Adolescent use and co-use of tobacco and cannabis in California: The roles of local policy and density of tobacco, vape, and cannabis retailers around schools.

Adolescent tobacco use (particularly vaping) and co-use of cannabis and tobacco have increased, leading some jurisdictions to implement policies intended to reduce youth access to these products; however, their impacts remain unclear. We examine associations between local policy, density of tobacco, vape, and cannabis retailers around schools, and adolescent use and co-use of tobacco/vape and cannabis. We combined 2018 statewide California (US) data on: (a) jurisdiction-level policies related to tobacco and cannabis retail environments, (b) jurisdiction-level sociodemographic composition, (c) retailer locations (tobacco, vape, and cannabis shops), and (d) survey data on 534,176 middle and high school students (California Healthy Kids Survey). Structural equation models examined how local policies and retailer density near schools are associated with frequency of past 30-day cigarette smoking or vaping, cannabis use, and co-use of tobacco/vape and cannabis, controlling for jurisdiction-, school-, and individual-level confounders. Stricter retail environment policies were associated with lower odds of past-month use of tobacco/vape, cannabis, and co-use of tobacco/vape and cannabis. Stronger tobacco/vape policies were associated with higher tobacco/vape retailer density near schools, while stronger cannabis policies and overall policy strength (tobacco/vape and cannabis combined) were associated with lower cannabis and combined retailer densities (summed tobacco/vape and cannabis), respectively. Tobacco/vape shop density near schools was positively associated with tobacco/vape use odds, as was summed retailer density near schools and co-use of tobacco, cannabis. Considering jurisdiction-level tobacco and cannabis control policies are associated with adolescent use of these substances, policymakers may proactively leverage such policies to curb youth tobacco and cannabis use.

A large-scale sentiment analysis of tweets pertaining to the 2020 US presidential election.

We capture the public sentiment towards candidates in the 2020 US Presidential Elections, by analyzing 7.6 million tweets sent out between October 31st and November 9th, 2020. We apply a novel approach to first identify tweets and user accounts in our database that were later deleted or suspended from Twitter. This approach allows us to observe the sentiment held for each presidential candidate across various groups of users and tweets: accessible tweets and accounts, deleted tweets and accounts, and suspended or inaccessible tweets and accounts. We compare the sentiment scores calculated for these groups and provide key insights into the differences. Most notably, we show that deleted tweets, posted after the Election Day, were more favorable to Joe Biden, and the ones posted leading to the Election Day, were more positive about Donald Trump. Also, the older a Twitter account was, the more positive tweets it would post about Joe Biden. The aim of this study is to highlight the importance of conducting sentiment analysis on all posts captured in real time, including those that are now inaccessible, in determining the true sentiments of the opinions around the time of an event.

Examining the utility of nonlinear machine learning approaches versus linear regression for predicting body image outcomes: The U.S. Body Project I.

Most body image studies assess only linear relations between predictors and outcome variables, relying on techniques such as multiple Linear Regression. These predictor variables are often validated multi-item measures that aggregate individual items into a single scale. The advent of machine learning has made it possible to apply Nonlinear Regression algorithms-such as Random Forest and Deep Neural Networks-to identify potentially complex linear and nonlinear connections between a multitude of predictors (e.g., all individual items from a scale) and outcome (output) variables. Using a national dataset, we tested the extent to which these techniques allowed us to explain a greater share of the variance in body-image outcomes (adjusted R2) than possible with Linear Regression. We examined how well the connections between body dissatisfaction and dieting behavior could be predicted from demographic factors and measures derived from objectification theory and the tripartite-influence model. In this particular case, although Random Forest analyses sometimes provided greater predictive power than Linear Regression models, the advantages were small. More generally, however, this paper demonstrates how body image researchers might harness the power of machine learning techniques to identify previously undiscovered relations among body image variables.

Unsupervised Machine Learning for Identifying Challenging Behavior Profiles to Explore Cluster-Based Treatment Efficacy in Children With Autism Spectrum Disorder: Retrospective Data Analysis Study.

