Search for: All records

There has been growing interest in developing ubiquitous technologies to analyze adult-child speech in naturalistic settings such as free play in order to support children's social and academic development, language acquisition, and parent-child interactions. However, these technologies often rely on off-the-shelf speech processing tools that have not been evaluated on child speech or child-directed adult speech, whose unique characteristics might result in significant performance gaps when using models trained on adult speech. This work introduces the Playlogue dataset containing over 33 hours of long-form, naturalistic, play-based adult-child conversations from three different corpora of preschool-aged children. Playlogue enables researchers to train and evaluate speaker diarization and automatic speech recognition models on child-centered speech. We demonstrate the lack of generalizability of existing state-of-the-art models when evaluated on Playlogue, and show how fine-tuning models on adult-child speech mitigates the performance gap to some extent but still leaves considerable room for improvement. We further annotate over 5 hours of the Playlogue dataset with 8668 validated adult and child speech act labels, which can be used to train and evaluate models to provide clinically relevant feedback on parent-child interactions. We investigate the performance of state-of-the-art language models at automatically predicting these speech act labels, achieving significant accuracy with simple chain-of-thought prompting or minimal fine-tuning. We use inhome pilot data to validate the generalizability of models trained on Playlogue, demonstrating its utility in improving speech and language technologies for child-centered conversations. The Playlogue dataset is available for download at https://huggingface.co/datasets/playlogue/playlogue-v1. 

Emotion dysregulation in early childhood is known to be associated with a higher risk of several psychopathological conditions, such as ADHD and mood and anxiety disorders. In developmental neuroscience research, emotion dysregulation is characterized by low neural activation in the prefrontal cortex during frustration. In this work, we report on an exploratory study with 94 participants aged 3.5 to 5 years, investigating whether behavioral measures automatically extracted from facial videos can predict frustration-related neural activation and differentiate between low- and high-risk individuals. We propose a novel multi-scale instance fusion framework to develop EarlyScreen - a set of classifiers trained on behavioral markers during emotion regulation. Our model successfully predicts activation levels in the prefrontal cortex with an area under the receiver operating characteristic (ROC) curve of 0.85, which is on par with widely-used clinical assessment tools. Further, we classify clinical and non-clinical subjects based on their psychopathological risk with an area under the ROC curve of 0.80. Our model's predictions are consistent with standardized psychometric assessment scales, supporting its applicability as a screening procedure for emotion regulation-related psychopathological disorders. To the best of our knowledge, EarlyScreen is the first work to use automatically extracted behavioral features to characterize both neural activity and the diagnostic status of emotion regulation-related disorders in young children. We present insights from mental health professionals supporting the utility of EarlyScreen and discuss considerations for its subsequent deployment. 

Background With nearly 20% of the US adult population using fitness trackers, there is an increasing focus on how physiological data from these devices can provide actionable insights about workplace performance. However, in-the-wild studies that understand how these metrics correlate with cognitive performance measures across a diverse population are lacking, and claims made by device manufacturers are vague. While there has been extensive research leading to a variety of theories on how physiological measures affect cognitive performance, virtually all such studies have been conducted in highly controlled settings and their validity in the real world is poorly understood. Objective We seek to bridge this gap by evaluating prevailing theories on the effects of a variety of sleep, activity, and heart rate parameters on cognitive performance against data collected in real-world settings. Methods We used a Fitbit Charge 3 and a smartphone app to collect different physiological and neurobehavioral task data, respectively, as part of our 6-week-long in-the-wild study. We collected data from 24 participants across multiple population groups (shift workers, regular workers, and graduate students) on different performance measures (vigilant attention and cognitive throughput). Simultaneously, we used a fitness tracker to unobtrusively obtain physiological measures that could influence these performance measures, including over 900 nights of sleep and over 1 million minutes of heart rate and physical activity metrics. We performed a repeated measures correlation (rrm) analysis to investigate which sleep and physiological markers show association with each performance measure. We also report how our findings relate to existing theories and previous observations from controlled studies. Results Daytime alertness was found to be significantly correlated with total sleep duration on the previous night (rrm=0.17, P<.001) as well as the duration of rapid eye movement (rrm=0.12, P<.001) and light sleep (rrm=0.15, P<.001). Cognitive throughput, by contrast, was not found to be significantly correlated with sleep duration but with sleep timing—a circadian phase shift toward a later sleep time corresponded with lower cognitive throughput on the following day (rrm=–0.13, P<.001). Both measures show circadian variations, but only alertness showed a decline (rrm=–0.1, P<.001) as a result of homeostatic pressure. Both heart rate and physical activity correlate positively with alertness as well as cognitive throughput. Conclusions Our findings reveal that there are significant differences in terms of which sleep-related physiological metrics influence each of the 2 performance measures. This makes the case for more targeted in-the-wild studies investigating how physiological measures from self-tracking data influence, or can be used to predict, specific aspects of cognitive performance.