Cluster-based Approach to Improve Affect Recognition from Passively Sensed Data

TL;DR

This paper presents a group-based modeling approach to improve passive negative affect recognition from mobile sensor data, outperforming generalized models in personal mental health monitoring.

Contribution

It introduces a novel group-based modeling method that personalizes affect recognition using mobility, communication, and activity data, enhancing accuracy over generalized models.

Findings

Group models outperform generalized models in affect recognition.

Personalized models improve detection accuracy.

Two weeks of daily life data sufficed for effective modeling.

Abstract

Negative affect is a proxy for mental health in adults. By being able to predict participants' negative affect states unobtrusively, researchers and clinicians will be better positioned to deliver targeted, just-in-time mental health interventions via mobile applications. This work attempts to personalize the passive recognition of negative affect states via group-based modeling of user behavior patterns captured from mobility, communication, and activity patterns. Results show that group models outperform generalized models in a dataset based on two weeks of users' daily lives.