Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis

Nikolaos Koutsouleris*, Lana Kambeitz-ilankovic, Stephan Ruhrmann, Marlene Rosen, Anne Ruef, Dominic B. Dwyer, Marco Paolini, Katharine Chisholm, Joseph Kambeitz, Theresa Haidl, André Schmidt, John Gillam, Frauke Schultze-lutter, Peter Falkai, Maximilian Reiser, Anita Riecher-rössler, Rachel Upthegrove, Jarmo Hietala, Raimo K. R. Salokangas, Christos PantelisEva Meisenzahl, Stephen J. Wood, Dirk Beque, Paolo Brambilla, Stefan Borgwardt

*Corresponding author for this work

Research output: Contribution to journalArticle

Abstract

Importance Social and occupational impairments contribute to the burden of psychosis and depression. There is a need for risk stratification tools to inform personalized functional-disability preventive strategies for individuals in at-risk and early phases of these illnesses.

Objective To determine whether predictors associated with social and role functioning can be identified in patients in clinical high-risk (CHR) states for psychosis or with recent-onset depression (ROD) using clinical, imaging-based, and combined machine learning; assess the geographic, transdiagnostic, and prognostic generalizability of machine learning and compare it with human prognostication; and explore sequential prognosis encompassing clinical and combined machine learning.

Design, Setting, and Participants This multisite naturalistic study followed up patients in CHR states, with ROD, and with recent-onset psychosis, and healthy control participants for 18 months in 7 academic early-recognition services in 5 European countries. Participants were recruited between February 2014 and May 2016, and data were analyzed from April 2017 to January 2018.

Main Outcomes and Measures Performance and generalizability of prognostic models.

Results A total of 116 individuals in CHR states (mean [SD] age, 24.0 [5.1] years; 58 [50.0%] female) and 120 patients with ROD (mean [SD] age, 26.1 [6.1] years; 65 [54.2%] female) were followed up for a mean (SD) of 329 (142) days. Machine learning predicted the 1-year social-functioning outcomes with a balanced accuracy of 76.9% of patients in CHR states and 66.2% of patients with ROD using clinical baseline data. Balanced accuracy in models using structural neuroimaging was 76.2% in patients in CHR states and 65.0% in patients with ROD, and in combined models, it was 82.7% for CHR states and 70.3% for ROD. Lower functioning before study entry was a transdiagnostic predictor. Medial prefrontal and temporo-parieto-occipital gray matter volume (GMV) reductions and cerebellar and dorsolateral prefrontal GMV increments had predictive value in the CHR group; reduced mediotemporal and increased prefrontal-perisylvian GMV had predictive value in patients with ROD. Poor prognoses were associated with increased risk of psychotic, depressive, and anxiety disorders at follow-up in patients in the CHR state but not ones with ROD. Machine learning outperformed expert prognostication. Adding neuroimaging machine learning to clinical machine learning provided a 1.9-fold increase of prognostic certainty in uncertain cases of patients in CHR states, and a 10.5-fold increase of prognostic certainty for patients with ROD.

Conclusions and Relevance Precision medicine tools could augment effective therapeutic strategies aiming at the prevention of social functioning impairments in patients with CHR states or with ROD.
Original languageEnglish
Article number1156
JournalJAMA Psychiatry
Volume75
Issue number11
Early online date26 Sep 2018
DOIs
Publication statusPublished - 1 Nov 2018

Fingerprint

Psychotic Disorders
Depression
Neuroimaging
Machine Learning
Precision Medicine
Structural Models
Depressive Disorder
Anxiety Disorders
Healthy Volunteers
Outcome Assessment (Health Care)

