Towards a machine-learning assisted diagnosis of psychiatric disorders and their operationalization in preclinical research: Evidence from studies on addiction-like behaviour in individual rats

Abstract

Over the last few decades, there has been a progressive transition from a categorical to a dimensional approach to psychiatric disorders. Especially in the case of substance use disorders, interest in the individual vulnerability to transition from controlled to compulsive drug taking warrants the development of novel dimension-based objective stratification tools. Here we drew on a multidimensional preclinical model of addiction, namely the 3-criteria model, previously developed to identify the neurobehavioural basis of the individual's vulnerability to switch from controlled to compulsive drug taking, to test a machine-learning assisted classifier objectively to identify individual subjects as vulnerable/resistant to addiction. Datasets from our previous studies on addiction-like behaviour for cocaine or alcohol were fed into a variety of machine-learning algorithms to develop a classifier that identifies resilient and vulnerable rats with high precision and reproducibility irrespective of the cohort to which they belong. A classifier based on K-median or K-mean-clustering (for cocaine or alcohol, respectively) followed by artificial neural networks emerged as a highly reliable and accurate tool to predict if a single rat is vulnerable/resilient to addiction. Thus, each rat previously characterized as displaying 0-criterion (i.e., resilient) or 3-criteria (i.e., vulnerable) in individual cohorts was correctly labelled by this classifier. The present machine-learning-based classifier objectively labels single individuals as resilient or vulnerable to developing addiction-like behaviour in a multisymptomatic preclinical model of addiction-like behaviour in rats. This novel dimension-based classifier increases the heuristic value of these preclinical models while providing proof of principle to deploy similar tools for the future of diagnosis of psychiatric disorders.

Keywords: addiction; clustering; individual vulnerability; machine learning; neural networks; substance use disorder.

FIGURE 1

Workflow of the machine learning classifier. The steps are illustrated as numbers in the circles. Clustering algorithms used are Gaussian mixture method and K‐mean/K‐median clustering. Classification algorithms used are K nearest neighbor, logistic regression, support vector machines and artificial neural networks. The blue arrows indicate the clustering algorithms and the green arrows indicate the classification algorithms

FIGURE 2

Illustration of an artificial neural network. The hidden layer consists of neurons with the ELU (exponential linear unit) activation function. Feed forward network entails multiple forward passes through the hidden layers. One forward pass consists of consecutive matrix multiplications at each layer by utilizing random weights to initialize the training, which are then adjusted during backpropagation to minimize the cross entropy loss function. Maintained drug use despite aversive consequences, increased motivation to take the drug and inability to refrain from drug seeking during signalled unavailability are represented as behaviour 1, behaviour 2 and behaviour 3, respectively.

FIGURE 3

Performance evaluation metrics of the machine learning classifier of the addiction‐like behavior for cocaine in rats. FIGURE 3A–3D depict the accuracy, precision, recall and ROC AUC score, respectively of the GMM clustering‐based classifier followed by four supervised machine learning algorithms. 3E–3H depict the accuracy, precision, recall and ROC AUC score respectively of the K‐median clustering‐based classifier followed by the four supervised machine learning algorithms. GMM: Gaussian mixture method, ML: Machine learning, KNN: K‐nearest neighbor, LR: Logistic regression, SVM: Support vector machines, ANN: Artificial neural networks.

FIGURE 4

Performance evaluation metrics of the machine learning classifier of the addiction‐like behavior for alcohol in rats. FIGURE 4A‐D depict the accuracy, precision, recall and ROC AUC scores respectively of the GMM clustering based classifier followed by four supervised machine learning algorithms. 4E–4H depict the accuracy, precision, recall and ROC AUC scores respectively of the K‐mean clustering‐based classifier followed by four supervised machine learning algorithms. GMM: Gaussian mixture method, ML: Machine learning, KNN: K nearest neighbor, LR: Logistic regression, SVM: Support vector machines, ANN: Artificial neural networks.

FIGURE 5

Flowchart for the labelling of any future rat as resilient or vulnerable to develop substance use disorder. The steps are illustrated as numbers in the circles. Having established that the classifier based on K‐median/K‐mean clustering followed by ANN gives the best predictive accuracy, the addiction vulnerability status of a single rat irrespective of the cohort it is trained with. The blue arrows indicate the clustering algorithms and the green arrows indicate the classification algorithms. ANN: Artificial neural network