Introduction to Machine Learning in Pharmaceutical Sciences (Theory)

Foundations of Machine Learning Definition and scope of Artificial Intelligence, Machine Learning, and Data Science Types of machine learning: supervised, unsupervised, and reinforcement learning Features and labels, training data and testing data Concept of model building and prediction Train-test split, overfitting and underfitting Conceptual explanation of bias-variance trade-off Overview of the basic ML workflow

Regression Models in Healthcare Overview of the concepts in predictive modeling for continuous outcomes, with emphasis on interpretation of outputs, and applications in pharmaceutical sciences. Linear regression and multiple linear regression Model coefficients and their interpretation Residuals and goodness-of-fit Performance metrics such as Mean Squared Error (MSE), Root Mean Squared Error (RMSE), R² score Implementation of regression models on pharmaceutical data (e.g. dose response relationships, predicting drug dissolution rates, estimating PK parameters), and demonstration of predictive modeling using python libraries such as sklearn

Classification Models in Clinical Applications Overview of the following ML models used for categorical prediction, with emphasis on conceptual understanding, interpretation of outputs and applications in pharmaceutical sciences. Logistic regression Probability output and threshold selection Confusion matrix Accuracy, sensitivity, specificity, precision, recall Intuitive understanding of ROC curve and AUC, and k- Nearest Neighbors Demonstration of predictive modeling using pharmaceutical and clinical examples such as predicting adverse drug reactions, disease risk prediction, binary therapeutic outcome modeling.

Tree-Based Models & Ensemble Learning Overview of the following ML models with emphasis on intuitive and conceptual understanding. Decision Trees (structure and splitting criteria), interpretation of decision paths and feature importance Random Forest, overview of ensemble concept Advantages and limitations of tree-based models Demonstration of these models for pharmaceutical and clinical applications such as ADR risk stratification, Patient classification, Predicting treatment outcomes etc.

Unsupervised Learning & Practical Case Studies Overview of the unsupervised learning with emphasis on intuitive and conceptual understanding of pattern recognition and clustering, and their applications in pharmaceutical sciences. Concept of unsupervised learning Clustering overview, K-means clustering, choosing number of clusters, Interpreting cluster outputs Mini-case study integrating regression, classification, or clustering on healthcare datasets to demonstrate Patient segmentation, and drug grouping based on properties.