BACKGROUND: Challenging behaviors are prevalent among individuals with autism spectrum disorder; however, research exploring the impact of challenging behaviors on treatment response is lacking. OBJECTIVE: The purpose of this study was to identify types of autism spectrum disorder based on engagement in different challenging behaviors and evaluate differences in treatment response between groups. METHODS: Retrospective data on challenging behaviors and treatment progress for 854 children with autism spectrum disorder were analyzed. Participants were clustered based on 8 observed challenging behaviors using k means, and multiple linear regression was performed to test interactions between skill mastery and treatment hours, cluster assignment, and gender. RESULTS: Seven clusters were identified, which demonstrated a single dominant challenging behavior. For some clusters, significant differences in treatment response were found. Specifically, a cluster characterized by low levels of stereotypy was found to have significantly higher levels of skill mastery than clusters characterized by self-injurious behavior and aggression (P<.003). CONCLUSIONS: These findings have implications on the treatment of individuals with autism spectrum disorder. Self-injurious behavior and aggression were prevalent among participants with the worst treatment response, thus interventions targeting these challenging behaviors may be worth prioritizing. Furthermore, the use of unsupervised machine learning models to identify types of autism spectrum disorder shows promise.

Racial and ethnic disparities in opioid use for adolescents at US emergency departments.

BACKGROUND: Racial/ethnic disparities in the use of opioids to treat pain disorders have been previously reported in the emergency department (ED). Further research is needed to better evaluate the impact race/ethnicity may have on the use of opioids in adolescents for the management of pain disorders in the ED. METHODS: This was a cross-sectional study using data from the National Hospital Ambulatory Medical Care Survey from 2006 to 2016. Multivariate models were used to evaluate the role of race/ethnicity in the receipt of opioid agonists while in the ED. All ED visits with patients aged 11-21 years old were analyzed. Races/ethnicities were stratified as non-Hispanic Whites, non-Hispanic Blacks, and Hispanics. In addition to race, statistical analysis included the following covariates: pain score, pain diagnosis, age, region, sex, and payment method. RESULTS: There was a weighted total of 189,256,419 ED visits. Those visits involved 109,826,315 (58%) non-Hispanic Whites, 46,314,977 (24%) non-Hispanic Blacks, and 33,115,127 (18%) Hispanics, with 21.6% (95% CI, 21.1%-22.1), 15.2% (95% CI, 14.6-15.9%), and 17.4% (95% CI, 16.5-18.2%) of those visits reporting use of opioids, respectively. Regardless of age, sex, and region, non-Hispanic Whites received opioids at a higher rate than non-Hispanic Blacks and Hispanics. Based on diagnosis, non-Hispanic Whites received opioids at a higher rate in multiple pain diagnoses. Additionally, non-Hispanic Blacks and Hispanics were less likely to receive an opioid when reporting moderate pain (aOR = 0.738, 95% CI 0.601-0.906, aOR = 0.739, 95% CI 0.578-0.945, respectively) and severe pain (aOR = 0.580, 95% CI 0.500-0.672, aOR = 0.807, 95% CI 0.685-0.951, respectively) compared to non-Hispanic Whites. CONCLUSIONS: Differences in the receipt of opioid agonists in EDs among the races/ethnicities exist, with more non-Hispanic Whites receiving opioids than their minority counterparts. Non-Hispanic Black women may be an especially marginalized population. Further investigation into sex-based and regional differences are needed.

Association Between Coffee Intake and Incident Heart Failure Risk: A Machine Learning Analysis of the FHS, the ARIC Study, and the CHS.

BACKGROUND: Coronary heart disease, heart failure (HF), and stroke are complex diseases with multiple phenotypes. While many risk factors for these diseases are well known, investigation of as-yet unidentified risk factors may improve risk assessment and patient adherence to prevention guidelines. We investigated the diet domain in FHS (Framingham Heart Study), CHS (Cardiovascular Heart Study), and the ARIC study (Atherosclerosis Risk in Communities) to identify potential lifestyle and behavioral factors associated with coronary heart disease, HF, and stroke. METHODS: We used machine learning feature selection based on random forest analysis to identify potential risk factors associated with coronary heart disease, stroke, and HF in FHS. We evaluated the significance of selected variables using univariable and multivariable Cox proportional hazards analysis adjusted for known cardiovascular risks. Findings from FHS were then validated using CHS and ARIC. RESULTS: We identified multiple dietary and behavioral risk factors for cardiovascular disease outcomes including marital status, red meat consumption, whole milk consumption, and coffee consumption. Among these dietary variables, increasing coffee consumption was associated with decreasing long-term risk of HF congruently in FHS, ARIC, and CHS. CONCLUSIONS: Higher coffee intake was found to be associated with reduced risk of HF in all three studies. Further study is warranted to better define the role, possible causality, and potential mechanism of coffee consumption as a potential modifiable risk factor for HF.

