Factores de predicción para la detección temprana de trastornos neurocognitivos en personas institucionalizadas en centros de protección social pertenecientes a la beneficencia de Cundinamarca

Abstract

Neurocognitive disorders are increasing significantly in the elderly population. elderly population. From this point of view, it is important to understand this phenomenon as the result of a multiplicity of variables as the result of a multiplicity of variables in which the social context and intellectual impairment and intellectual impairment establish elements of study in line with the identification of risk and/or contributory factors. risk and/or contributory factors. Individuals with intellectual developmental disorder with mild impairment are more likely to be with mild impairment are more likely to acquire a specific type of dementia, which generally begins to show early signs of dementia, which usually begins to show early signs. The present study aimed to identify potential cases in people institutionalized in Social Protection Centers of the Beneficencia de Cundinamarca. of the Beneficencia de Cundinamarca in order to establish early therapeutic actions to control the to control mental deterioration and the costs associated with the treatments. The study took into account 32 predictor variables obtained through the Wechsler Scale of Intelligence. Wechsler Adult Intelligence Scale and Mini-Mental State Examination. Based on the analysis of existing data in the Social Protection Centers of the Beneficencia de Cundinamarca in 2018, in Cundinamarca in 2018, 2020 and 2022 (n = 294), statistical models were trained which divided the objective variable: diagnosis of dementia in 2022 into independent components oriented to the oriented to the prediction of potential cases according to the prevalence of some type of according to the prevalence of some type of impairment. Once the data were processed, the classification model that performed best was the Random classification model that presented the best performance was the Random Forest Classifier, yielding a 2 AUC of 0.97, 0.01 higher than the Light Gradient Boosting Machine, and 0.1 higher than the Logisctic Regression initially proposed. The variables that best contributed to the prediction were related to the evocative memory were related to evocative memory, temporal organization, immediate memory, attention and calculation, baseline diagnosis, executive quotient and age.