AI-Driven System to Accelerate Immunotherapy Design Receives NSF Funding

The US National Science Foundation has awarded $581,503 to Natasa Miskov-Zivanov, assistant professor of electrical and computer engineering at the University of Pittsburgh, to develop an artificial intelligence-driven system for designing more effective immunotherapies. The project aims to overcome limitations in treating solid tumours with current chimeric antigen receptor (CAR) T cell therapies by automating the process of identifying promising lymphocyte designs. Miskov-Zivanov’s system will integrate data from scientific literature and experimental results using natural language processing, large language models, and graph neural networks to construct knowledge graphs and predict therapeutic efficacy, ultimately accelerating the design of both CAR T cells and tumour infiltrating lymphocytes (TILs). The research also includes the development of a new graduate-level course focused on knowledge graphs and their application.

Accelerating Immunotherapy Development with Computational Modeling and Workforce Training

Researchers develop a novel framework integrating knowledge graphs, graph neural networks, and natural language processing to revolutionise immunotherapy design and accelerate the discovery of effective cancer treatments. The system’s ability to process and interpret data from diverse sources, including scientific literature, facilitates a deeper understanding of complex biological interactions and accelerates the development of personalised immunotherapies. This innovative system addresses a critical bottleneck in immunotherapy development by efficiently translating complex biological data into testable hypotheses, reducing reliance on costly and time-intensive laboratory validation.

The automated framework systematically organises biological information, including genes, proteins, receptors, experimental conditions, and outcomes, into a structured network that facilitates comprehensive analysis and insight generation. Graph neural networks then process this organised data, identifying patterns and relationships that correlate with therapeutic efficacy, enabling researchers to predict the performance of novel cell designs in silico. By prioritising computational modelling, the team aims to significantly reduce the time and resources required to identify promising immunotherapy candidates, paving the way for faster clinical translation and improved patient outcomes.

The system effectively integrates both traditional natural language processing techniques and contemporary large language models, extracting nuanced information from the vast landscape of scientific literature. This combined approach fosters a deeper understanding of complex biological interactions, allowing researchers to pinpoint promising therapeutic candidates that might otherwise remain obscured. Researchers believe this comprehensive data integration, facilitated by the knowledge graph serving as a central repository, will unlock new avenues for immunotherapy development and personalised cancer treatment.

The computational approach enables a systematic exploration of the design space, identifying optimal cell designs that balance critical factors such as efficacy, safety, and manufacturability. This is particularly crucial in the realm of complex immunotherapies, where the interplay between various cell components and the tumour microenvironment presents significant challenges. By accurately predicting therapeutic outcomes, the system minimises the need for extensive trial-and-error experimentation, accelerating the development process and reducing overall costs.

The framework’s modular design ensures adaptability and continuous improvement, allowing for the seamless incorporation of new data and algorithms as the field of immunotherapy evolves. The framework’s adaptability and open architecture promote collaboration and ensure the system remains at the forefront of research. This forward-thinking approach guarantees the system remains at the forefront of research, addressing the rapidly changing landscape of cancer biology and immunotherapy.

Recognising the importance of a skilled workforce, the project incorporates a dedicated graduate-level course focused on knowledge graphs and their applications in immunotherapy. This educational initiative cultivates the expertise necessary to effectively utilise these advanced technologies, empowering the next generation of researchers to drive innovation in the field. By training future engineers in computational immunotherapy, the project ensures the long-term impact of its advancements.