Dataset Information

The potential of accelerating early detection of autism through content analysis of YouTube videos.

ABSTRACT: Autism is on the rise, with 1 in 88 children receiving a diagnosis in the United States, yet the process for diagnosis remains cumbersome and time consuming. Research has shown that home videos of children can help increase the accuracy of diagnosis. However the use of videos in the diagnostic process is uncommon. In the present study, we assessed the feasibility of applying a gold-standard diagnostic instrument to brief and unstructured home videos and tested whether video analysis can enable more rapid detection of the core features of autism outside of clinical environments. We collected 100 public videos from YouTube of children ages 1-15 with either a self-reported diagnosis of an ASD (N?=?45) or not (N?=?55). Four non-clinical raters independently scored all videos using one of the most widely adopted tools for behavioral diagnosis of autism, the Autism Diagnostic Observation Schedule-Generic (ADOS). The classification accuracy was 96.8%, with 94.1% sensitivity and 100% specificity, the inter-rater correlation for the behavioral domains on the ADOS was 0.88, and the diagnoses matched a trained clinician in all but 3 of 22 randomly selected video cases. Despite the diversity of videos and non-clinical raters, our results indicate that it is possible to achieve high classification accuracy, sensitivity, and specificity as well as clinically acceptable inter-rater reliability with nonclinical personnel. Our results also demonstrate the potential for video-based detection of autism in short, unstructured home videos and further suggests that at least a percentage of the effort associated with detection and monitoring of autism may be mobilized and moved outside of traditional clinical environments.

SUBMITTER: Fusaro VA

PROVIDER: S-EPMC3989176 | biostudies-literature | 2014

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

The potential of accelerating early detection of autism through content analysis of YouTube videos.

Fusaro Vincent A VA Daniels Jena J Duda Marlena M DeLuca Todd F TF D'Angelo Olivia O Tamburello Jenna J Maniscalco James J Wall Dennis P DP

PloS one 20140416 4

Autism is on the rise, with 1 in 88 children receiving a diagnosis in the United States, yet the process for diagnosis remains cumbersome and time consuming. Research has shown that home videos of children can help increase the accuracy of diagnosis. However the use of videos in the diagnostic process is uncommon. In the present study, we assessed the feasibility of applying a gold-standard diagnostic instrument to brief and unstructured home videos and tested whether video analysis can enable m ...[more]

PMID: 24740236

Similar Datasets

Project description:BackgroundYouTube (Alphabet Incorporated) has become the most popular video-sharing platform in the world. The Fukushima Daiichi Nuclear Power Plant (FDNPP) disaster resulted in public anxiety toward nuclear power and radiation worldwide. YouTube is an important source of information about the FDNPP disaster for the world.ObjectiveThis study's objectives were to examine the characteristics of YouTube videos related to the FDNPP disaster, analyze the content and comments of videos with a quantitative method, and determine which features contribute to making a video popular with audiences. This study is the first to examine FDNPP disaster-related videos on YouTube.MethodsWe searched for the term "Fukushima nuclear disaster" on YouTube on November 2, 2019. The first 60 eligible videos in the relevance, upload date, view count, and rating categories were recorded. Videos that were irrelevant, were non-English, had inappropriate words, were machine synthesized, and were <3 minutes long were excluded. In total, 111 videos met the inclusion criteria. Parameters of the videos, including the number of subscribers, length, the number of days since the video was uploaded, region, video popularity (views, views/day, likes, likes/day, dislikes, dislikes/day, comments, comments/day), the tone of the videos, the top ten comments, affiliation, whether Japanese people participated in the video, whether the video recorder visited Fukushima, whether the video contained theoretical knowledge, and whether the video contained information about the recent situation in Fukushima, were recorded. By using criteria for content and technical design, two evaluators scored videos and grouped them into the useful (score: 11-14), slightly useful (score: 6-10), and useless (score: 0-5) video categories.ResultsOf the 111 videos, 43 (38.7%) videos were useful, 43 (38.7%) were slightly useful, and 25 (22.5%) were useless. Useful videos had good visual and aural effects, provided vivid information on the Fukushima disaster, and had a mean score of 12 (SD 0.9). Useful videos had more views per day (P<.001), likes per day (P<.001), and comments per day (P=.02) than useless and slightly useful videos. The popularity of videos had a significant correlation with clear sounds (likes/day: P=.001; comments/day: P=.02), vivid information (likes/day: P<.001; comments/day: P=.007), understanding content (likes/day: P=.001; comments/day: P=.04). There was no significant difference in likes per day (P=.72) and comments per day (P=.11) between negative and neutral- and mixed-tone videos. Videos about the recent situation in Fukushima had more likes and comments per day. Video recorders who personally visited Fukushima Prefecture had more subscribers and received more views and likes.ConclusionsThe possible features that made videos popular to the public included video quality, videos made in Fukushima, and information on the recent situation in Fukushima. During risk communication on new forms of media, health institutes should increase publicity and be more approachable to resonate with international audiences.

