Person-Centered Predictions of Psychological Constructs with Social Media Contextualized by Multimodal Sensing

Abstract

Personalized predictions have shown promises in various disciplines but they are fundamentally constrained in their ability to generalize across individuals. These models are often trained on limited datasets which do not represent the fluidity of human functioning. In contrast, generalized models capture normative behaviors between individuals but lack precision in predicting individual outcomes. This paper aims to balance the tradeoff between one-for-each and one-for-all models by clustering individuals on mutable behaviors and conducting cluster-specific predictions of psychological constructs in a multimodal sensing dataset of 754 individuals. Specifically, we situate our modeling on social media that has exhibited capability in inferring psychosocial attributes. We hypothesize that complementing social media data with offline sensor data can help to personalize and improve predictions. We cluster individuals on physical behaviors captured via Bluetooth, wearables, and smartphone sensors. We build contextualized models predicting psychological constructs trained on each cluster's social media data and compare their performance against generalized models trained on all individuals' data. The comparison reveals no difference in predicting affect and a decline in predicting cognitive ability, but an improvement in predicting personality, anxiety, and sleep quality. We construe that our approach improves predicting psychological constructs sharing theoretical associations with physical behavior. We also find how social media language associates with offline behavioral contextualization. Our work bears implications in understanding the nuanced strengths and weaknesses of personalized predictions, and how the effectiveness may vary by multiple factors. This work reveals the importance of taking a critical stance on evaluating the effectiveness before investing efforts in personalization.

Biological Breakdown of Sports in Athletics Based on Multimedia Image Acquisition Techniques

“…However, the rapid development of remote sensing technology has also brought about an explosive growth of remote sensing data volume. Therefore, how to decode the massive remote sensing data intelligently and rapidly and how to extract useful key information from these massive multisource heterogeneous remote sensing data have become an inevitable requirement in the era of remote sensing big data [ 4 ]. Given that different types of optical remote sensing data acquired by different sensors not only have their data characteristics but also have different feature distribution characteristics, different interpretation strategies and methods should be applied to different types of optical remote sensing data.…”

Observer Effect in Social Media Use

“…To measure the observer effect in social media use, we needed to investigate individual-level changes due to the heterogeneity of social media behaviors. However, social media data is sparse and is prone to high variance across individuals; so, it is challenging to extrapolate from individual behaviors, but the extrapolation at a group level is more reliable 31 . Therefore, we adopted a middle ground between fully personalized and fully generalized approaches by clustering individuals on self-reported intrinsic traits and examining the changes per cluster.…”

“…Although personalized approaches can help overcome the above challenge 67;68 , it is hard to conduct personalized examinations on social media data because of sparsity issues, which compromises the statistical power. Therefore, drawing on prior work 31 , we clustered individuals with self-reported intrinsic traits and then examined the outcomes per cluster. This approach is known to account for both between-individual homogeneity and within-individual heterogeneity in our behaviors 31 .…”

“…Therefore, drawing on prior work 31 , we clustered individuals with self-reported intrinsic traits and then examined the outcomes per cluster. This approach is known to account for both between-individual homogeneity and within-individual heterogeneity in our behaviors 31 . As a robustness test of our findings, we also conducted variable-centered regression analyses with the intrinsic traits, described in Supplementary Information.…”

Stepping Into the Next Decade of Ubiquitous and Pervasive Computing: UbiComp and ISWC 2021

“…22 The Health sessions, chaired by Professors Akane Sano, Nabil Alshurafa, Gang Zhou, and Judy Kay, showcased many interesting works in health and wellbeing. Many works presented mHealth and smartphone-based solutions for various health issues ranging from controlling drinking behavior among Dysphagia Patients 24 to respiration monitoring 25 and predicting psychological constructs 26 among others. Varun Mishra and his team presented a chatbot-based digital coach to predict when a person is the most receptive to Just-In-Time Adaptive Health Interventions.…”