Skip to main content
ISOP ACoP 2025 Annual Conference Main banner
ViewAttendees 3
Tutorial

Big Data/AI/Machine Learning

T1 - Strategies for Selecting Explanatory Variables in Exposure-Response Analysis Using Machine Learning

Wednesday, October 22, 2025
8:00 AM - 12:00 PM MDT
Location: Willow Lake 3 - 5

Disclosure(s):
Souvik Bhattacharya, PhD: No financial relationships to disclose
Hu Huang, PhD: No relevant disclosure to display
Jiawei Zhou, PhD: No financial relationships to disclose
Ahmed Elmokadem, PhD: No financial relationships to disclose
Lorenzo Contento, PhD: No financial relationships to disclose

Description of session (include background & scientific importance): With the advances in digital platforms, AI/ML is evolving as an important tool in drug development. Broadening the knowledge of clinical pharmacology by leveraging data and using modelling and statistical tools has proven to be effective in drug development by reducing both the cost and the time required to bring a drug to patients. The progress in the field of AI/ML has supported model-informed drug development (MIDD) by promoting synergistic techniques alongside the traditional approaches of exposure-response analyses. A widely used application of AI/ML, which proved to be advantageous in several aspects of drug development, is covariate selection. Effective covariate selection is essential for developing robust, interpretable, and generalizable models from the complex datasets commonly encountered in pharmacometrics.
This tutorial will connect general machine learning principles with their specific applications in pharmacometrics and modelling, with a focus on practical variable selection strategies.

Draft agenda of the tutorial is as follows:

1) Introduction to strategies for selecting explanatory variables in Exposure-Response Analysis Using Machine Learning
Souvik Bhattacharya
2) Use Machine Learning for Covariate Selection in Population Mode
Jiawei Zhou
3) Hierarchical Deep Compartment Modeling: Integrating Deep Learning, SHAP Analysis, and Bayesian Inference in Julia
Ahmed Elmokadem
4) Machine learning applied to Hepatitis B virtual patients suggests prognostic biomarker signatures for stratifying responders to standard-of-care therapies.
Rajat Desikan

Hands-on AI/ML application to select covariates organized by Souvik Bhattacharya and Hu Huang.

Learning Objectives:

  • Gain a solid foundation in ML principles relevant to pharmacometrics, including the challenges associated with high-dimensional and sparse data. Become familiar with a variety of feature selection techniques, along with an understanding their strengths and weaknesses. Recognize the importance of model interpretability and gain experience using techniques like SHAP values to understand feature contributions. Participants will develop a clear understanding of how machine learning and feature selection can enhance various pharmacometrics applications, such as PK/PD modeling, disease progression modeling, and exposure-response analysis. Engage in data preprocessing, feature selection, model building, and evaluation within a pharmacometrics context using R or Python.

Presentations:

  • 8:00 AM - 9:00 AM MDT
    Introduction to strategies for selecting explanatory variables in Exposure-Response Analysis Using Machine Learning

    Instructor: Souvik Bhattacharya, PhD (he/him/his) – Astellas Pharma

  • 9:00 AM - 9:30 AM MDT
    Use Machine Learning for Covariate Selection in Population Model

    Instructor: Jiawei Zhou, PhD (she/her/hers) – UNC-Chapel Hill

    Instructor: Rohit Rao, NA – Pfizer Inc.

  • 9:30 AM - 10:00 AM MDT
    Hierarchical Deep Compartment Modeling: Integrating Deep Learning, SHAP Analysis, and Bayesian Inference in Julia

    Instructor: Ahmed Elmokadem, PhD – Metrum Research Group

  • 10:15 AM - 10:45 AM MDT
    Machine learning applied to Hepatitis B virtual patients suggests prognostic biomarker signatures for stratifying responders to standard-of-care therapies.

    Instructor: Rajat Desikan, PhD (he/him/his) – GSK (UK)

  • 10:45 AM - 11:15 AM MDT
    Machine learning for exploratory data analysis and sub-population analysis in pharmacometrics using Julia

    Instructor: Lorenzo Contento, PhD (he/him/his) – Pumas-AI Inc.

  • 11:15 AM - 12:00 PM MDT
    Strategies for selecting explanatory variables

    Instructor: Souvik Bhattacharya, PhD (he/him/his) – Astellas Pharma