Evaluating the 0-10 Point Pain Scale on Adolescent Opioid Use in US Emergency Departments.

OBJECTIVE: To evaluate trends in national emergency department (ED) adolescent opioid use in relation to reported pain scores. METHODS: A retrospective, cross-sectional analysis on National Hospital Ambulatory Medical Care Survey (NHAMCS) data was conducted on ED visits involving patients aged 11-21 from 2008-2017. Crude observational counts were extrapolated to weighted estimates matching total population counts. Multivariate models were used to evaluate the role of a pain score in the reported use of opioids. Anchors for pain scores were 0 (no pain) and 10 (worst pain imaginable). RESULTS: 31,355 observations were captured, which were extrapolated by the NHAMCS to represent 162,515,943 visits nationwide. Overall, patients with a score of 10 were 1.35 times more likely to receive an opioid than patients scoring a 9, 41.7% (CI95 39.7-43.8%) and 31.0% (CI95 28.8-33.3%), respectively. Opioid use was significantly different between traditional pain score cutoffs of mild (1-3) and moderate pain (4-6), where scores of 4 were 1.76 times more likely to receive an opioid than scores of 3, 15.5% (CI95 13.7-17.3%) and 8.8% (CI95 7.1-10.6%), respectively. Scores of 7 were 1.33 times more likely to receive opioids than scores of 6, 24.7% (CI95 23.0-26.3%) and 18.5% (CI95 16.9-20.0%), respectively. Fractures had the highest likelihood of receiving an opioid, as 49.2% of adolescents with a fracture received an opioid (CI95 46.4-51.9%). Within this subgroup, only adolescents reporting a fracture pain score of 10 had significantly higher opioid use than adjacent pain scores, where fracture patients scoring a 10 were 1.4 times more likely to use opioids than those scoring 9, 82.2% (CI95 76.1-88.4%) and 59.8% (CI95 49.0-70.5%), respectively. CONCLUSIONS: While some guidelines in the adult population have revised cut-offs and groupings of the traditional tiers on a 0-10 point pain scale, the adolescent population may also require further examination to potentially warrant a similar adjustment.

Developing Employment Environments Where Individuals with ASD Thrive: Using Machine Learning to Explore Employer Policies and Practices.

An online survey instrument was developed to assess employers' perspectives on hiring job candidates with Autism Spectrum Disorder (ASD). The investigators used K-means clustering to categorize companies in clusters based on their hiring practices related to individuals with ASD. This methodology allowed the investigators to assess and compare the various factors of businesses that successfully hire employees with ASD versus those that do not. The cluster analysis indicated that company structures, policies and practices, and perceptions, as well as the needs of employers and employees, were important in determining who would successfully hire individuals with ASD. Key areas that require focused policies and practices include recruitment and hiring, training, accessibility and accommodations, and retention and advancement.

Learning in the machine: To share or not to share?

Weight-sharing is one of the pillars behind Convolutional Neural Networks and their successes. However, in physical neural systems such as the brain, weight-sharing is implausible. This discrepancy raises the fundamental question of whether weight-sharing is necessary. If so, to which degree of precision? If not, what are the alternatives? The goal of this study is to investigate these questions, primarily through simulations where the weight-sharing assumption is relaxed. Taking inspiration from neural circuitry, we explore the use of Free Convolutional Networks and neurons with variable connection patterns. Using Free Convolutional Networks, we show that while weight-sharing is a pragmatic optimization approach, it is not a necessity in computer vision applications. Furthermore, Free Convolutional Networks match the performance observed in standard architectures when trained using properly translated data (akin to video). Under the assumption of translationally augmented data, Free Convolutional Networks learn translationally invariant representations that yield an approximate form of weight-sharing.

ML-MEDIC: A Preliminary Study of an Interactive Visual Analysis Tool Facilitating Clinical Applications of Machine Learning for Precision Medicine.

Accessible interactive tools that integrate machine learning methods with clinical research and reduce the programming experience required are needed to move science forward. Here, we present Machine Learning for Medical Exploration and Data-Inspired Care (ML-MEDIC), a point-and-click, interactive tool with a visual interface for facilitating machine learning and statistical analyses in clinical research. We deployed ML-MEDIC in the American Heart Association (AHA) Precision Medicine Platform to provide secure internet access and facilitate collaboration. ML-MEDIC's efficacy for facilitating the adoption of machine learning was evaluated through two case studies in collaboration with clinical domain experts. A domain expert review was also conducted to obtain an impression of the usability and potential limitations.

