We test the hypothesis that changes in preceding physiological arousal can be used to predict imminent aggression proximally before it occurs in youth with autism spectrum disorder (ASD) who are minimally verbal (MV-ASD). We evaluate this hypothesis through statistical analyses performed on physiological biosensor data wirelessly recorded from 20 MV-ASD youth over 69 independent naturalistic observations in a hospital inpatient unit. Using ridge-regularized logistic regression, results demonstrate that, on average, our models are able to predict the onset of aggression 1 minute before it occurs using 3 minutes of prior data with a 0.71 AUC for global, and a 0.84 AUC for person-dependent models.

It has been suggested that changes in physiological arousal precede potentially dangerous aggressive behavior in youth with autism spectrum disorder (ASD) who are minimally verbal (MV-ASD). The current work tests this hypothesis through time-series analyses on biosignals acquired prior to proximal aggression onset. We implement ridge-regularized logistic regression models on physiological biosensor data wirelessly recorded from 15 MV-ASD youth over 64 independent naturalistic observations in a hospital inpatient unit. Our results demonstrate proof-of-concept, feasibility, and incipient validity predicting aggression onset 1 minute before it occurs using global, person-dependent, and hybrid classifier models.

Background Research suggests there is heterogeneity in treatment response for internet-delivered cognitive behavioral therapy (iCBT) users, but few studies have investigated the trajectory of individual symptom change across iCBT treatment. Large patient data sets using routine outcome measures allows the investigation of treatment effects over time as well as the relationship between outcomes and platform use. Understanding trajectories of symptom change, as well as associated characteristics, may prove important for tailoring interventions or identifying patients who may not benefit from the intervention. Objective We aimed to identify latent trajectories of symptom change during the iCBT treatment course for depression and anxiety and to investigate the patients’ characteristics and platform use for each of these classes. Methods This is a secondary analysis of data from a randomized controlled trial designed to examine the effectiveness of guided iCBT for anxiety and depression in the UK Improving Access to Psychological Therapies (IAPT) program. This study included patients from the intervention group (N=256) and followed a longitudinal retrospective design. As part of the IAPT’s routine outcome monitoring system, patients were prompted to complete the Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder-7 (GAD-7) after each supporter review during the treatment period. Latent class growth analysis was used to identify the underlying trajectories of symptom change across the treatment period for both depression and anxiety. Differences in patient characteristics were then evaluated between these trajectory classes, and the presence of a time-varying relationship between platform use and trajectory classes was investigated. Results Five-class models were identified as optimal for both PHQ-9 and GAD-7. Around two-thirds (PHQ-9: 155/221, 70.1%; GAD-7: 156/221, 70.6%) of the sample formed various trajectories of improvement classes that differed in baseline score, the pace of symptom change, and final clinical outcome score. The remaining patients were in 2 smaller groups: one that saw minimal to no gains and another with consistently high scores across the treatment journey. Baseline severity, medication status, and program assigned were significantly associated (P<.001) with different trajectories. Although we did not find a time-varying relationship between use and trajectory classes, we found an overall effect of time on platform use, suggesting that all participants used the intervention significantly more in the first 4 weeks (P<.001). Conclusions Most patients benefit from treatment, and the various patterns of improvement have implications for how the iCBT intervention is delivered. Identifying predictors of nonresponse or early response might inform the level of support and monitoring required for different types of patients. Further work is necessary to explore the differences between these trajectories to understand what works best for whom and to identify early on those patients who are less likely to benefit from treatment.

BACKGROUND Research suggests there is heterogeneity on treatment response for internet-delivered CBT (iCBT) users, but few studies have investigated the trajectory of individual symptom change across iCBT treatment. Large patient datasets using routine outcome measures provides an opportunity to investigate treatments effect over time, as well as the relationship between outcomes and platform usage. Understanding these different trajectories of symptom change, as well as associated characteristics, may prove important for tailoring interventions or identifying patients who may not benefit from the intervention. OBJECTIVE The aim was to identify latent trajectories of symptom change during the iCBT treatment course for depression and anxiety and to investigate the patients’ characteristics and platform usage for each of these classes. METHODS This is a secondary analysis of data from a Randomized Controlled Trial designed to examine the effectiveness of guided-iCBT for anxiety and depression in the UK’s Improving Access to Psychological Therapies (IAPT) program. The present work includes patients from the intervention group (N=256) and follows a longitudinal, retrospective design. As part of IAPT’s Routine Outcome Monitoring system, patients were prompted to complete PHQ-9 and GAD-7 after each supporter review during treatment period. Latent class growth analysis was used to identify underlying trajectories of symptom change across the treatment period for both, depression and anxiety. Differences in patient characteristics were then evaluated between these trajectory classes, and the presence of a time-varying relationship between platform usage and trajectory classes was investigated. RESULTS Five-class models were identified as the optimal ones for both depression and anxiety as measured using the PHQ-9 and GAD-7 respectively. Two thirds (70%) of the sample formed various trajectories of improvement classes that differed on baseline score, pace of symptom change, and final clinical outcome score. The remaining patients were in two smaller groups, one that saw minimal to no gains and another with consistently high scores across the treatment journey. No significant associations were found between different types of trajectories and baseline patient characteristics (age, sex, employment status, presence of long-term condition). While we did not find a time-varying relationship between usage and trajectory classes, we found an overall effect of time on platform usage suggesting that all participants used the intervention significantly more in the first four weeks. CONCLUSIONS Most patients benefit from treatment and the various patterns of improvement have implications for how the iCBT intervention is delivered. Identifying predictors of non-response or early response might inform the level of support and monitoring required for different types of patients. Further work is necessary to explore differences between these trajectories to understand what works best for whom and to identify early-on those patients who are less likely to benefit from treatment.

No abstract