gQSPTwin: A MATLAB-Based Application for Automation and Standardization of the QSP Digital Twin Generation Workflow

Objectives: Quantitative Systems Pharmacology (QSP) models are increasingly employed in clinical development to enhance the understanding of drug effects and patient variability 1 . A significant challenge when simulating clinical trials is accurately representing patient variability, especially for novel therapies or new patient populations where biological and pharmacological parameter distributions are not well understood. Digital twins offer a promising solution by creating model parameterizations that replicate the physiological and pathological characteristics of each individual patient, allowing for the evaluation of clinical outcomes with limited data2-4. This approach has been successfully applied in the Phase 1 development of T-cell engaging bispecific (TCB) antibodies to characterize clinical dose-response and propose predictive biomarkers2, as well as in TCR-engineered cell therapy to predict T-cell kinetics and explore the role of Tscm biology3. Methods: We have developed the MATLAB-based gQSPTwin toolbox to automate the digital twin generation workflow, streamlining the process for more efficient use of QSP models during clinical trials. It facilitates the creation of candidate digital twins for each patient, simulates and selects a diverse set of patient-specific digital twins, and provides diagnostic visualizations throughout the workflow. Results: The gQSPTwin workflow includes key steps such as preparing input files, generating candidate digital twins, visualizing the generation process, simulating digital twins, selecting top digital twins, and visualizing selections via a diagnostic dashboard, culminating in an output log file. We demonstrate the application of the gQSPTwin toolbox through two case studies: TCR-engineered cell therapy and TCB.