TY - GEN
T1 - Reading Between the Lines
T2 - 2024 CHI Conference on Human Factors in Computing Sytems, CHI 2024
AU - O'Dea, Bridianne
AU - Braund, Taylor A.
AU - Batterham, Philip J.
AU - Larsen, Mark E.
AU - Glozier, Nick
AU - Whitton, Alexis E.
PY - 2024/5/11
Y1 - 2024/5/11
N2 - Stratifying depressed individuals may help to improve recovery rates by identifying the subgroups who would benefit from targeted treatments. Detecting depressed individuals with prominent anhedonia (i.e. lack of pleasure) may be one effective approach, given these individuals experience poorer treatment outcomes. This paper explores the linguistic features associated with anhedonia among depressed adults. Over 9 weeks, 218 individuals with depressive symptoms completed a fortnightly psychometric measure of depression (PHQ-9) and provided text data (SMS, social media posts, expressive essays, emotion diaries, personal letters). Linguistic features were examined using LIWC-22. Greater use of discrepancy words was significantly associated with higher anhedonia, but in SMS data only. Machine learning showed some utility for predicting increased anhedonia, with discrepancy words the most important linguistic feature in the model. Discrepancy words were not found to be associated with overall depression scores. These results suggest that this linguistic feature may show some promise for the stratification of anhedonic depression.
AB - Stratifying depressed individuals may help to improve recovery rates by identifying the subgroups who would benefit from targeted treatments. Detecting depressed individuals with prominent anhedonia (i.e. lack of pleasure) may be one effective approach, given these individuals experience poorer treatment outcomes. This paper explores the linguistic features associated with anhedonia among depressed adults. Over 9 weeks, 218 individuals with depressive symptoms completed a fortnightly psychometric measure of depression (PHQ-9) and provided text data (SMS, social media posts, expressive essays, emotion diaries, personal letters). Linguistic features were examined using LIWC-22. Greater use of discrepancy words was significantly associated with higher anhedonia, but in SMS data only. Machine learning showed some utility for predicting increased anhedonia, with discrepancy words the most important linguistic feature in the model. Discrepancy words were not found to be associated with overall depression scores. These results suggest that this linguistic feature may show some promise for the stratification of anhedonic depression.
KW - Anhedonia
KW - Depression
KW - Digital phenotype
KW - Linguistics
KW - Stratification
UR - http://www.scopus.com/inward/record.url?scp=85194825444&partnerID=8YFLogxK
UR - http://purl.org/au-research/grants/NHMRC/1165233
UR - http://purl.org/au-research/grants/NHMRC/1197249
U2 - 10.1145/3613904.3642478
DO - 10.1145/3613904.3642478
M3 - Conference contribution
AN - SCOPUS:85194825444
T3 - Conference on Human Factors in Computing Systems - Proceedings
BT - CHI 2024 - Proceedings of the 2024 CHI Conference on Human Factors in Computing Sytems
PB - Association for Computing Machinery
CY - New York, USA
Y2 - 11 May 2024 through 16 May 2024
ER -