Machine learning regression models for internal shame.

Area of Science:

• Psychology
• Machine Learning
• Clinical Psychology

Background:

• Internal Shame (IS) is a significant psychological construct.
• Understanding predictors of IS is crucial for mental health interventions.
• Childhood trauma and emotional regulation skills are implicated in IS development.

Purpose of the Study:

• To predict Internal Shame (IS) using various psychological factors.
• To evaluate the performance of multiple machine learning models for IS prediction.
• To identify the most significant predictors of IS in an Iranian population.

Main Methods:

• A sample of Iranian individuals was assessed for childhood trauma, social emotional competence, cognitive flexibility, distress tolerance, and alexithymia.
• Nine machine learning regression techniques were initially employed.
• Model performance was evaluated, leading to the selection of five models for detailed analysis, including XGBoost.

Main Results:

• Distress tolerance was identified as the most significant predictor of Internal Shame (IS).
• Cognitive flexibility showed the least impact on Internal Shame.
• The XGBoost regression model exhibited superior predictive performance compared to other evaluated machine learning methods.

Conclusions:

• Distress tolerance is a key factor in the development of Internal Shame.
• Machine learning, particularly XGBoost, offers a powerful approach for predicting Internal Shame.
• These findings have implications for targeted therapeutic interventions for individuals experiencing Internal Shame.