Detecting and Classifying Self-injurious Behavior in Autism Spectrum Disorder Using Machine Learning Techniques

Cantin-Garside, Kristine D.; Kong, Zhenyu; White, Susan W.; Antezana, Ligia; Kim, Sunwook

doi:10.1007/s10803-020-04463-x

Cited by 40 publications

(34 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MLR here had higher accuracy, specificity, precision, and recall compared to several commonly-used machine learning algorithms. These machine learning algorithms also detected SIB with high accuracy in our previous study using featureless data, though accuracy greatly decreased at the group-level (see Cantin-Garside et al for further details) 34 . Multi-level regression with both variable slopes and intercept may be preferred for group data with variable behaviors that could be specific to an individual, and it may also be more accessible to interpretation than other machine learning algorithms.…”

Section: Discussionmentioning

confidence: 67%

“…Multiple researchers annotated and discussed the video data before labeling raw accelerometry files (see Fig. 1 for an overview of the modeling process, and Cantin-Garside et al 34 for details on consensus-building for SIB labels) 34 . Raw sensor data were filtered using a 4th order, low-pass, recursive Butterworth filter, with a cutoff frequency of 20 Hz.…”

Section: Methodsmentioning

confidence: 99%

“…Data used here were obtained in Cantin-Garside et al 34 and are briefly summarized below, with additional information in Supplementary Material. Children with SIB and ASD were recruited through the university-affiliated psychology clinic and through the authors’ networks.…”

Section: Methodsmentioning

confidence: 99%

“…In our previous study, we examined featureless classification methods to detect the presence of SIB and classify the type of SIB among individuals and groups of individuals with ASD 34 . Using data from wearable accelerometers as input, accuracy was up to 99.1% for individual models, and up to 94.6% for models specific to SIB type.…”

mentioning

confidence: 99%

“…However, the detection models that incorporated all participant data had substantially poorer performance (accuracy = 48.8%), likely due to the inter-and intra-individual variability in SIB and activity levels. The current study is a secondary analysis of data described in Cantin-Garside et al 34 , and employed a multi-level modeling technique that includes motor variability features to address this limitation 34 .…”

mentioning

confidence: 99%

See 4 more Smart Citations

Multi-level modeling with nonlinear movement metrics to classify self-injurious behaviors in autism spectrum disorder

Cantin-Garside

Srinivasan

Ranganathan

et al. 2020

Sci Rep

Self Cite

View full text Add to dashboard Cite

Self-injurious behavior (SIB) is among the most dangerous concerns in autism spectrum disorder (ASD), often requiring detailed and tedious management methods. Sensor-based behavioral monitoring could address the limitations of these methods, though the complex problem of classifying variable behavior should be addressed first. We aimed to address this need by developing a group-level model accounting for individual variability and potential nonlinear trends in SIB, as a secondary analysis of existing data. Ten participants with ASD and SIB engaged in free play while wearing accelerometers. Movement data were collected from > 200 episodes and 18 different types of SIB. Frequency domain and linear movement variability measures of acceleration signals were extracted to capture differences in behaviors, and metrics of nonlinear movement variability were used to quantify the complexity of SIB. The multi-level logistic regression model, comprising of 12 principal components, explained > 65% of the variance, and classified SIB with > 75% accuracy. Our findings imply that frequency-domain and movement variability metrics can effectively predict SIB. Our modeling approach yielded superior accuracy than commonly used classifiers (~ 75 vs. ~ 64% accuracy) and had superior performance compared to prior reports (~ 75 vs. ~ 69% accuracy) This work provides an approach to generating an accurate and interpretable group-level model for SIB identification, and further supports the feasibility of developing a real-time SIB monitoring system.

show abstract

Section: Discussionmentioning

confidence: 67%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Multi-level modeling with nonlinear movement metrics to classify self-injurious behaviors in autism spectrum disorder

Cantin-Garside

Srinivasan

Ranganathan

et al. 2020

Sci Rep

Self Cite

View full text Add to dashboard Cite

show abstract

Characterizing associations between emotion dysregulation, anxiety, and repetitive behaviors in autistic youth with intellectual disability

Ferguson,

Spackman,

Cai

et al. 2024

Autism Research

View full text Add to dashboard Cite

There is a paucity of research that explores associations between emotion dysregulation and the expression and severity of core and co‐occurring characteristics of autism spectrum disorder (ASD), especially in autistic youth with co‐occurring intellectual disability (ID). This study explored the interplay between developmental characteristics and emotion dysregulation, anxiety, and specific subtypes of RRBs that are elevated in autistic youth with co‐occurring ID. Generalized additive models demonstrated that age, sex, language level, dysregulation, and anxiety showed unique patterns of association with subtypes of RRBs in a sample of 150 autistic youth with co‐occurring ID who are non‐speaking or minimally verbal. More specifically, higher anxiety levels were significantly associated with elevations in repetitive sensory motor behaviors (RSMB), self‐injurious behaviors (SIB), insistence on sameness (IS), and unusual interests (UI). While emotion dysregulation was a significant predictor of UI, it demonstrated positive, albeit not significant, associations with the intensity of SIB and RSMB. Language level was a significant predictor of RSMB, such that the intensity of RSMB was higher for individuals who were non‐speaking relative to those who spoke in single words. These findings provide preliminary insights into patterns of associations between emotion dysregulation, anxiety, and specific subdomains of RRBs in autistic youth with ID.

show abstract

Exploring the Heterogeneity of Self‐Injurious Behaviors in Autistic Youth: Patterns, Predictors, and Implications for Intervention

Ferguson,

Spackman,

Cai

et al. 2024

Autism Research

View full text Add to dashboard Cite

Self‐injurious behaviors (SIB) encompass a heterogeneous set of self‐inflicted aggressive behaviors that are highly prevalent in autistic youth. Existing research on SIB in autism spectrum disorder (ASD) has been limited by significant methodological and conceptual inconsistencies. Thus, the current study leveraged item‐level data capturing the severity of unique SIB topographies to further understanding of factors associated with distinct SIB in a sample of 582 autistic youth (Mage = 12.12, SDage = 3.68; range: 3–19 years; 13% females). Results suggest variation in severity endorsements for specific SIB topographies amongst autistic youth, such that 30%–50% of caregivers endorsed slight to very serious concern regarding the SIB topographies of bites nails/skin/fingers, scratches self, hits head/face/neck, bangs head against things, and picks skin. Generalized additive models demonstrated distinct patterns of associations between each SIB topography and dysregulation, sensory hypersensitivity, age, sex, IQ, and language level. Findings underscore the importance of exploring SIB as a multifaceted construct to capture unique correlates of distinct SIB that vary in severity and functional impact, which is critical for the development of effective interventions. This study represents an important step towards more individualized characterization of SIB and support for diverse presentations of these behaviors in autistic youth.

show abstract

Detecting and Classifying Self-injurious Behavior in Autism Spectrum Disorder Using Machine Learning Techniques

Cited by 40 publications

References 54 publications

Multi-level modeling with nonlinear movement metrics to classify self-injurious behaviors in autism spectrum disorder

Multi-level modeling with nonlinear movement metrics to classify self-injurious behaviors in autism spectrum disorder

Characterizing associations between emotion dysregulation, anxiety, and repetitive behaviors in autistic youth with intellectual disability

Exploring the Heterogeneity of Self‐Injurious Behaviors in Autistic Youth: Patterns, Predictors, and Implications for Intervention

Contact Info

Product

Resources

About