Free-text responses improve prediction of mental health outcomes after psychedelic experiences

Mental health surveys and studies regularly ask participants to describe their experiences in their own words. These free-text responses are information-rich but rarely analysed, as quantifying them has traditionally been effortful. Recent open-weight language models offer a new, scalable alternative: they represent any text as numerical vectors (text embeddings), which can be entered as predictors into standard statisticalanalyses. In three datasets on psychedelic experiences (229–958 participants), we tested whether text embeddings of participants’ free-text responses improve held-out predictions of five mental health outcomes beyond 20–45 demographic, setting, and questionnairepredictors. In repeated nested cross-validation, free-text responses reliably increased explained variance by 3–10 percentage points (relative gains of 8–32%) for changes in well-being and psychological flexibility and for post-traumatic symptom severity, but notfor current well-being or post-traumatic growth. Open-ended questions thus capture valuable information that standard questionnaires miss, and we recommend including them routinely and analysing the responses.

Authors

Publication Details

Journal
PsyArXiv (OSF Preprints)
Published
2026-11-19
DOI
https://doi.org/10.31234/osf.io/e9wzk_v2
Primary Topic
Psychedelics and Drug Studies
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Free-text responses improve prediction of mental health outcomes after psychedelic experiences

Ulrike Lueken, Abdo Uyar, Ricarda Evens, Fritz Guenther et al.
2 links
PsyArXiv (OSF Preprints)
Psychedelics and Drug Studies
preprint

Free-text responses improve prediction of mental health outcomes after psychedelic experiences

Ulrike Lueken, Abdo Uyar, Ricarda Evens, Fritz Guenther, Max Wolff, Mathis Parent-Couture, Klara Wieczorke
preprint en

Abstract

Mental health surveys and studies regularly ask participants to describe their experiences in their own words. These free-text responses are information-rich but rarely analysed, as quantifying them has traditionally been effortful. Recent open-weight language models offer a new, scalable alternative: they represent any text as numerical vectors (text embeddings), which can be entered as predictors into standard statisticalanalyses. In three datasets on psychedelic experiences (229–958 participants), we tested whether text embeddings of participants’ free-text responses improve held-out predictions of five mental health outcomes beyond 20–45 demographic, setting, and questionnairepredictors. In repeated nested cross-validation, free-text responses reliably increased explained variance by 3–10 percentage points (relative gains of 8–32%) for changes in well-being and psychological flexibility and for post-traumatic symptom severity, but notfor current well-being or post-traumatic growth. Open-ended questions thus capture valuable information that standard questionnaires miss, and we recommend including them routinely and analysing the responses.

PsyArXiv (OSF Preprints)
Good health and well-being
Psychedelics and Drug Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon