 disease (COVID-19) pandemic affects availability and performance of neurophysiological diagnostic methods, including EEG. Our objective was to outline the current situation regarding EEG-based investigations across Europe. A web-based survey was distributed to centres within the European Reference Network on rare and complex epilepsies (ERN EpiCARE). Responses were collected between April 9 and May 15, 2020. Results were analysed with Microsoft Excel, Python Pandas and SciPy. Representants from 47 EpiCARE centres from 22 countries completed the survey. At the time of completing the survey, inpatient video-EEGs had been stopped or restricted in most centres (61.7% vs. 36.2% for adults, and 38.3% vs. 53.2% for children). Invasive investigations and epilepsy surgery were similarly affected. Acute EEGs continued to be performed, while indications for outpatient EEGs were limited and COVID-19 triage put in place. The strictness of measures varied according to extent of the outbreak in a given country. The results indicate a profound impact of COVID-19 on neurophysiological diagnostics, especially inpatient video-EEGs, invasive investigations, and epilepsy surgery. The COVID-19 pandemic may hamper care for patients in need of EEG-based investigations, particularly patients with seizure disorders. ERN EpiCARE will work on recommendations on how to rapidly adapt to such situations in order to alleviate consequences for our patients.

Evaluating the Quality of Overactive Bladder Patient Education Material on YouTube: A Pilot Study Using the Patient Education Materials Assessment Tool.

OBJECTIVES: To assess the characteristics and quality of patient education content for overactive bladder (OAB) on YouTube. METHODS: We searched YouTube on September 10, 2019 and reviewed the top 25 search results for "overactive bladder." For comparison, we reviewed 3 Urology Care Foundation (UCF) OAB education videos on YouTube. Videos were scored using the Agency for Healthcare Research and Quality's Patient Education Materials Assessment Tool (PEMAT) for Audiovisual materials, which is subdivided into understandability and actionability domains, each with a maximum score of 100%. RESULTS: The mean length was 9.5 (0.9-34) minutes and the mean number of views was 151,382 (127-2,032,441). Of the top 25 search results, the mean PEMAT understandability score was 74% ± 16% (range 41%-100%). The mean PEMAT actionability score was 57% ± 39% (range 0%-100%). Seventy-two percent of the reviewed videos featured a physician. In comparison, the 3 UCF videos had an average of 5840 views, 3 minutes length, and PEMAT understandability and actionability scores of 93% and 100%. All UCF OAB videos featured urologists. CONCLUSION: The quality of OAB patient education materials on YouTube varies significantly in understandability and actionability. Although the UCF videos scored much higher on the PEMAT than our reviewed videos, they did not appear within the top 25 search results for OAB. There is great opportunity for the American Urological Association (AUA)/UCF and Society of Urodynamics, Female Pelvic Medicine & Urogenital Reconstruction (SUFU) to increase its outreach to patients on platforms like YouTube.

'Both useful in their own way': Video podcasts and typed solutions as feedback on undergraduate pharmaceutical calculations skills assessment.

INTRODUCTION: Accurately completing pharmaceutical calculations is a core professional skill for pharmacists. To date, few studies have focused on to providing feedback on calculations, or the role of technology in feedback provision. This study aimed to develop a theory-informed video podcast-based method of providing formative feedback and evaluate student perceptions. METHODS: First-year pharmacy students (n = 53) completed a formative pharmaceutical calculations assessment. Two forms of feedback were designed and evaluated; typed solutions (traditional format commonly used/seen in textbooks) and video podcasts informed by instructional design theory (novel format). RESULTS: A survey was completed by 70% (37/53) of students. Specific features of video podcasts reported useful included hearing reasoning, and the ability to pause and rewind. Most (76%) reported positive attitudes towards video podcasts, considered them useful (75%) and helpful for learning (79%). A comparable number (76% and 71% respectively) felt positively about typed solutions. The majority (51%) preferred to receive both types rather than podcasts alone (24%), or typed solutions alone (8%). Free-text responses indicated both were used in different ways; typed solutions for quick verification and video podcasts for clarification. CONCLUSIONS: Video podcasts appear to be a potentially helpful additional method of delivering feedback that afford specific advantages. They can be readily developed by faculty with minimal expense/difficulty. However, as respondents indicated that they used both kinds of feedback in different ways to support their learning, and indicated a preference to receive both types, they should be considered an addition rather than replacement for typed solutions.

Adopting YouTube to Promote Health: Analysis of State Health Departments.

