Person-Centered Predictions of Psychological Constructs with Social Media Contextualized by ...
summary
Uses multimodal sensing (Bluetooth, wearables, phone) plus Facebook language from 754 participants to build cluster-specific models; hierarchical clustering yields three behavioral clusters and models are evaluated with five-fold CV and a held-out set of 50 participants. Contextualized models improve prediction of personality traits, anxiety, and sleep quality, leave affect unchanged, and worsen cognitive ability prediction. Code uses standard ML comparisons and pooled Pearson/S-MAPE metrics, showing personalization helps when constructs link to physical behavior.
AI-written summary · about
related talks
05mA Multisensor Person-Centered Approach to Understand the Role of Daily Activities in Job ...
38mMeasuring Psychological Traits using Social Media
06mPredicting Subjective Measures of Social Anxiety from Sparsely Collected Mobile Sensor Data
05mInterview with Alex Willcock
07mPredicting Brain Functional Connectivity Using Mobile Sensing
20mAugmenting Mental Health Care in the Digital Age: Machine Learning as a Therapist Assistant
22mMaking Natural Language Processing Robust to Sociolinguistic Variation
25mClaudia Perlich, Chief Scientist, Dstillery, NYC 2017