Identification and analysis of behavioral phenotypes in autism spectrum disorder via unsupervised machine learning.

BACKGROUND AND OBJECTIVE: Autism spectrum disorder (ASD) is a heterogeneous disorder. Research has explored potential ASD subgroups with preliminary evidence supporting the existence of behaviorally and genetically distinct subgroups; however, research has yet to leverage machine learning to identify phenotypes on a scale large enough to robustly examine treatment response across such subgroups. The purpose of the present study was to apply Gaussian Mixture Models and Hierarchical Clustering to identify behavioral phenotypes of ASD and examine treatment response across the learned phenotypes. MATERIALS AND METHODS: The present study included a sample of children with ASD (N = 2400), the largest of its kind to date. Unsupervised machine learning was applied to model ASD subgroups as well as their taxonomic relationships. Retrospective treatment data were available for a portion of the sample (n = 1034). Treatment response was examined within each subgroup via regression. RESULTS: The application of a Gaussian Mixture Model revealed 16 subgroups. Further examination of the subgroups through Hierarchical Agglomerative Clustering suggested 2 overlying behavioral phenotypes with unique deficit profiles each composed of subgroups that differed in severity of those deficits. Furthermore, differentiated response to treatment was found across subtypes, with a substantially higher amount of variance accounted for due to the homogenization effect of the clustering. DISCUSSION: The high amount of variance explained by the regression models indicates that clustering provides a basis for homogenization, and thus an opportunity to tailor treatment based on cluster memberships. These findings have significant implications on prognosis and targeted treatment of ASD, and pave the way for personalized intervention based on unsupervised machine learning.

Exploring the relationship of digital information sources and medication adherence.

We present a retrospective analysis of data collected in the United States from the 2015 National Consumer Survey on the Medication Experience and Pharmacists' Role in order to model the relationship between health information sources and medication adherence and perception. Our results indicate that while the digital age has presented prescription users with many non-traditional alternatives for health information, the use of digital content has a significant negative correlation with pharmaceutical adherence and attitudes toward medication. These findings along with previous research suggest that in order to fully realize the potential benefits of the digital age in regards to patient health, positive patient-provider discussions regarding information found online, efforts to improve general health literacy and improvements in the quality and accuracy of the information found are key. Given that higher reliance on digital content is correlated with younger age, the analysis suggests that proactive measures should be taken to educate younger prescription users about the merits and pitfalls of information seeking techniques as they pertain to health literacy.

Exploring Age-Related Metamemory Differences using Modified Brier Scores and Hierarchical Clustering.

Older adults (OAs) typically experience memory failures as they age. However, with some exceptions, studies of OAs' ability to assess their own memory functions-Metamemory (MM) - find little evidence that this function is susceptible to age-related decline. Our study examines OAs' and young adults' (YAs) MM performance and strategy use. Groups of YAs (N = 138) and OAs (N = 79) performed a MM task that required participants to place bets on how likely they were to remember words in a list. Our analytical approach includes hierarchical clustering, and we introduce a new measure of MM-the modified Brier-in order to adjust for differences in scale usage between participants. Our data indicate that OAs and YAs differ in the strategies they use to assess their memory and in how well their MM matches with memory performance. However, there was no evidence that the chosen strategies were associated with differences in MM match, indicating that there are multiple strategies that might be effective (i.e. lead to similar match) in this MM task.

Evaluating Factors Impacting Medication Adherence Among Rural, Urban, and Suburban Populations.

PURPOSE: To evaluate differences in prescription medication adherence rates, as well as influencing factors, in rural and urban adults. METHODS: This is a retrospective analysis of the 2015 National Consumer Survey on the Medication Experience and Pharmacists' Role. A total of 26,173 participants completed the survey and provided usable data. Participants using between 1 and 30 prescription medications and living more than 0 miles and up to 200 miles from their nearest pharmacy were selected for the study, resulting in a total of 15,933 participants. Data from the 2010 US Census and Rural Health Research Center were used to determine the population density of each participant's ZIP code. Participant adherence to reported chronic medications was measured based on the 8-item Morisky Medication Adherence Scale (MMAS-8). FINDINGS: Overall adherence rates did not differ significantly between rural and urban adults with average adherence based on MMAS-8 scores of 5.58 and 5.64, respectively (P = .253). Age, income, education, male sex, and white race/ethnicity were associated with higher adherence rates. While the overall adherence rates between urban and rural adults were not significantly different, the factors that influenced adherence varied between age-specific population density groupings. CONCLUSION: These analyses suggest that there is no significant difference in adherence between rural and urban populations; however, the factors contributing to medication adherence may vary based on age and population density. Future adherence intervention methods should be designed with consideration for these individualized factors.

