Frontiers in Medicine 2024 | AI in Oncology: Predicting Cancer’s Course

Summary

In this talk, Sylvia Plevritis explores the transformative role of AI in cancer research, particularly in understanding metastasis. She discusses how AI-enabled data lakes integrate clinical and molecular data, allowing for better predictions, treatment strategies, and clinical trial matches. Plevritis highlights the importance of lymph nodes in metastasis, previously considered passive, and how AI is helping to rethink their role. She also explains the innovative work being done to analyze tumor microenvironments using AI, advancing both clinical decision-making and our understanding of cancer biology.

Takeaways

• 😀 AI is transforming cancer research by improving our understanding of metastasis, which is one of the most life-threatening aspects of cancer.
• 😀 Lymph nodes, once thought to be passive bystanders in metastasis, are now being recognized as active players in the metastatic process and potential targets for reprogramming.
• 😀 AI-driven data lakes are integrating vast amounts of clinical and molecular data from electronic health records, lab tests, imaging, and patient notes to improve cancer care.
• 😀 The power of AI is allowing researchers to look at all aspects of cancer care, from prevention to treatment and advanced disease management, in a more holistic and interconnected way.
• 😀 AI is revolutionizing tumor boards by automating the integration of multimodal data, allowing specialists to make more informed decisions about cancer treatment.
• 😀 AI models are being used to predict clinical outcomes, find similar patients, and identify optimal treatment strategies, including matching patients to clinical trials.
• 😀 Multimodal foundation models, which integrate text, images, and clinical data, are being trained to better understand the complexities of cancer and its treatment.
• 😀 Cutting-edge technology is allowing for high-resolution tissue analysis, enabling the identification of 60 or even 5,000 different cellular markers, drastically expanding the amount of data available for research.
• 😀 AI is being used to analyze tumor ecosystems at a granular level, revealing patterns of cells and how they interact in ways that were previously undetectable.
• 😀 Data analysis has shown that certain cellular ecosystems around tumors, particularly those enriched with stromal cells, can create a protective barrier that hinders immune cells, influencing patient outcomes.
• 😀 The ultimate goal is to use AI to advance both clinical decision support and our understanding of tumor biology, enabling precision health and the potential to win the war on cancer.

Q & A

What is the primary focus of Sylvia Plevritis's research?

Sylvia Plevritis's research primarily focuses on the metastatic process in cancer, specifically how cancer cells spread from the primary tumor to lymph nodes and distant organs like the lung and liver.

What new perspective on lymph nodes has been uncovered through AI research?

AI research has revealed that lymph nodes are not just passive bystanders in metastasis but may actually play an active role as 'master orchestrators' in the metastatic process, opening up possibilities for reprogramming them to shut down metastasis.

How has AI improved the analysis of cancer research data?

AI has enhanced the ability to analyze large datasets by integrating diverse sources of data, such as clinical records, molecular data, and imaging, allowing researchers to discover associations that were previously unimaginable and make more informed clinical decisions.

What is a 'data lake,' and how does it help in cancer research?

A data lake is an AI-enabled platform that stores vast amounts of diverse cancer-related data—such as electronic health records, lab tests, imaging, and molecular data—from thousands of patients. It enables comprehensive analysis across different points in the cancer care continuum, from prevention to treatment and surveillance.

What role does AI play in the Tumor Board process?

AI assists the Tumor Board by automating the integration of multimodal data (clinical, imaging, genetic, etc.), allowing specialists to ask predictive questions, identify similar cases, and recommend optimal treatment strategies or clinical trials for patients with complex metastatic cancer.

What is a multimodal foundation model, and how is it used in cancer care?

A multimodal foundation model is a type of AI that can process different types of data (e.g., images, text, clinical data) and make predictions or recommendations. In cancer care, it is used to answer questions like predicting clinical events, finding similar patient cases, and helping select treatment plans.

How does AI assist in analyzing tumor images?

AI is used to analyze tumor images at a cellular level, identifying patterns and structures that might not be visible to the human eye. For example, AI can highlight cancer cells and understand how they interact with other cells in the tumor microenvironment, which is essential for developing targeted therapies.

What are tumor ecosystems, and how are they related to treatment outcomes?

Tumor ecosystems refer to the distinct patterns of cell types within a tumor, such as cancer, immune, and stromal cells. AI has identified that certain ecosystems, particularly those with a high concentration of stromal cells, are associated with worse treatment outcomes. Understanding these ecosystems can help in developing therapeutic strategies to improve patient prognosis.

How does the AI analysis of tumor microenvironments help in cancer treatment?

AI helps identify how various cell types in the tumor microenvironment, such as immune and stromal cells, interact with cancer cells. This understanding can lead to strategies to disrupt these interactions, potentially improving the effectiveness of treatments and breaking down barriers to immune cell infiltration.

Why is AI considered crucial in advancing our understanding of metastasis?

AI is crucial in advancing metastasis research because it allows for the integration and analysis of large, complex datasets, such as molecular, clinical, and imaging data. This enables researchers to uncover insights into how metastasis occurs and how different factors contribute to its progression, ultimately leading to more precise and effective treatments.