Cite this

Koutsouleris, Nikolaos ; Kambeitz-ilankovic, Lana ; Ruhrmann, Stephan ; Rosen, Marlene ; Ruef, Anne ; Dwyer, Dominic B. ; Paolini, Marco ; Chisholm, Katharine ; Kambeitz, Joseph ; Haidl, Theresa ; Schmidt, André ; Gillam, John ; Schultze-lutter, Frauke ; Falkai, Peter ; Reiser, Maximilian ; Riecher-rössler, Anita ; Upthegrove, Rachel ; Hietala, Jarmo ; Salokangas, Raimo K. R. ; Pantelis, Christos ; Meisenzahl, Eva ; Wood, Stephen J. ; Beque, Dirk ; Brambilla, Paolo ; Borgwardt, Stefan. / Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression : A Multimodal, Multisite Machine Learning Analysis. In: JAMA Psychiatry. 2018 ; Vol. 75, No. 11.
@article{b84a9169fd3e4d9da840a9e431274ad7,
title = "Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis",
abstract = "Importance Social and occupational impairments contribute to the burden of psychosis and depression. There is a need for risk stratification tools to inform personalized functional-disability preventive strategies for individuals in at-risk and early phases of these illnesses.Objective To determine whether predictors associated with social and role functioning can be identified in patients in clinical high-risk (CHR) states for psychosis or with recent-onset depression (ROD) using clinical, imaging-based, and combined machine learning; assess the geographic, transdiagnostic, and prognostic generalizability of machine learning and compare it with human prognostication; and explore sequential prognosis encompassing clinical and combined machine learning.Design, Setting, and Participants This multisite naturalistic study followed up patients in CHR states, with ROD, and with recent-onset psychosis, and healthy control participants for 18 months in 7 academic early-recognition services in 5 European countries. Participants were recruited between February 2014 and May 2016, and data were analyzed from April 2017 to January 2018.Main Outcomes and Measures Performance and generalizability of prognostic models.Results A total of 116 individuals in CHR states (mean [SD] age, 24.0 [5.1] years; 58 [50.0{\%}] female) and 120 patients with ROD (mean [SD] age, 26.1 [6.1] years; 65 [54.2{\%}] female) were followed up for a mean (SD) of 329 (142) days. Machine learning predicted the 1-year social-functioning outcomes with a balanced accuracy of 76.9{\%} of patients in CHR states and 66.2{\%} of patients with ROD using clinical baseline data. Balanced accuracy in models using structural neuroimaging was 76.2{\%} in patients in CHR states and 65.0{\%} in patients with ROD, and in combined models, it was 82.7{\%} for CHR states and 70.3{\%} for ROD. Lower functioning before study entry was a transdiagnostic predictor. Medial prefrontal and temporo-parieto-occipital gray matter volume (GMV) reductions and cerebellar and dorsolateral prefrontal GMV increments had predictive value in the CHR group; reduced mediotemporal and increased prefrontal-perisylvian GMV had predictive value in patients with ROD. Poor prognoses were associated with increased risk of psychotic, depressive, and anxiety disorders at follow-up in patients in the CHR state but not ones with ROD. Machine learning outperformed expert prognostication. Adding neuroimaging machine learning to clinical machine learning provided a 1.9-fold increase of prognostic certainty in uncertain cases of patients in CHR states, and a 10.5-fold increase of prognostic certainty for patients with ROD.Conclusions and Relevance Precision medicine tools could augment effective therapeutic strategies aiming at the prevention of social functioning impairments in patients with CHR states or with ROD.",
author = "Nikolaos Koutsouleris and Lana Kambeitz-ilankovic and Stephan Ruhrmann and Marlene Rosen and Anne Ruef and Dwyer, {Dominic B.} and Marco Paolini and Katharine Chisholm and Joseph Kambeitz and Theresa Haidl and Andr{\'e} Schmidt and John Gillam and Frauke Schultze-lutter and Peter Falkai and Maximilian Reiser and Anita Riecher-r{\"o}ssler and Rachel Upthegrove and Jarmo Hietala and Salokangas, {Raimo K. R.} and Christos Pantelis and Eva Meisenzahl and Wood, {Stephen J.} and Dirk Beque and Paolo Brambilla and Stefan Borgwardt",
year = "2018",
month = "11",
day = "1",
doi = "10.1001/jamapsychiatry.2018.2165",
language = "English",
volume = "75",
journal = "JAMA Psychiatry",
issn = "2168-622X",
publisher = "American Medical Association",
number = "11",