Trends in Opioid Use in Pediatric Patients in US Emergency Departments From 2006 to 2015.

Importance: The use of opioids to treat pain in pediatric patients has been viewed as necessary; however, this practice has raised concerns regarding opioid abuse and the effects of opioid use. To effectively adjust policy regarding opioids in the pediatric population, prescribing patterns must be better understood. Objective: To evaluate opioid prescribing patterns in US pediatric patients and factors associated with opioid prescribing. Design, Setting, and Participants: This cross-sectional study used publicly available data from the National Hospital Ambulatory Medical Care Survey from January 1, 2006, to December 31, 2015. Analysis included the use of bivariate and multivariate models to evaluate factors associated with opioid prescribing. Practitioners from emergency departments throughout the United States were surveyed, and data were collected using a representative sample of visits to hospital emergency departments. The study analyzed all emergency department visits included in the National Hospital Ambulatory Medical Care Survey for patients younger than 18 years. All statistical analysis was completed in June of 2018 and updated upon receiving reviewer feedback in October of 2018. Exposures: Information regarding participants' medications was collected at time of visit. Participants who reported taking 1 or more opioids were identified. Main Outcomes and Measures: Evaluation of opioid prescribing patterns across demographic factors and pain diagnoses. Results: A total of 69 152 visits with patients younger than 18 years (32 727 female) were included, which were extrapolated by the National Hospital Ambulatory Medical Care Survey to represent 293 528 632 visits nationwide, with opioid use representing 21 276 831 (7.25%) of the extrapolated visits. Factors including geographic region, race, age, and payment method were associated with statistically significant differences in opioid prescribing. The Northeast reported an opioid prescribing rate of 4.69% (95% CI, 3.69%-5.70%) vs 8.84% (95% CI, 6.82%-10.86%) in the West (P = .004). White individuals were prescribed an opioid at 8.11% (95% CI, 7.23%-8.99%) of visits vs 5.31% (95% CI, 4.31%-6.32%) for nonwhite individuals (P < .001). Those aged 13 to 17 years were significantly more likely to receive opioid prescriptions (16.20%; 95% CI, 14.29%-18.12%) than those aged 3 to 12 years (6.59%; 95% CI, 5.75%-7.43%) or 0 to 2 years (1.70%; 95% CI, 1.42%-1.98%). Patients using Medicaid for payment were less likely to receive an opioid than those using private insurance (5.47%; 95% CI, 4.79%-6.15% vs 9.73%; 95% CI, 8.56%-10.90%). There was no significant difference in opioid prescription across sexes. Opioid prescribing rates decreased when comparing 2006 to 2010 with 2011 to 2015 (8.23% [95% CI, 6.75%-9.70%] vs 6.30% [95% CI, 5.44%-7.17%]; P < .001); however, opioid prescribing rates remained unchanged in specific pain diagnoses, including pelvic and back pain. Conclusions and Relevance: This research demonstrated an overall reduction in opioid use among pediatric patients from 2011 to 2015 compared with the previous 5 years; however, there appear to be variations in factors associated with opioid prescribing. The association of location, race, payment method, and pain diagnoses with rates of prescribing of opioids suggests areas of potential quality improvement and further research.

A Program Evaluation of Home and Center-Based Treatment for Autism Spectrum Disorder.

The present study aimed to retrospectively compare the relative rates of mastery of exemplars for individuals with ASD (N = 313) who received home-based and center-based services. A between-group analysis found that participants mastered significantly more exemplars per hour when receiving center-based services than home-based services. Likewise, a paired-sample analysis found that participants who received both home and center-based services had mastered 100 % more per hour while at the center than at home. These analyses indicated that participants demonstrated higher rates of learning during treatment that was provided in a center setting than in the participant's home.

T-Time: A data repository of T cell and calcium release-activated calcium channel activation imagery.