INTRODUCTION: We describe videos posted to the YouTube video-sharing Web site by US state health departments (SHDs) and associated institutional factors. METHODS: YouTube channels from SHDs were identified, their data retrieved, and their videos saved to a playlist on January 10, 2016. Ten randomly sampled videos from each channel were manually coded for topics. The 2012 Association of State and Territorial Health Officials profile survey was used to obtain information on staff, expenditure, and top 5 priorities for each SHD. Descriptive statistics and univariable regression were conducted. RESULTS: Forty-three SHDs had YouTube channels. Together, all SHDs posted 3957 videos, accumulated 12,151,720 views, and gained 6302 subscribers. In total, 415 videos were manually coded. Information about the agency (17.6%), communicable diseases (12.5%), and mother/infant health (8.9%) comprised the largest share of topics. No statistically significant association was observed between the log-transformed number of videos posted on an SHD's YouTube channel and any of the explanatory variables of SHD staffing and expenditure in 2011. The number of full-time employees (r = 0.34, p = 0.03), number of epidemiologists and biostatisticians (r = 0.41, p = 0.01), and 2011 total year expenditure (r = 0.38, p = 0.02) were positively correlated with the log-transformed number of views per YouTube video posted by SHDs. No meaningful patterns of statistical association were observed between the percentage of expenditure on a specific program area and the topics of videos. CONCLUSION: Most SHDs are using YouTube, which provides a unique opportunity for SHDs to disseminate health messages.

Keyframe extraction from laparoscopic videos based on visual saliency detection.

BACKGROUND AND OBJECTIVE: Laparoscopic surgery offers the potential for video recording of the operation, which is important for technique evaluation, cognitive training, patient briefing and documentation. An effective way for video content representation is to extract a limited number of keyframes with semantic information. In this paper we present a novel method for keyframe extraction from individual shots of the operational video. METHODS: The laparoscopic video was first segmented into video shots using an objectness model, which was trained to capture significant changes in the endoscope field of view. Each frame of a shot was then decomposed into three saliency maps in order to model the preference of human vision to regions with higher differentiation with respect to color, motion and texture. The accumulated responses from each map provided a 3D time series of saliency variation across the shot. The time series was modeled as a multivariate autoregressive process with hidden Markov states (HMMAR model). This approach allowed the temporal segmentation of the shot into a predefined number of states. A representative keyframe was extracted from each state based on the highest state-conditional probability of the corresponding saliency vector. RESULTS: Our method was tested on 168 video shots extracted from various laparoscopic cholecystectomy operations from the publicly available Cholec80 dataset. Four state-of-the-art methodologies were used for comparison. The evaluation was based on two assessment metrics: Color Consistency Score (CCS), which measures the color distance between the ground truth (GT) and the closest keyframe, and Temporal Consistency Score (TCS), which considers the temporal proximity between GT and extracted keyframes. About 81% of the extracted keyframes matched the color content of the GT keyframes, compared to 77% yielded by the second-best method. The TCS of the proposed and the second-best method was close to 1.9 and 1.4 respectively. CONCLUSIONS: Our results demonstrated that the proposed method yields superior performance in terms of content and temporal consistency to the ground truth. The extracted keyframes provided highly semantic information that may be used for various applications related to surgical video content representation, such as workflow analysis, video summarization and retrieval.

Electronic and Social Media-based Radiology Learning Initiative: Development, Implementation, Viewership Trends, and Assessment at 1 Year.

RATIONALE AND OBJECTIVES: We report the development of a new "Case of the Day" (COTD) educational initiative using email, social media (SoMe), and a website to disseminate content, as well as its trends in viewership and assessment of utility for the first year of implementation. MATERIALS/METHODS: Using an image-rich format, a new unknown case was disseminated to radiology trainees and attendings at our institution by email twice per week, including history, salient images, and follow-up questions. Simultaneously, content was externally disseminated on Twitter and a publicly viewable departmental website. On subsequent days, the answer was posted via email, Twitter, and website in the form of a brief YouTube video lecture. Viewership data were collected over the first 12 months (July 1, 2016 to June 30, 2017), and an anonymous survey of participants was performed. RESULTS: Sixty-five COTDs had complete viewership data and were included in our analysis, yielding 4911 "case" email views (mean = 76), 3798 "answer" email views (mean = 58), 68,034 "case" Twitter impressions (mean = 1047), 75,724 "answer" Twitter impressions (mean = 1164), 5465 "case" Twitter engagements (mean = 84), and 5307 "answer" Twitter engagements (mean = 82). COTD YouTube video lectures garnered 3657 views (mean = 61) amounting to 10,358 minutes of total viewing time. Viewers were very satisfied with COTD quality, with 97% (n = 63) reporting the quality as "good" or "excellent." CONCLUSIONS: Email and SoMe can serve as effective tools for disseminating radiology educational content. SoMe offers substantial external visibility and branding potential for programs.