}

Koutsouleris, N, Kambeitz-ilankovic, L, Ruhrmann, S, Rosen, M, Ruef, A, Dwyer, DB, Paolini, M, Chisholm, K, Kambeitz, J, Haidl, T, Schmidt, A, Gillam, J, Schultze-lutter, F, Falkai, P, Reiser, M, Riecher-rössler, A, Upthegrove, R, Hietala, J, Salokangas, RKR, Pantelis, C, Meisenzahl, E, Wood, SJ, Beque, D, Brambilla, P & Borgwardt, S 2018, 'Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis', JAMA Psychiatry, vol. 75, no. 11, 1156. https://doi.org/10.1001/jamapsychiatry.2018.2165

Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression : A Multimodal, Multisite Machine Learning Analysis. / Koutsouleris, Nikolaos; Kambeitz-ilankovic, Lana; Ruhrmann, Stephan; Rosen, Marlene; Ruef, Anne; Dwyer, Dominic B.; Paolini, Marco; Chisholm, Katharine; Kambeitz, Joseph; Haidl, Theresa; Schmidt, André; Gillam, John; Schultze-lutter, Frauke; Falkai, Peter; Reiser, Maximilian; Riecher-rössler, Anita; Upthegrove, Rachel; Hietala, Jarmo; Salokangas, Raimo K. R.; Pantelis, Christos; Meisenzahl, Eva; Wood, Stephen J.; Beque, Dirk; Brambilla, Paolo; Borgwardt, Stefan.

In: JAMA Psychiatry, Vol. 75, No. 11, 1156, 01.11.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression

T2 - A Multimodal, Multisite Machine Learning Analysis

AU - Koutsouleris, Nikolaos

AU - Kambeitz-ilankovic, Lana

AU - Ruhrmann, Stephan

AU - Rosen, Marlene

AU - Ruef, Anne

AU - Dwyer, Dominic B.

AU - Paolini, Marco

AU - Chisholm, Katharine

AU - Kambeitz, Joseph

AU - Haidl, Theresa

AU - Schmidt, André

AU - Gillam, John

AU - Schultze-lutter, Frauke

AU - Falkai, Peter

AU - Reiser, Maximilian

AU - Riecher-rössler, Anita

AU - Upthegrove, Rachel

AU - Hietala, Jarmo

AU - Salokangas, Raimo K. R.

AU - Pantelis, Christos

AU - Meisenzahl, Eva

AU - Wood, Stephen J.

AU - Beque, Dirk

AU - Brambilla, Paolo

AU - Borgwardt, Stefan

PY - 2018/11/1

Y1 - 2018/11/1

N2 - Importance Social and occupational impairments contribute to the burden of psychosis and depression. There is a need for risk stratification tools to inform personalized functional-disability preventive strategies for individuals in at-risk and early phases of these illnesses.Objective To determine whether predictors associated with social and role functioning can be identified in patients in clinical high-risk (CHR) states for psychosis or with recent-onset depression (ROD) using clinical, imaging-based, and combined machine learning; assess the geographic, transdiagnostic, and prognostic generalizability of machine learning and compare it with human prognostication; and explore sequential prognosis encompassing clinical and combined machine learning.Design, Setting, and Participants This multisite naturalistic study followed up patients in CHR states, with ROD, and with recent-onset psychosis, and healthy control participants for 18 months in 7 academic early-recognition services in 5 European countries. Participants were recruited between February 2014 and May 2016, and data were analyzed from April 2017 to January 2018.Main Outcomes and Measures Performance and generalizability of prognostic models.Results A total of 116 individuals in CHR states (mean [SD] age, 24.0 [5.1] years; 58 [50.0%] female) and 120 patients with ROD (mean [SD] age, 26.1 [6.1] years; 65 [54.2%] female) were followed up for a mean (SD) of 329 (142) days. Machine learning predicted the 1-year social-functioning outcomes with a balanced accuracy of 76.9% of patients in CHR states and 66.2% of patients with ROD using clinical baseline data. Balanced accuracy in models using structural neuroimaging was 76.2% in patients in CHR states and 65.0% in patients with ROD, and in combined models, it was 82.7% for CHR states and 70.3% for ROD. Lower functioning before study entry was a transdiagnostic predictor. Medial prefrontal and temporo-parieto-occipital gray matter volume (GMV) reductions and cerebellar and dorsolateral prefrontal GMV increments had predictive value in the CHR group; reduced mediotemporal and increased prefrontal-perisylvian GMV had predictive value in patients with ROD. Poor prognoses were associated with increased risk of psychotic, depressive, and anxiety disorders at follow-up in patients in the CHR state but not ones with ROD. Machine learning outperformed expert prognostication. Adding neuroimaging machine learning to clinical machine learning provided a 1.9-fold increase of prognostic certainty in uncertain cases of patients in CHR states, and a 10.5-fold increase of prognostic certainty for patients with ROD.Conclusions and Relevance Precision medicine tools could augment effective therapeutic strategies aiming at the prevention of social functioning impairments in patients with CHR states or with ROD.