BACKGROUND: A fundamental understanding of live-cell dynamics is necessary in order to advance scientific techniques and personalized medicine. For this understanding to be possible, image processing techniques, probes, tracking algorithms and many other methodologies must be improved. Currently there are no large open-source datasets containing live-cell imaging to act as a standard for the community. As a result, researchers cannot evaluate their methodologies on an independent benchmark or leverage such a dataset to formulate scientific questions. FINDINGS: Here we present T-Time, the largest free and publicly available data set of T cell phase contrast imagery designed with the intention of furthering live-cell dynamics research. T-Time consists of over 40 GB of imagery data, and includes annotations derived from these images using a custom T cell identification and tracking algorithm. The data set contains 71 time-lapse sequences containing T cell movement and calcium release activated calcium channel activation, along with 50 time-lapse sequences of T cell activation and T reg interactions. The database includes a user-friendly web interface, summary information on the time-lapse images, and a mechanism for users to download tailored image datasets for their own research. T-Time is freely available on the web at http://ttime.mlatlab.org . CONCLUSIONS: T-Time is a novel data set of T cell images and associated metadata. It allows users to study T cell interaction and activation.

Remote sensing observation of annual dust cycles and possible causality of Kawasaki disease outbreaks in Japan.

Kawasaki disease (KD) is a rare vascular disease that, if left untreated, can result in irreparable cardiac damage in children. While the symptoms of KD are well-known, as are best practices for treatment, the etiology of the disease and the factors contributing to KD outbreaks remain puzzling to both medical practitioners and scientists alike. Recently, a fungus known as Candida, originating in the farmlands of China, has been blamed for outbreaks in China and Japan, with the hypothesis that it can be transported over long ranges via different wind mechanisms. This paper provides evidence to understand the transport mechanisms of dust at different geographic locations and the cause of the annual spike of KD in Japan. Candida is carried along with many other dusts, particles or aerosols, of various sizes in major seasonal wind currents. The evidence is based upon particle categorization using the Moderate Resolution Imaging Spectrometer (MODIS) Aerosol Optical Depth (AOD), Fine Mode Fraction (FMF) and Ångström Exponent (AE), the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) attenuated backscatter and aerosol subtype, and the Aerosol Robotic Network's (AERONET) derived volume concentration. We found that seasonality associated with aerosol size distribution at different geographic locations plays a role in identifying dominant abundance at each location. Knowing the typical size of the Candida fungus, and analyzing aerosol characteristics using AERONET data reveals possible particle transport association with KD events at different locations. Thus, understanding transport mechanisms and accurate identification of aerosol sources is important in order to understand possible triggers to outbreaks of KD. This work provides future opportunities to leverage machine learning, including state-of-the-art deep architectures, to build predictive models of KD outbreaks, with the ultimate goal of early forecasting and intervention within a nascent global health early-warning system.

Intensity and Learning Outcomes in the Treatment of Children With Autism Spectrum Disorder.

Ample research has shown that intensive applied behavior analysis (ABA) treatment produces robust outcomes for individuals with autism spectrum disorder (ASD); however, little is known about the relationship between treatment intensity and treatment outcomes. The current study was designed to evaluate this relationship. Participants included 726 children, ages 1.5 to 12 years old, receiving community-based behavioral intervention services. Results indicated a strong relationship between treatment intensity and mastery of learning objectives, where higher treatment intensity predicted greater progress. Specifically, 35% of the variance in mastery of learning objectives was accounted for by treatment hours using standard linear regression, and 60% of variance was accounted for using artificial neural networks. These results add to the existing support for higher intensity treatment for children with ASD.

An Evaluation of the Impact of Supervision Intensity, Supervisor Qualifications, and Caseload on Outcomes in the Treatment of Autism Spectrum Disorder.

Ample research has shown the benefits of intensive applied behavior analysis (ABA) treatment for autism spectrum disorder (ASD); research that investigates the role of treatment supervision, however, is limited. The present study examined the relationship between mastery of learning objectives and supervision hours, supervisor credentials, years of experience, and caseload in a large sample of children with ASD (N = 638). These data were retrieved from a large archival database of children with ASD receiving community-based ABA services. When analyzed together via a multiple linear regression, supervision hours and treatment hours accounted for only slightly more of the observed variance (r2 = 0.34) than treatment hours alone (r2 = 0.32), indicating that increased supervision hours do not dramatically increase the number of mastered learning objectives. In additional regression analyses, supervisor credentials were found to have a significant impact on the number of mastered learning objectives, wherein those receiving supervision from a Board Certified Behavior Analyst (BCBA) mastered significantly more learning objectives. Likewise, the years of experience as a clinical supervisor showed a small but significant impact on the mastery of learning objectives. A supervisor's caseload, however, was not a significant predictor of the number of learning objectives mastered. These findings provide guidance for best practice recommendations.