Usability of American Sign Language Videos for Presenting Mathematics Assessment Content.

There is considerable interest in determining whether high-quality American Sign Language videos can be used as an accommodation in tests of mathematics at both K-12 and postsecondary levels; and in learning more about the usability (e.g., comprehensibility) of ASL videos with two different types of signers - avatar (animated figure) and human. The researchers describe the results of administering each of nine pre-college mathematics items in both avatar and human versions to each of 31 Deaf participants with high school and post-high school backgrounds. This study differed from earlier studies by obliging the participants to rely on the ASL videos to answer the items. While participants preferred the human version over the avatar version (apparently due largely to the better expressiveness and fluency of the human), there was no discernible relationship between mathematics performance and signed version.

Reduction in Fall Rate in Dementia Managed Care Through Video Incident Review: Pilot Study.

BACKGROUND: Falls of individuals with dementia are frequent, dangerous, and costly. Early detection and access to the history of a fall is crucial for efficient care and secondary prevention in cognitively impaired individuals. However, most falls remain unwitnessed events. Furthermore, understanding why and how a fall occurred is a challenge. Video capture and secure transmission of real-world falls thus stands as a promising assistive tool. OBJECTIVE: The objective of this study was to analyze how continuous video monitoring and review of falls of individuals with dementia can support better quality of care. METHODS: A pilot observational study (July-September 2016) was carried out in a Californian memory care facility. Falls were video-captured (24×7), thanks to 43 wall-mounted cameras (deployed in all common areas and in 10 out of 40 private bedrooms of consenting residents and families). Video review was provided to facility staff, thanks to a customized mobile device app. The outcome measures were the count of residents' falls happening in the video-covered areas, the acceptability of video recording, the analysis of video review, and video replay possibilities for care practice. RESULTS: Over 3 months, 16 falls were video-captured. A drop in fall rate was observed in the last month of the study. Acceptability was good. Video review enabled screening for the severity of falls and fall-related injuries. Video replay enabled identifying cognitive-behavioral deficiencies and environmental circumstances contributing to the fall. This allowed for secondary prevention in high-risk multi-faller individuals and for updated facility care policies regarding a safer living environment for all residents. CONCLUSIONS: Video monitoring offers high potential to support conventional care in memory care facilities.

An exploratory assessment of weight loss videos on YouTube™.

OBJECTIVE: To describe the variation in contents between different video sources on YouTube ™ regarding weight loss. STUDY DESIGN: The 100 most widely viewed weight loss videos on YouTube™ were selected for manual coding. METHODS: Videos were manually coded based on sources and contents, we also identify themes for each of the videos. Univariate logistic regression was used to assess difference in contents between videos from different sources. RESULTS: After excluding two videos, a sample of 98 videos remained. Among them, there were 74 consumer-generated videos, 12 news videos, and 12 from commercial television and companies. Collectively, the 98 videos were viewed more than 365 million times. The odds of mentioning ongoing lifestyle that includes long-term changes in daily eating was 90% less in commercial and company videos compared to consumer videos (odds ratio [OR] = 0.10, 95% confidence interval [CI] 0.01-0.85, P = 0.03). The odds of showing a female talking in the video about weight loss was 81% less in news videos compared to consumer videos (OR = 0.19, 95% CI 0.05-0.68, P = 0.01). In addition, commercial and company videos were found to be 18 times as likely to carry a theme on workout as consumer videos (OR = 18.13, 95% CI 3.60-91.22, P < 0.01). CONCLUSION: Public health professionals need to be more creative to make their videos popular if they want to reach a wide audience.

Social media and scientific research are complementary-YouTube and shrikes as a case study.

