Table 1. Artificial intelligence (AI) techniques and uses in pharmacological research.

LASSO: Least Absolute Shrinkage and Selection Operation

Technique	Methods	Process
Machine learning	Support vector machine (SVM)	It finds the best hyperplane that separates different classes with the maximum margin, allowing effective classification even in complex datasets. Identifies decision boundaries in high-dimensional data and can handle non-linear relationships. It is used for classification and regression tasks in pharmacological research.
	Random forest	An ensemble learning method, that works by combining multiple decision trees to improve predictive accuracy, feature importance analysis, and identification of potential drug candidates. Primarily used for feature selection and classification tasks in drug discovery and toxicity prediction.
	Supervised learning	It involves training models with labeled data to yield desired outputs, enabling accurate classification and prediction tasks through algorithms, such as neural networks, support vector machines, and random forests. Drug-target interaction prediction, virtual screening, and toxicity prediction are its main uses.
	Unsupervised learning principal component analysis (PCA)	It involves finding patterns and structures in unlabeled data without explicit human supervision. It includes tasks like clustering, association, and dimensionality reduction to discover insights and extract meaningful information from the data. It is utilized for target identification, lead identification and optimization, pharmacovigilance and adverse drug event (ADE) detection, drug repurposing, bioactivity prediction
	Reinforcement learning	Reinforcement learning is the science of decision-making through trial and error, where an agent learns optimal behavior in an environment to maximize reward. It involves an agent exploring and interacting with an environment, learning from outcomes, and adjusting its actions based on feedback. It has great value in personalized medicine and dose optimization.
	Feature selection recursive feature elimination (RFE) LASSO regression	Recursive feature elimination (RFE) is a feature selection algorithm that iteratively eliminates less important features from a dataset based on their relevance in predicting the target variable. It starts with all features and removes them one by one until the desired number of features is reached. LASSO regression, also known as L1 regularization, is a linear regression technique that performs both feature selection and regularization by adding a penalty term to the loss function. It encourages sparsity in the coefficients, effectively shrinking less important features to zero, and keeping only the most relevant features in the model. They are primarily used to select relevant molecular or clinical descriptors for drug-target interaction prediction or patient stratification.
	Dimensionality reduction principal component analysis (PCA) t-SNE (t-distributed stochastic neighbor embedding)	Utilized to transform high-dimensional data into lower-dimensional representations while preserving essential information. It has value in E data exploration and visualization, feature selection, clustering and classification, noise reduction in data, and pre-processing for machine learning.
Deep learning	Neural networks	Mimic the structure and function of biological neurons.
	Convolutional neural networks (CNNs)	They employ convolution, a mathematical operation, to process pixel data. By breaking down images into smaller features and progressively combining them into more complex patterns, CNNs efficiently learn and extract abstract representations, minimizing overfitting. These have revolutionized image analysis tasks, enabling accurate image classification, segmentation, and object detection.
	Recurrent neural networks (RNNs)	Sequence-based tasks like protein structure prediction.