Classification of Adolescent Psychiatric Patients at High Risk of Suicide Using the Personality Assessment Inventory by Machine Learning

Abstract

Objective There are growing interests on suicide risk screening in clinical settings and classifying high-risk groups of suicide with suicidal ideation is crucial for a more effective suicide preventive intervention. Previous statistical techniques were limited because they tried to predict suicide risk using a simple algorithm. Machine learning differs from the traditional statistical techniques in that it generates the most optimal algorithm from various predictors.Methods We aim to analyze the Personality Assessment Inventory (PAI) profiles of child and adolescent patients who received outpatient psychiatric care using machine learning techniques, such as logistic regression (LR), random forest (RF), artificial neural network (ANN), support vector machine (SVM), and extreme gradient boosting (XGB), to develop and validate a classification model for individuals with high suicide risk.Results We developed prediction models using seven relevant features calculated by Boruta algorithm and subsequently tested all models using the testing dataset. The area under the ROC curve of these models were above 0.9 and the RF model exhibited the best performance.Conclusion Suicide must be assessed based on multiple aspects, and although Personality Assessment Inventory for Adolescent assess an array of domains, further research is needed for predicting high suicide risk groups.

Machine learning and the prediction of suicide in psychiatric populations: a systematic review

“…Most relevant features: onset age, Parietal, VLPFC/OFC, DMPFC, Precuneus, and DLPFC GMV. Kim et al, [ 76 ] 44 (19/25) high-risk; 80 (34/46) low-risk adolescents Adolescents with a psychiatric diagnoses (5 psychotic, 40 mood disorders, 42 ANX, 16 ADHD, 35 others) Not specified LR, RF, ANN, SVM, XGB (10-folds CV) 256 clinical and sociodemographic features Low vs hig risk, based on suicide scale of the PAI-A LR Acc: 0.89 Sens: 0.77 Spec: 0.96 RF Acc: 0.89 Sens: 0.92 Spec: 0.88 ANN Acc: 0.78 Sens: 0.77 Spec: 0.79 SVM Acc: 0.89 Sens: 0.85 Spec: 0.92 XGB Acc: 0.86 Sens: 0.92 Spec: 0.83 Most relevant features from the PAI-A contributing to the predictions: anxiety and anxiety-related scores; depression; nonsupport in social life; treatment rejection. Ji et al, [ 74 ] 44 (34/10) 48 (28/20) MDD with suicide; MDD without suicide; 51 (28/23) HC MDD Not specified SVM, AdaBoost, NB (10-folds CV) 105 clinical variables from clinical questionnaires (BDI, BSI, BHS, TDPPS, TEPS, STAI, BIS) Suicide SVM Acc: 0.88 Sens: 0.87 Spec: 0.89 AdaBoost Acc: 0.82 Sens: 0.87 Spec: 0.78 NB Acc: 0.82 Sens: 0.75 Spec: 0.89 Clinical questionnaires features selected with LR proved to be able to classify the groups with good accuracy.…”

“…In the studies that compared more than one algorithm, ML methods always performed better than LR. Moreover, RF [ 32 , 57 , 73 ] and SVM [ 74 , 75 ] resulted among the best-performing algorithms, often with comparable results [ 65 , 76 ], when compared to other methods. Finally, when present, CNN outperformed other ML methods [ 49 , 50 , 62 , 77 ], including SVM and RF (please see Supplementary Table 4 for further details).…”

Risk Factors and Trends in Adolescent’s Suicide Attempt Rates Before and After the Coronavirus Disease 2019 Pandemic

“…In addition, while this study addressed risk factors associated with the COVID era, it did not account for other confounding factors, including adverse childhood experiences, personality, and protective factors. 6 13 37 …”

“…Adolescence is a vulnerable period in which individuals undergo significant transitions. 5 6 Mental health issues are more common than physical issues during this period and can impact individuals’ mental health status into adulthood. 7 During adolescence, individuals establish the direction of their behaviors and develop self-concept and self-worth based on their experiences of peer acceptance and rejection, highlighting the importance of peer interactions for mental health.…”

Suicides in Turkey: 25‐year trend (1995–2019)

“…In most studies that have investigated the relationship between education level and suicide, the risk of suicide has been shown to be higher in people with low education level, especially in males, and that the risk of suicide decreases with a higher educational level [40][41][42][43][44][45]. In some studies, it has been stated that suicidal tendency increases inversely with low IQ level and decreased ability to compete for jobs and thus a reduced ability to cope with economic difficulties and solve problems in these people with low IQ may have an impact on high suicide rates [46][47][48]. Although it was not possible to evaluate directly in the current study, it has been stated that the rate of committing crimes is higher among people with low IQ, due to the low academic achievement levels [49].…”