AB - Importance Social and occupational impairments contribute to the burden of psychosis and depression. There is a need for risk stratification tools to inform personalized functional-disability preventive strategies for individuals in at-risk and early phases of these illnesses.Objective To determine whether predictors associated with social and role functioning can be identified in patients in clinical high-risk (CHR) states for psychosis or with recent-onset depression (ROD) using clinical, imaging-based, and combined machine learning; assess the geographic, transdiagnostic, and prognostic generalizability of machine learning and compare it with human prognostication; and explore sequential prognosis encompassing clinical and combined machine learning.Design, Setting, and Participants This multisite naturalistic study followed up patients in CHR states, with ROD, and with recent-onset psychosis, and healthy control participants for 18 months in 7 academic early-recognition services in 5 European countries. Participants were recruited between February 2014 and May 2016, and data were analyzed from April 2017 to January 2018.Main Outcomes and Measures Performance and generalizability of prognostic models.Results A total of 116 individuals in CHR states (mean [SD] age, 24.0 [5.1] years; 58 [50.0%] female) and 120 patients with ROD (mean [SD] age, 26.1 [6.1] years; 65 [54.2%] female) were followed up for a mean (SD) of 329 (142) days. Machine learning predicted the 1-year social-functioning outcomes with a balanced accuracy of 76.9% of patients in CHR states and 66.2% of patients with ROD using clinical baseline data. Balanced accuracy in models using structural neuroimaging was 76.2% in patients in CHR states and 65.0% in patients with ROD, and in combined models, it was 82.7% for CHR states and 70.3% for ROD. Lower functioning before study entry was a transdiagnostic predictor. Medial prefrontal and temporo-parieto-occipital gray matter volume (GMV) reductions and cerebellar and dorsolateral prefrontal GMV increments had predictive value in the CHR group; reduced mediotemporal and increased prefrontal-perisylvian GMV had predictive value in patients with ROD. Poor prognoses were associated with increased risk of psychotic, depressive, and anxiety disorders at follow-up in patients in the CHR state but not ones with ROD. Machine learning outperformed expert prognostication. Adding neuroimaging machine learning to clinical machine learning provided a 1.9-fold increase of prognostic certainty in uncertain cases of patients in CHR states, and a 10.5-fold increase of prognostic certainty for patients with ROD.Conclusions and Relevance Precision medicine tools could augment effective therapeutic strategies aiming at the prevention of social functioning impairments in patients with CHR states or with ROD.

UR - http://archpsyc.jamanetwork.com/article.aspx?doi=10.1001/jamapsychiatry.2018.2165

U2 - 10.1001/jamapsychiatry.2018.2165

DO - 10.1001/jamapsychiatry.2018.2165

M3 - Article

VL - 75

JO - JAMA Psychiatry

JF - JAMA Psychiatry

SN - 2168-622X

IS - 11

M1 - 1156

ER -