The Session-specific Risk Index (SRI) in psychotherapy: Developing longitudinal session-by-session machine learning algorithms to predict the likelihood of non-response to treatment

INTRODUCTION: This study aimed to develop and introduce an implementation-oriented, session-specific metric called the Session-specific Risk Index (SRI), which defines the likelihood of treatment non-response in psychotherapy. The SRI integrates baseline information with real-time clinical data from each session, providing a continuously updated probability of non-response to treatment. METHODS: We used observational data from an outpatient clinic, including 1,010 psychotherapy patients. We compared the performance of elastic net regressions and gradient boosting machines for the prediction of non-response by session 20 using patients' initial sociodemographic, diagnostic and clinical information, combined with data from their symptoms' clinical evolution. Finally, we used the best performing algorithms to define the SRIs. RESULTS: Elastic net regressions without interaction effects performed the best across the first ten sessions (accuracy = .65-.70; area-under-the-curve = .70 [.70, .71] -.77 [.76, .78]). The use of maximum Youden indices provided optimal SRI thresholds, balancing sensitivity and specificity, to identify patients at high-risk of non-response. Midpoints of SRI groups strongly correlated with proportion of patients not responding to treatment within those groups (weighted correlation= .85, 95%CI [.83-.86]). CONCLUSION: These findings positioned the SRI as an implementation-oriented, timely-sensitive index to determine the risk of non-response to therapy session-by-session. If implemented in clinical practice, SRIs might enhance precision in mental health care going beyond baseline models of treatment personalization.

Authors

Publication Details

Journal
Psychotherapy and Psychosomatics
Published
2026-11-17
DOI
https://doi.org/10.1159/pps/aemag026
Primary Topic
Psychotherapy Techniques and Applications
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

The Session-specific Risk Index (SRI) in psychotherapy: Developing longitudinal session-by-session machine learning algorithms to predict the likelihood of non-response to treatment

Steffen T. Eberhardt, Julian A. Rubel, Juan Martín Gómez Penedo, Wolfgang Lutz et al.
3 links
Psychotherapy and Psychosomatics
Psychotherapy Techniques and Applications
article

The Session-specific Risk Index (SRI) in psychotherapy: Developing longitudinal session-by-session machine learning algorithms to predict the likelihood of non-response to treatment

Steffen T. Eberhardt, Julian A. Rubel, Juan Martín Gómez Penedo, Wolfgang Lutz, Kristin Jankowsky, Juan Segundo Peña Loray, Christoph Flückiger
article en

Abstract

INTRODUCTION: This study aimed to develop and introduce an implementation-oriented, session-specific metric called the Session-specific Risk Index (SRI), which defines the likelihood of treatment non-response in psychotherapy. The SRI integrates baseline information with real-time clinical data from each session, providing a continuously updated probability of non-response to treatment. METHODS: We used observational data from an outpatient clinic, including 1,010 psychotherapy patients. We compared the performance of elastic net regressions and gradient boosting machines for the prediction of non-response by session 20 using patients' initial sociodemographic, diagnostic and clinical information, combined with data from their symptoms' clinical evolution. Finally, we used the best performing algorithms to define the SRIs. RESULTS: Elastic net regressions without interaction effects performed the best across the first ten sessions (accuracy = .65-.70; area-under-the-curve = .70 [.70, .71] -.77 [.76, .78]). The use of maximum Youden indices provided optimal SRI thresholds, balancing sensitivity and specificity, to identify patients at high-risk of non-response. Midpoints of SRI groups strongly correlated with proportion of patients not responding to treatment within those groups (weighted correlation= .85, 95%CI [.83-.86]). CONCLUSION: These findings positioned the SRI as an implementation-oriented, timely-sensitive index to determine the risk of non-response to therapy session-by-session. If implemented in clinical practice, SRIs might enhance precision in mental health care going beyond baseline models of treatment personalization.

Psychotherapy and Psychosomatics
Good health and well-being
Psychotherapy Techniques and Applications
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