# Prediction of the relationship between childhood trauma and psychological disturbances among younger adults using machine learning algorithms

## TL;DR

This study uses machine learning to show that childhood trauma is strongly linked to psychological issues in young adults, with females being more affected.

## Contribution

The study introduces a machine learning approach to predict the relationship between childhood trauma and psychological disturbances in younger adults.

## Key findings

- 29% of the variance in DASS-CT links is explained by Multiple Linear Regression.
- Random Forest achieved 71% accuracy and 76% AUC in predicting severity of psychological disturbances.
- Females show greater sensitivity to psychological disturbances linked to childhood trauma.

## Abstract

Childhood Trauma (CT) is a prevalent issue with significant health implications throughout an individual’s life and is believed to affect the psychological well-being of younger adults.

To explore the relationship between CT and psychological disturbances among younger adults, including sociodemographic factors.

A descriptive cross-sectional study was conducted from October 2024 to March 2025, with a sample of 331 participants in Egypt. Data were collected via paper questionnaires assessing youth characteristics and psychological disturbances using Depression, Anxiety and Stress Scale-Short Form (DASS-21) and CT using Childhood Trauma Questionnaire (CTQ). Statistical methods, including the Chi-squared test, Machine Learning (ML) regression and classification models, were used to explore the association of Depression, Anxiety and Stress (DASS) and CT within the data, and model validations were performed for internal stability. The ML models used include Multiple Linear Regression (MLR), Random Forest (RF), Support Vector Machine (SVM), a Deep Neural Network (DNN), and Stochastic Gradient Descent (SGD). The algorithms’ performance was assessed and balanced for more realistic reporting of findings. All CTQ subscales and gender were used in analysis and the importance of the features was compared.

We find that (19.94%) and (25.68%) of the younger adults experienced moderate and severe CT, respectively. In addition, (59.82%) and (14.8%) experienced moderate and severe DASS, respectively. The MLR is highly significant (p = 0.000), elucidating \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${R}^{2}=29\%$$\end{document} variance \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$${f}^{2}=41\%)$$\end{document} in continuous DASS-CT links, while RF classification attained (71%) Accuracy and (76%) Area Under the ROC Curve (AUC) for severity prediction.

There are highly significant interrelations between DASS, CT and sex. Younger adults who experienced CT, experience some level of DASS and females exhibit greater sensitivity to DASS. The ML approach proves effective in predicting relationships between DASS, CT, and sex. Bolstered by RF’s 76% AUC for severity stratification, the findings support the practical utility of ML in guiding trauma-informed therapy and early intervention programs.