Fascination with animals and their behaviour is one the most prominent patterns persisting in all human cultures. During the last decades, however, technological development and public access to the Internet have increased the speed and the extent of information sharing at an unprecedented rate, in some cases even challenging the traditional methods used in science. In order to understand the extent of this influence, we focused on the behaviour of shrikes. Shrikes are an enigmatic group of songbirds with a unique behaviour of impaling prey. We employed an extensive Internet search on YouTube (YT), a very popular and increasingly important source of information worldwide, for videos recording shrikes. Our analyses revealed that the number of shrike videos on YT is strongly positively correlated with classical knowledge on shrikes from books and scientific articles. Our results also suggest that in some cases YT may provide an alternative source of information on shrike ecology and behaviour. YT videos may thus provide new insights into the study of certain species or subjects and help identify gaps in ecological studies, especially in poorly studied species.

Agreement between video footage and fall incident reports on the circumstances of falls in long-term care.

OBJECTIVE: Incident reports guide fall prevention efforts in long-term care (LTC) facilities, often based on descriptions of how falls occurred. The validity of these reports is poorly understood. We examined agreement on fall characteristics between fall incident reports and analysis of video footage of real-life falls in LTC. DESIGN/SETTING/PARTICIPANTS: Video capture of 863 falls (by 309 individuals) over 6 years in common areas of 2 LTC facilities in British Columbia. MEASUREMENTS: We reviewed each fall video with a previously validated questionnaire to determine the cause of imbalance leading to the fall, activity at the time of falling, and use of mobility aids. These data were compared with corresponding information recorded by staff on fall incident reports. RESULTS: There was agreement between video analysis and incident reports on the cause of imbalance in 45.5% of falls (kappa = 0.25), on activity at time of falling in 45.1% of falls (kappa = 0.22), and on use of mobility aids in 79.5% of falls (kappa = 0.59). When compared with video analysis, incident reports overreported falls due to slips, and falling while rising and while using a wheelchair or walker. Incident reports also underreported falls due to hit/bump and loss-of-support, and falling while standing and sitting down. CONCLUSION: In more than 50% of falls, we found discrepancies between fall incident reports and analysis of video footage on the cause of imbalance and activity while falling. Emerging technologies incorporating video capture or wearable sensors should improve our ability to understand the mechanisms and improve the prevention of falls in LTC.

Frequency and factors associated with use of videolaryngostroboscopy in voice disorder assessment.

OBJECTIVES/HYPOTHESIS: Videolaryngostroboscopy (VLS) is considered an important diagnostic tool in the evaluation of patients with laryngeal/voice disorders. We evaluated the frequency of, diagnoses associated with, and factors related to VLS use in the evaluation of outpatients with laryngeal/voice disorders. STUDY DESIGN: Retrospective analysis of a large, national administrative U.S. claims database. METHODS: Patients with a laryngeal disorder based on International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes from January 1, 2004, to December 31, 2008, seen by an otolaryngologist were included. Patient age, gender, geographic region, laryngeal diagnosis, comorbid conditions, and whether laryngoscopy or VLS was performed during the outpatient otolaryngology visit were collected. RESULTS: A total of 168,444 unique patients saw an otolaryngologist for 272,112 outpatient visits. Of those, 6.2% of outpatient otolaryngology visits had a VLS performed. Patient age was related to VLS use, with lower odds in the elderly (≥ 65 years of age) and those 0 to 17 years of age. Geographic variation was noted, with higher odds of VLS use in urban versus rural areas and greater odds in the Northeast versus the South. Laryngeal diagnosis was associated with VLS use with greatest odds for patients with multiple diagnoses, vocal fold paralysis, and paresis-followed by nonspecific dysphonia and benign vocal fold/laryngeal pathology and followed by acute and chronic laryngitis and laryngeal cancer. Patients with gastro-esophageal reflux (GER) had greater odds of VLS use than patients without GER. CONCLUSIONS: VLS was used in 6.2% of outpatient otolaryngology outpatient visits; and its use was influenced by multiple factors.

Kaiser Permanente Northern California: current experiences with internet, mobile, and video technologies.

The US health care system has been slow to adopt Internet, mobile, and video technologies, which have the capability to engage patients in their own care, increase patients' access to providers, and possibly improve the quality of care while reducing costs. Nevertheless, there are some pockets of progress, including Kaiser Permanente Northern California (KPNC). In 2008 KPNC implemented an inpatient and ambulatory care electronic health record system for its 3.4 million members and developed a suite of patient-friendly Internet, mobile, and video tools. KPNC has achieved many successes. For example, the number of virtual "visits" grew from 4.1 million in 2008 to an estimated 10.5 million in 2013. This article describes KPNC's experience with Internet, mobile, and video technologies and the obstacles faced by other health care providers interested in embracing them. The obstacles include the predominant fee-for-service payment model, which does not reimburse for virtual visits; the considerable investment needed to deploy these technologies; and physician buy-in.

