Application of Machine Learning Techniques to Help in the Feature Selection Related to Hospital Readmissions of Suicidal Behavior

Suicide was the main source of death from external causes in Spain in 2020, with 3,941 cases. The importance of identifying those mental disorders that influenced hospital readmissions will allow us to manage the health care of suicidal behavior. The feature selection of each hospital in this region was carried out by applying Machine learning (ML) and traditional statistical methods. The results of the characteristics that best explain the readmissions of each hospital after assessment by the psychiatry specialist are presented. Adjustment disorder, alcohol abuse, depressive syndrome, personality disorder, and dysthymic disorder were selected for this region. The most influential methods or characteristics associated with suicide were benzodiazepine poisoning, suicidal ideation, medication poisoning, antipsychotic poisoning, and suicide and/or self-harm by jumping. Suicidal behavior is a concern in our society, so the results are relevant for hospital management and decision-making for its prevention.

Automated summary

In 2020, suicide was the leading cause of death from external causes in Spain. It is important to identify mental disorders that contribute to hospital readmissions in order to effectively manage suicidal behavior. Machine learning and traditional statistical methods were used to select the characteristics of each hospital that explained readmissions. These characteristics included adjustment disorder, alcohol abuse, depressive syndrome, personality disorder, and dysthymic disorder. The most influential methods or characteristics associated with suicide were benzodiazepine poisoning, suicidal ideation, medication poisoning, antipsychotic poisoning, and suicide and/or self-harm by jumping.