Advancements in cell and gene therapies have prompted a new era of personalized medicine, offering potential therapies for several diseases. However, challenges such as limited efficacy, off-target effects, and lack of long-term resilience within the host organism hinder their widespread adoption. We believe that composite AI and biocomputing will improve our understanding and engineering of resilient phenotypes, through the integration of diverse and contextual biological data, followed by the identification of key operational principles for advanced modeling of adaptive behavior and enhanced engineering of modulators of adaptation.

The intersection of AI/ML and cellular engineering presents an extraordinary opportunity. As we unlock the complexities of cellular behavior, composite AI emerges as a powerful approach to decode, predict, and enhance cellular resilience. The integration of biological principles with advanced AI architectures opens new possibilities for understanding phenotype resilience and for computing those phenotypes in cell models for therapeutic development.

Living cells are remarkable information processing units, constantly integrating multiple streams of data to maintain function and adapt to challenges. This inherent sophistication demands new approaches to understanding and engineering cellular behavior—dynamic approaches that can match the intricacy of natural systems while providing actionable insights for therapeutic development.

Neural-symbolic AI merges neural networks' learning with symbolic reasoning. We believe that next-gen AI systems are bio-symbolic AI, advancing this further by implementing biological computation principles—moving from brain-inspired models to incorporate cellular processing, evolutionary adaptations, and ecosystems interactions—creating systems that mirror nature's adaptive architectures.

Bio-symbolic composite AI emerges as a powerful tool for decoding cellular intelligence by contextualizing diverse biological data. From the genetic code to metabolic states, cells speak in multiple molecular languages simultaneously. Our system processes this symphony of information—genomics, proteomics, metabolomics, and more—creating a dynamic picture of cellular behavior that reveals patterns of adaptation emerging across different scales (Figure 1).

The power lies not just in collecting this data, but in understanding how these different layers interact to create resilient behaviors. When cells encounter stress, they orchestrate complex responses involving transcriptional changes, protein interactions, and metabolic shifts. By modeling these interactions through composite AI architectures, we can begin to understand how cellular resilience emerges from these coordinated responses.

Contextual Data Integration

Biosymbolic composite AI combines bio-inspired computing, symbolic reasoning, and composite AI architectures to model and simulate complex biological systems. By integrating data-driven AI with symbolic and bio-inspired components, this approach can capture the multifaceted nature of biological processes, providing deeper insights and more robust solutions.

One of the foremost challenges in bioengineering is integrating vast and heterogeneous datasets, including genomics, proteomics, metabolomics, and context-specific data.

• Multi-Omics Integration: Composite AI can handle multi-omics data, identifying patterns and correlations that single-layered AI might miss.
• Context-Specific Modeling: Incorporating specific data allows for personalized solutions.

Simulation of Biological Principles

Natural cellular adaptation follows sophisticated operational principles that we can encode into computational frameworks. Cells employ intricate networks to process environmental signals and mount appropriate responses. Our composite AI architectures model these networks, simulating how signals propagate through cellular systems and predict emergent behaviors that arise from these interactions.

Biosymbolic composite AI enables the simulation of complex biological principles by combining data-driven models with symbolic reasoning and bio-inspired algorithms.

• Modeling Cellular Processes: Simulating gene regulation, metabolic pathways, and cell signaling networks.
• Dynamic System Representation: Capturing the evolution of cellular states under various conditions.

Perhaps most intriguingly, cells can "remember" past experiences and adjust their future responses accordingly. This cellular memory, essential for sustained therapeutic effectiveness, emerges from complex molecular interactions. Our frameworks model how these memory states establish and maintain themselves, providing insights for engineering more persistent therapeutic cells.

Understanding cellular behavior leads naturally to engineering enhanced resilience. Our system helps design therapeutic cells that not only survive but thrive in challenging environments. By identifying key modulators of cellular resilience, we can engineer cells with improved adaptive capabilities—cells that respond intelligently to local conditions and maintain their therapeutic function over time.

Concept of Cellular Memory

Cellular memory refers to a cell's ability to retain information about past states or stimuli, influencing future responses. In the context of therapies:

• Epigenetic Modifications: Stable changes in gene expression without altering DNA sequences.
• Synthetic Circuits: Engineered genetic constructs that enable cells to "remember" past signals.

Enhancing Therapy Resilience Through Cellular Memory

By engineering cellular memory, therapies can become more robust and adaptable within the host environment.

• Adaptive Responses: Cells can adjust their behavior in response to new stimuli based on stored information.
• Sustained Functionality: Maintaining therapeutic effects over extended periods.

Example: Designing CAR-T cells with synthetic circuits that enhance survival and function upon encountering immunosuppressive signals in the tumor microenvironment.

Implementing Cellular Memory in Therapies

• Memory Circuit Design: Using composites to model and optimize genetic circuits that implement memory functions.
• Identify mechanisms enhancing epigenetic memory in key transcriptional hubs.
• Safety Mechanisms: Incorporating kill switches or controllable elements to mitigate potential adverse effects.

As we look forward, the potential of this approach extends beyond current capabilities. We're moving toward precisely engineered cellular memory systems and adaptive therapeutic cells that evolve in response to disease conditions. The future promises self-optimizing cellular networks and therapeutic cells that communicate and coordinate their responses. These approaches can also be enhanced by the implementation of foundational cellular models and multiscale integration of contextual data.

The impact could revolutionize multiple therapeutic areas. In cancer immunotherapy, we envision T-cells with enhanced persistence and memory-enabled immune responses. Regenerative medicine could benefit from smart stem cells that respond appropriately to local cues. For metabolic disorders, engineered cells could provide robust, long-term therapeutic solutions.

Biosymbolic composite AI holds immense promise in revolutionizing cell and gene therapies by providing sophisticated tools for data integration, simulation, and optimization. By engineering cellular memory, we can enhance the resilience and adaptability of therapies within the host, leading to more effective and long-lasting treatments. As we continue to develop and refine these technologies, collaboration across disciplines will be essential to unlock their full potential.