Koren Iton, A.; Iton, E.; Michaelson, D. M.; Blinder, P.

Recent advances in neuroscience have enabled simultaneous collection of multimodal datasets, including behavioral tracking, physiological signals, and neuronal recordings. Yet most analysis tools process these streams independently, often requiring manual alignment and custom code. This fragmentation limits reproducibility and interpretability at the experiment level, particularly in models where cognitive impairments are subtle, and reflects a broader tendency to analyze behavior as isolated events rather than as a structured process. Topological methods, though successful in other domains, remain underused in behavioral neuroscience, leaving the architecture of decision-making largely unexplored. To address these, we developed NaviGraph (Navigation on the Graph), a flexible, open-source pipeline designed to integrate diverse data types into a unified graph-based framework suited for spatial decision-making studies. By modeling decision points as graph nodes and populating them with behavioral, neuronal, and physiological parameters, NaviGraph offers a holistic, multi-layered perspective on cognition, enabling computation of spatial and topological metrics. This approach shifts the focus from isolated modalities to comprehensive interpretation, making behavioral patterns more accessible and easier to validate across datastreams. We applied NaviGraph to a trial-based spatial memory task in a complex maze and uncovered nuanced sex- and genotype-specific differences. In a knock-in model of Apolipoprotein {epsilon}E (apoE4), the most prevalent genetic risk factor for Alzheimer\'s disease, females exhibited deficits detectable only through topological metrics, including inefficient navigation and increased visits to decision points, aligning with the heightened cognitive vulnerability observed in female apoE4 human carriers. Wildtype females, in contrast, displayed more direct recall navigation compared to males. To demonstrate NaviGraph\'s multimodal capabilities, we mapped neuronal activity from head-mounted miniaturized microscope calcium imaging in the retrosplenial cortex, an area extensively involved in path integration, alongside head orientation dynamics and behavioral trajectories onto the graph structure. This unified framework enabled visualization of decision-point-specific neuronal activity patterns, subpopulation dynamics linked to path familiarity, and physiological- behavioral alignment. With its modular architecture, NaviGraph supports diverse maze configurations and data types, providing a holistic, interpretable, and high-resolution tool for spatial navigation and cognitive research.