Prediction of patient admission and readmission in adults from a Colombian cohort with bipolar disorder using artificial intelligence.

Introduction: Bipolar disorder (BD) is a chronically progressive mental condition, associated with a reduced quality of life and greater disability. Patient admissions are preventable events with a considerable impact on global functioning and social adjustment. While machine learning (ML) approaches have proven prediction ability in other diseases, little is known about their utility to predict patient admissions in this pathology. Aim: To develop prediction models for hospital admission/readmission within 5 years of diagnosis in patients with BD using ML techniques. Methods: The study utilized data from patients diagnosed with BD in a major healthcare organization in Colombia. Candidate predictors were selected from Electronic Health Records (EHRs) and included sociodemographic and clinical variables. ML algorithms, including Decision Trees, Random Forests, Logistic Regressions, and Support Vector Machines, were used to predict patient admission or readmission. Survival models, including a penalized Cox Model and Random Survival Forest, were used to predict time to admission and first readmission. Model performance was evaluated using accuracy, precision, recall, F1 score, area under the receiver operating characteristic curve (AUC) and concordance index. Results: The admission dataset included 2,726 BD patients, with 354 admissions, while the readmission dataset included 352 patients, with almost half being readmitted. The best-performing model for predicting admission was the Random Forest, with an accuracy score of 0.951 and an AUC of 0.98. The variables with the greatest predictive power in the Recursive Feature Elimination (RFE) importance analysis were the number of psychiatric emergency visits, the number of outpatient follow-up appointments and age. Survival models showed similar results, with the Random Survival Forest performing best, achieving an AUC of 0.95. However, the prediction models for patient readmission had poorer performance, with the Random Forest model being again the best performer but with an AUC below 0.70. Conclusion: ML models, particularly the Random Forest model, outperformed traditional statistical techniques for admission prediction. However, readmission prediction models had poorer performance. This study demonstrates the potential of ML techniques in improving prediction accuracy for BD patient admissions.

Determinación de los factores asociados con el reingreso temprano de pacientes con enfermedad mental en dos ciudades de Colombia durante 2018 / Determination of Factors Associated with Early Readmission of Patients with Mental Illness in Two Cities in Colombia During 2018

RESUMEN Introducción: La rehospitalización temprana en unidades de salud mental (USM) es la necesidad de hospitalización de un paciente en los primeros 30 días tras el alta, principalmente por descompensación recurrente de su enfermedad mental. Este fenómeno se relaciona con un peor pronóstico y tiene impacto en el entorno familiar, social y laboral. El ausentismo laboral y las estancias hospitalarias adicionales son gastos para el sistema de salud y de empleo que han hecho de la rehospitalización un fenómeno de especial interés. El presente estudio se llevó a cabo con el objetivo de explorar los factores asociados con el reingreso de los pacientes con enfermedad psiquiátrica atendidos en 2 USM en 2018, así como aquellos modificables que actúen como protección contra esta condición. Métodos: Estudio observacional descriptivo con componente analítico de tipo casos y controles en 2 USM de distintas ciudades de Colombia. Se obtuvo información por medio de una ficha de recolección de datos tomados de los registros de historias clínicas de los pacientes que ingresaron entre el 1 de enero y el 31 de diciembre de 2018. La recolección de datos se hizo del 20 de febrero al 27 de mayo de 2019. Compusieron la muestra todos los pacientes que cumplían los criterios de reingreso temprano en ambas instituciones. El grupo de estudio estuvo conformado por 113 pacientes: 28 casos y 85 controles, emparejados por las variables edad, sexo, lugar de hospitalización y diagnóstico. Resultados: En las 2 USM hospitalarias, los diagnósticos encontrados fueron: depresión (15,5%), trastorno afectivo bipolar (33,1%) y esquizofrenia (37,3%); en Bogotá la más prevalente fue la depresión (31,1%) y en Tunja, la esquizofrenia (44,8%). Para ambas instituciones, el factor que más se asocia con el reingreso es el consumo de alcohol, pero otras variables de tratamiento, núcleo familiar e intervención individual también se asociaron con mayor probabilidad de reingreso temprano. Conclusiones: Se pudo demostrar que el uso de antipsicóticos atípicos y/o de depósito, las hospitalizaciones de más de 15 días y la prescripción de menos de 3 medicamentos al alta disminuyen el número de reingresos tempranos en las USM.

ABSTRACT Introduction: Early rehospitalisation in mental health units (SMHUs) is when a patient needs to be readmitted in the first 30 days after receiving discharge, and is mainly due to recurrent decompensation of their mental illness. This phenomenon is related to a worse prognosis and has an impact on the family, social and work environment. Absenteeism from work and additional hospital time are expenses for the health and employment system which have made rehospitalisation a phenomenon of special interest. The present study was carried out with the objective of exploring the factors associated with the readmission of patients with psychiatric illnesses treated in two MHUs during 2018, as well as those modifiable factors that act as protection for this condition. Methods: Observational, descriptive study with analytical component of cases and controls in two MHUs in different cities of Colombia. Information was obtained by collecting data from the medical records of patients who were admitted between 1 January 2018 and 31 December 2018. The data were collected between 20 February and 27 May 2019. The sample was composed of all the patients who met the criteria for early readmission in both institutions. The study group consisted of 113 patients: (28 cases and 85 controls), matched by the variables: age, sex, place of hospitalisation and diagnosis. Results: In the two hospital MHUs the diagnoses found were: depression (15.5%), bipolar affective disorder (33.1%) and schizophrenia (37.3%). In Bogotá, the most prevalent was depression (31.1%) and, in Tunja, it was schizophrenia (44.8%). For both institutions, the factor most associated with readmission was alcohol consumption, but other variables of treatment, family nucleus, and individual intervention were also associated with a greater probability of early readmission. Conclusions: It was possible to demonstrate that the use of atypical and/or depot antipsychotics, hospitalisations longer than 15 days, and prescriptions of less than three drugs at discharge, reduce the number of early readmissions to MHUs.