Functional lesions in dysphagia due to acute stroke: discordance between abnormal findings of bedside swallowing assessment and aspiration on videofluorography.

INTRODUCTION: Bedside swallowing assessments are often used to assess dysphagia. However, in some patients, aspiration pneumonia occurs without any problems on bedside swallowing assessments and some patients do not suffer aspiration pneumonia despite abnormal results of bedside swallowing assessments in acute stroke. To detect the differences of lesions related to bedside swallowing assessment abnormality and aspiration, we investigated swallowing-related functional lesions in terms of cerebral blood flow in patients with dysphagia after stroke. METHODS: The study included 50 acute stroke patients who underwent bedside swallowing assessments and videofluorography as well as single-photon emission computed tomography (CT) at approximately the same time. Bedside swallowing assessments included repetitive saliva swallowing test and modified water swallowing test as dry and wet swallowing tasks. The presence or absence of aspiration was assessed using videofluorography. We divided patients into three subgroups based on the outcomes of the bedside swallowing assessments and presence or absence of aspiration. Statistical image analysis was performed using single-photon emission CT to determine their relationship with bedside swallowing assessments and videofluorography results. RESULTS: Twenty-seven (54.0%) and 28 (56.0%) patients had abnormal repetitive saliva swallowing test and modified water swallowing test results. Videofluorography indicated aspiration in 35 (70.0%) patients. In comparing patients with and without abnormal results on each test, the groups with abnormal repetitive saliva swallowing test, abnormal modified water swallowing test, and aspiration demonstrated lower cerebral blood flow in the left precuneus, left insula, and anterior cingulate gyrus, respectively. CONCLUSIONS: Based on the analysis of cerebral blood flow, functional lesions differed across abnormal repetitive saliva swallowing test and abnormal modified water swallowing test findings and aspiration on videofluorography, and each test may assess different functions among the many processes involved in swallowing.

Usability and feasibility of smartphone video capabilities for telehealth care in the U.S. military.

The two-way audio/visual capabilities on the latest smartphone platforms bring new possibilities for the delivery of healthcare services to users. Because this technology is so new, the feasibility and the basic usability of the technology need to be evaluated before more research is conducted on its application in the telehealth field. The purpose of this study was to conduct preliminary usability testing of smartphone two-way video capabilities for potential telehealth use for U.S. military service members. Seven service member volunteers communicated with research staff at a large military installation via Apple's (Cupertino, CA) FaceTime® app on the iPhone® 4 smartphone platform and conducted basic usability testing of the technology. Preferences for potential use of the technology were also assessed. The results showed that the technology has both limitations as well as potential for telehealth applications that warrant additional research.

High-fidelity, low-cost, automated method to assess laparoscopic skills objectively.

BACKGROUND: We sought to define the extent to which a motion analysis-based assessment system constructed with simple equipment could measure technical skill objectively and quantitatively. METHODS: An "off-the-shelf" digital video system was used to capture the hand and instrument movement of surgical trainees (beginner level = PGY-1, intermediate level = PGY-3, and advanced level = PGY-5/fellows) while they performed a peg transfer exercise. The video data were passed through a custom computer vision algorithm that analyzed incoming pixels to measure movement smoothness objectively. RESULTS: The beginner-level group had the poorest performance, whereas those in the advanced group generated the highest scores. Intermediate-level trainees scored significantly (p < 0.04) better than beginner trainees. Advanced-level trainees scored significantly better than intermediate-level trainees and beginner-level trainees (p < 0.04 and p < 0.03, respectively). CONCLUSIONS: A computer vision-based analysis of surgical movements provides an objective basis for technical expertise-level analysis with construct validity. The technology to capture the data is simple, low cost, and readily available, and it obviates the need for expert human assessment in this setting.

Cervical cancer control: deaf and hearing women's response to an educational video.

Deaf people encounter barriers to accessing cancer information. In this study, a graphically enriched educational video about cervical cancer was created in American Sign Language, with English open captioning and voice overlay. Deaf (n = 127) and hearing (n = 106) women completed cancer knowledge surveys before and after viewing the video. Hearing women yielded higher scores before the intervention. Both groups demonstrated a significant increase in general and cervical cancer knowledge after viewing the video, rendering posttest knowledge scores nearly equal between the groups. These findings indicate that this video is an effective strategy for increasing cervical cancer knowledge among deaf women.