Project description:BACKGROUND:As user modification can alter the addictiveness and toxicity of electronic nicotine delivery systems (ENDS), more research is needed to understand the types, motivations, risks, and information sources that lead to these product alterations. YouTube has been identified as a major platform where ENDS users obtain and share information about ENDS products and modifications. However, a comprehensive study of ENDS modification videos on YouTube is lacking. OBJECTIVE:This study aimed to analyze the content of YouTube videos depicting modifications of ENDS. METHODS:YouTube was searched in March 2019 to identify videos depicting ENDS modifications. Search terms were derived from interviews with ENDS users and current literature. We used 28 search phrases that combined the words vape and vaping with modification-related key terms (eg, custom build, modification, and dripping). The final sample included 168 videos. RESULTS:Videos were 1 to 108 min long (median 9.55). Presenters were largely male (117/168, 69.6%), white (94/168, 56.0%), and older than 25 years (94/168, 56.0%). Most videos gave how to instructions (148/168, 88.1%), but few offered warnings (30/168, 17.9%) or mentioned commercial alternatives to modifications they presented (16/168, 9.5%). The ENDS devices most often featured were drippers (63/168, 37.5%) and refillable tanks (37/168, 22.0%). The most often modified ENDS components were coils (82/168, 48.8%) and e-liquids (34/168, 20.2%), which included adding other substances, such as cannabis, to the e-liquids (6/168, 3.6%). Most videos portrayed ENDS modifications positively (106/168, 63.1% positive; 60/168, 35.7% neutral; and 2/168, 1.2% negative) and were either neutral or positive in their overall portrayal of ENDS devices (78/168, 46.4% positive; 89/168, 53.0% neutral; and 1/168, 0.6% negative). CONCLUSIONS:This study identified several concerning trends in popular YouTube videos on ENDS modifications, including lack of warnings, the addition of marijuana derivatives to e-liquids, and the positive portrayal of ENDS devices and modifications. By identifying the types of modifications (coil and e-liquid being the most prevalent), this study sets an agenda for research on the effects of modifications.

Project description:BackgroundPeople with gestational diabetes have enhanced learning requirements during pregnancy, and management of their disease often requires the translation of health information into new health behavior changes. Seeking information from the internet to augment learning from health professionals is becoming more common during pregnancy. YouTube is a popular free and accessible web-based resource, which may be particularly useful for individuals with low health literacy or other barriers to receiving high-quality health care; however, the quality and content of YouTube videos varies, and little is known about those covering gestational diabetes.ObjectiveWe aimed to systematically evaluate the quality, content, and reliability of YouTube videos about gestational diabetes.MethodsA systematic search of YouTube videos was conducted over the course of 1 week in April 2020 using the following keywords: "gestational diabetes," "gestational diabetes management," "gestational diabetes treatment," and "pregnancy and diabetes." The search results were displayed by relevance, replicating a default YouTube search attempt. The first 60 results from each keyword were reviewed (n=240). Exclusion criteria were videos unrelated to gestational diabetes, videos not in English, and those for which the full video was not available at the time of review. For each unique video, a gestational diabetes content score was used to rate video comprehensiveness and accuracy, and the DISCERN instrument, a validated metric to assess consumer health information, was used to evaluate the reliability of information presented. Videos were further categorized by quality: videos with DISCERN scores lower than 3 (out of 5) or a content score less than 4 (out of 7) were categorized as low quality, and all others were designated high quality. We performed descriptive analysis and compared video characteristics by source and quality rating.ResultsFor 115 unique videos, the mean content score (out of 7) was 3.5 (SD 2.0) , and the mean DISCERN score (out of 5) was 2.7 (SD 0.7), representing low to moderate information comprehensiveness and reliability respectively. Video sources were categorized as personal vlog (12/115, 10.4%), web-based education (37/115, 32.2%), medical (52/115, 45.2%), business or company (13/115, 11.3%), and media clip (1/115, 0.9%). DISCERN and content scores trended higher among medical and web-based education videos. The majority of videos (n=88) were categorized as low quality, while 27 videos were categorized as high quality. Video duration was longer for high-quality videos (P<.001); high- and low-quality videos otherwise had similar views and viewer interaction numbers.ConclusionsAlthough high-quality videos about gestational diabetes exist, reliability, accuracy, and comprehensiveness were low overall, and higher quality was not associated with increased viewer interaction. It is important to acknowledge the limitations of this platform and to assist patients in accessing high quality content and differentiating the quality of information sources.

Dataset Information

The potential of accelerating early detection of autism through content analysis of YouTube videos.

Publications

The potential of accelerating early detection of autism through content analysis of YouTube videos.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets