Screening of Mood Symptoms Using MMPI-2-RF Scales: An Application of Machine Learning Techniques

Area of Science:

• Psychiatry
• Psychology
• Machine Learning

Background:

• Mood disorders are prevalent globally, often presenting diagnostic challenges and poor prognoses.
• The Minnesota Multiphasic Personality Inventory-2-Restructured Form (MMPI-2-RF) is a widely validated tool for psychopathology assessment.
• Early detection of mood disorders remains difficult, impacting treatment outcomes.

Purpose of the Study:

• To evaluate the efficacy of machine learning techniques in predicting mood symptoms.
• To compare the predictive accuracy of MMPI-2-RF scales against traditional Restructured Clinical scales for mood disorders.
• To enhance the classification and early detection of depressive and hypomanic symptoms.

Main Methods:

• Analysis of 8645 participants' data.
• Utilized the Patient Health Questionnaire-9 (PHQ-9) for depressive symptoms and the Mood Disorder Questionnaire (MDQ) for hypomanic symptoms.
• Employed machine learning algorithms including k-nearest neighbor, linear discriminant analysis, and random forest classification on MMPI-2-RF scales.

Main Results:

• Machine learning predicted depressive symptoms with an Area Under the Curve (AUC) of 0.634-0.767 and hypomanic symptoms with an AUC of 0.770-0.840.
• Specific MMPI-2-RF scales (RCd for depression, RC9 for hypomania) combined with linear discriminant analysis yielded high AUC values (0.840 and 0.767, respectively).
• Machine learning models demonstrated superior prediction accuracy for mood symptoms compared to standard Restructured Clinical scales.

Conclusions:

• Machine learning analysis significantly improves the classification and prediction of mood symptoms.
• The MMPI-2-RF, when analyzed with machine learning, offers enhanced diagnostic capabilities for mood disorders.
• This study highlights the potential of AI-driven approaches for early detection and management of mood disorders.