An Integrated Simulation-HPC-Learning Approach to Create Cancer Patient Templates for Digital Twins
Event Type
Workshop
Applications
Computational Science
Education and Training and Outreach
HPC Community Collaboration
HPC Training and Education
Machine Learning and Artificial Intelligence
Performance
Workforce
W
TimeSunday, 14 November 202110:30am - 10:45am CST
Location223
DescriptionCancer patient digital twins (CPDTs) are personalized simulation models that can forecast individuals’ prognosis under a variety of treatment options. To successfully launch CPDTs, we must combine mechanistic modeling, artificial intelligence (AI), and high performance computing (HPC) into a platform that can seamlessly combine the patient’s data with accumulated knowledge, while continuously learning from successes and failures.
Building from recent COVID-19 modeling, we developed a multiscale agent-based model of melanoma micrometastases and immune response in lung tissue. The cellular components of the tumor microenvironment are formed by healthy epithelial cells from the lung, isolated melanoma cells, and immune cells: macrophages, dendritic cells, CD8+, and CD4+ T cells. The model also includes immune cell trafficking to and from the lymphatic system to drive an expanding immune response. Melanoma cells proliferate uncontrolled, causing mechanical stress in the region of the cell cluster. This mechanical factor leads to the death of healthy cells in the region, stimulating the activation of antigen-presenting cells (APCs). Apoptotic melanoma cell death can also activate APCs. Macrophages ingest dead cell debris and produce proinflammatory cytokines, recruiting more APCs to the region. Dendritic cells are activated when in contact with dying cells (lung and melanoma) and 'receive' their antigen signature. Activated dendritic cells migrate to the lymph node and recruit CD8+ and CD4+ T cells that can induce cancer cell death.
We analyzed the parameter space by using high-throughput model exploration on HPC to generate over 100k virtual patient trajectories, and demonstrated that the model can recapitulate a broad variety of virtual patient trajectories, including dormancy, uncontrolled growth, and partial and complete tumor response. Moreover, we used AI techniques to cluster the virtual trajectories into CPDT templates – the first step in fitting a personalized model to an individual patient. To fit a personalized model to an individual patient, we used several different bi-clustering techniques on CPDT patient templates and virtual trajectories. Our initial results suggest that some CPDT patient profiles may not be distinguishable in the initial stages, and that complete tumor elimination may represent a “lucky” stochastic event in the broader population whose cancer is otherwise partially controlled by the immune system, rather than a distinct subpopulation.
Building from recent COVID-19 modeling, we developed a multiscale agent-based model of melanoma micrometastases and immune response in lung tissue. The cellular components of the tumor microenvironment are formed by healthy epithelial cells from the lung, isolated melanoma cells, and immune cells: macrophages, dendritic cells, CD8+, and CD4+ T cells. The model also includes immune cell trafficking to and from the lymphatic system to drive an expanding immune response. Melanoma cells proliferate uncontrolled, causing mechanical stress in the region of the cell cluster. This mechanical factor leads to the death of healthy cells in the region, stimulating the activation of antigen-presenting cells (APCs). Apoptotic melanoma cell death can also activate APCs. Macrophages ingest dead cell debris and produce proinflammatory cytokines, recruiting more APCs to the region. Dendritic cells are activated when in contact with dying cells (lung and melanoma) and 'receive' their antigen signature. Activated dendritic cells migrate to the lymph node and recruit CD8+ and CD4+ T cells that can induce cancer cell death.
We analyzed the parameter space by using high-throughput model exploration on HPC to generate over 100k virtual patient trajectories, and demonstrated that the model can recapitulate a broad variety of virtual patient trajectories, including dormancy, uncontrolled growth, and partial and complete tumor response. Moreover, we used AI techniques to cluster the virtual trajectories into CPDT templates – the first step in fitting a personalized model to an individual patient. To fit a personalized model to an individual patient, we used several different bi-clustering techniques on CPDT patient templates and virtual trajectories. Our initial results suggest that some CPDT patient profiles may not be distinguishable in the initial stages, and that complete tumor elimination may represent a “lucky” stochastic event in the broader population whose cancer is otherwise partially controlled by the immune system, rather than a distinct subpopulation.
Author/Presenter
