Detection of formal thought disorders in child and adolescent psychosis using machine learning and neuropsychometric data

Aug 3, 2026

Upload by Rowan

The study evaluates whether neuropsychological testing combined with machine learning can distinguish formal thought disorder (FTD) in early-onset psychosis from cases without FTD. A cohort of 27 young participants underwent Iowa Gambling Task and Simple Reaction Time assessments, alongside medication load expressed in chlorpromazine equivalents. Multiple classifiers, including logistic regression, support vector machines, random forest, and XGBoost, were trained and cross-validated. Logistic regression achieved moderate-to-good discrimination (ROC AUC 0.850), supported by task performance and reaction-time variability, enabling timely identification.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages8

Reading Time20 min

AccessFree to download

Introduction

Formal thought disorders and early-onset psychosis

Limitations of clinical assessment

Methods

Participants and study design

Neuropsychological tasks and features

Machine learning classifiers

Results

Model performance and key predictive features

Discussion

Clinical feasibility and implications

Future validation needs

TYPE Brief Research Report PUBLISHED 17 March 2025 DOI 10.3389/fpsyt.2025.1550571 OPEN ACCESS EDITED BY Amit Singhal, Netaji Subhas University of Technology, India REVIEWED BY Animesh Kumar Paul, University of Alberta, Canada Alexandre Hudon, Montreal University, Canada *CORRESPONDENCE Przemysław T. Zakowicz [pzakowicz@neurologicalsociety.org](pzakowicz@neurologicalsociety.org) †These authors have contributed equally to this work and share ﬁrst authorship RECEIVED 23 December 2024 ACCEPTED 03 March 2025 PUBLISHED 17 March 2025 CITATION Zakowicz PT, Brzezicki MA, Levidiotis C, Kim S, Wejkuc´ O, Wisniewska Z, Biernaczyk D and Remberk B (2025) Detection of formal thought disorders in child and adolescent psychosis using machine learning and neuropsychometric data. Front. Psychiatry 16:1550571 . doi: 10.3389/fpsyt.2025.1550571 COPYRIGHT © 2025 Zakowicz, Brzezicki, Levidiotis, Kim, Wejkuc´, Wisniewska, Biernaczyk and Remberk. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Detection of formal thought disorders in child and adolescent psychosis using machine learning and neuropsychometric data Przemysław T. Zakowicz 1*†, Maksymilian A. Brzezicki 2†, Charalampos Levidiotis 3, Sojeong Kim 3, Oskar Wejku´c1, Zuzanna Wisniewska 1, Dominika Biernaczyk 1 and Barbara Remberk 4 1Collegium Medicum, University of Zielona Gora, Zielona Góra, Poland, 2 Nufﬁeld Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom, 3 Department of Medicine, Poznan University of Medical Sciences, Poznan, Poland, 4 Department of Child and Adolescent Psychiatry, Institute of Psychiatry and Neurology, Warsaw, Poland Introduction: Formal Thought Disorder (FTD) is a signiﬁcant clinical feature of earlyonset psychosis, often associated with poorer outcomes. Current diagnostic methods rely on clinical assessment, which can be subjective and timeconsuming. This study aimed to investigate the potential of neuropsychological tests and machine learning to differentiate individuals with and without FTD. Methods: A cohort of 27 young people with early-onset psychosis was included. Participants underwent neuropsychological assessment using the Iowa Gambling Task (IGT) and Simple Reaction Time (SRT) tasks. A range of machine learning models (Logistic Regression (LR), Support Vector Machines (SVM), Random Forest (RF) and eXtreme Gradient Boosting (XGBoost)) were employed to classify participants into FTD-positive and FTD-negative groups based on these neuropsychological measures and their antipsychotic regimen (medication load in chlorpromazine equivalents) . Results: The best performing machine learning model was LR with mean +/ -standard deviation of cross validation Receiver Operating Characteristic Area Under Curve (ROC AUC) score of 0 . 850 (+/ -0. 133), indicating moderate-togood discriminatory performance. Key features contributing to the model ’s accuracy included IGT card selections, SRT reaction time (most notably standard deviation) and chlorpromazine equivalent milligrams. The model correctly classiﬁed 24 out of 27 participants. Discussion: This study demonstrates the feasibility of using neuropsychological tests and machine learning to identify FTD in early-onset psychosis. Early identiﬁcation of FTD may facilitate targeted interventions and improve clinical outcomes. Further research is needed to validate these ﬁndings in larger, more diverse populations and to explore the underlying neurocognitive mechanisms. KEYWORDS child and adolescent psychosis, machine learning, cognitive psyc

FAQs

What is the study trying to detect in young patients with psychosis?

The study aims to detect formal thought disorders (FTD) in child and adolescent/early-onset psychosis by distinguishing FTD-positive from FTD-negative individuals.

Which neuropsychological measures and data were used for the machine learning models?

Participants completed the Iowa Gambling Task (IGT) and Simple Reaction Time (SRT) tasks, and the models also used antipsychotic regimen information expressed as medication load in chlorpromazine equivalents.

How well did the best model perform, and what contributed most to its accuracy?

Logistic regression performed best with a cross-validation ROC AUC of 0.850 (moderate-to-good discrimination). Key contributors included IGT card selections, SRT reaction-time variability (notably standard deviation), and chlorpromazine-equivalent milligrams.

Aug 3, 2026

Upload by Rowan