JMIR Publications

ABSTRACT

Background:

There is an urgent need to enhance understanding of how accessing a crisis helpline benefits help-seekers. Affective computing has the potential to transform this area of research, yet remains relatively unexplored, partly due to the scarcity of available helpline data.

Objective:

This study aims to explore the feasibility of using lexical analysis to explore changes in the mental state of help-seekers accessing a crisis helpline via online chat.

Methods:

6618 de-identified online chat transcripts from Lifeline Australia were examined using the validated Empath lexical categories of Positive Emotion, Negative Emotion, Suffering, and Optimism. Distress and Suicidality categories were also developed and analyzed as context specific for crisis support. One-way ANOVAs were used to assess change in each category across the beginning, middle, and end phases of conversation.

Results:

The context specific categories of Distress (d=.79) and Suicidality (d=.49) showed the strongest improvements across phases of conversation. The most frequently occurring member terms representing Distress were ‘hard’, ‘bad’, and ‘down’, and for Suicidality were ‘suicide’, ‘stop’, and ‘hurt’. The negatively framed Empath categories of Suffering (d=.49) and Negative Emotion (d=.39) also significantly reduced. The positively framed Empath categories also significantly reduced, however, with fewer terms associated with Positive Emotion (d=.15) and Optimism (d=.07) from the beginning to end of conversation.

Conclusions:

This study has demonstrated the feasibility of using lexical analysis to represent and monitor changes in mental state in online crisis support. The findings suggest that lexical analysis could be applied to improving crisis support service delivery and outcome measurement, with further validation.