Learning-accelerated discovery of immune-tumour interactions

Area of Science:

• Computational biology
• Cancer research
• Immunotherapy

Background:

• Cancer immunotherapies offer promising treatment avenues but require complex design optimization.
• Simulating intricate tumor-immune dynamics is computationally intensive.

Purpose of the Study:

• To present an integrated framework for dynamic exploration of cancer immunotherapy design spaces.
• To leverage high-performance computing for detailed dynamical simulations.

Main Methods:

• Combined PhysiCell (agent-based modeling) and EMEWS (model exploration) platforms.
• Developed an agent-based model of immunosurveillance against heterogeneous tumors.
• Implemented active learning and genetic algorithms with high-performance computing workflows.

Main Results:

• Adaptive sampling of the model parameter space was achieved.
• Iterative discovery of optimal cancer regression regions within biological constraints was enabled.
• Detailed spatial dynamics of stochastic tumor-immune interactions were simulated.

Conclusions:

• The integrated framework facilitates efficient exploration of complex cancer immunotherapy design spaces.
• This approach aids in identifying optimal treatment strategies by simulating tumor-immune dynamics.
• The framework supports adaptive sampling and discovery of cancer regression regions under clinical constraints.