Identifying cognitive deficits in cocaine dependence using standard tests and machine learning

Abstract

There is a growing need to address the variability in detecting cognitive deficits with standard tests in cocaine dependence (CD). The aim of the current study was to identify cognitive deficits by means of Machine Learning (ML) algorithms: Generalized Linear Model (Glm), Random forest (Rf) and Elastic Net (GlmNet), to allow more effective categorization of CD and Non-dependent controls (NDC and to address common methodological problems. For our validation, we used two independent datasets, the first consisted of 87 participants (53 CD and 34 NDC) and the second of 40 participants (20 CD and 20 NDC). All participants were evaluated with neuropsychological tests that included 40 variables assessing cognitive domains.-Using results from the cognitive evaluation, the three ML algorithms were trained in the first dataset and tested on the second to classify participants into CD and NDC. While the three algorithms had a receiver operating curve (ROC) performance over 50%, the GlmNet was superior in both the training (ROC = 0.71) and testing datasets (ROC = 0.85) compared to Rf and Glm. Furthermore, GlmNet was capable of identifying the eight main predictors of group assignment (CD or NCD) from all the cognitive domains assessed. Specific variables from each cognitive test resulted in robust predictors for accurate classification of new cases, such as those from cognitive flexibility and inhibition domains. These findings provide evidence of the effectiveness of ML as an approach to highlight relevant sections of standard cognitive tests in CD, and for the identification of generalizable cognitive markers.

Driving under the influence of drugs: Correlation between blood psychoactive drug concentrations and cognitive impairment. A narrative review taking into account forensic issues

“…A further concern is that in the literature sometimes it is assumed that statistically significant differences are also clinically significant: this leads to conclusion about impairments that are therefore based on statistical differences with respect to a limited number of tasks [ 137 ] with no comparison against a normative baseline that takes into account the demographic characteristics of the individual. Previous studies also stressed that other problems are related to the lack of clarity as to which task subtest are adequately assessed and inadequate statistical analysis [ 173 ]. Furthermore, it should be noted that the culpability status is almost always unknown, creating an interpretative bias caused by researchers treating culpability ORs as equivalent to crash ORs [ 174 , 175 ], an issue of pivotal importance in the interpretation of previous literature.…”

Enhancing data pipelines for forecasting student performance: integrating feature selection with cross-validation

“…The ensemble technique GLMNET significantly outperformed all other DMMs and improved the AUC metric when feature selection was applied as a preprocessing step to select six features. GLMNET is advantageous in binary classification due to its ability to select or exclude correlated covariates, and its use of regularization parameters to constrain the magnitudes of coefficients (Bertolini & Finch, in press;Jiménez et al, 2019;Kirpich et al, 2018;Lu & Petkova, 2014). However, it is important to note that some applied research found that this method performed worse than other ensemble techniques such as RF (Alexandro, 2018;Ransom et al, 2019).…”

SUDMEX CONN: The Mexican MRI dataset of patients with cocaine use disorder

“…Previous publications have used this dataset using diffusion kurtosis imaging analysis 11 , neurite orientation dispersion and density imaging analysis 12 , and a machine learning approach for the identification of cognitive markers in CUD 13 , among others , [14][15][16][17] . Overall, this dataset could contribute to the in-depth study of substance use disorders, particularly cocaine use.…”