CauseMap

CauseMap infers causal relationships in complex nonlinear time series using Convergent Cross Mapping (CCM) to analyze dynamical systems for biomedical and personalized medicine applications.

Key Features:

• Model-free CCM: Uses Convergent Cross Mapping to infer causality without relying on predefined parametric models.
• Robustness and directionality: Determines the direction of causation in systems with feedback loops and nonlinear interactions.
• High-dimensional system reconstruction: Reconstructs high-dimensional system dynamics from single-variable time series based on Takens' Theorem.
• Personalized-medicine analysis: Extracts causal relationships from individual-level longitudinal data to characterize person-specific dynamics.
• High-performance implementation: Implemented in Julia for efficient numerical computation on large time series datasets.

Scientific Applications:

• Biomedical research: Identifying causal links among physiological, molecular, and clinical time series in biomedical studies.
• Personalized medicine: Characterizing individual-specific causal dynamics to inform personalized diagnostics and interventions.
• Wearable device time series analysis: Inferring health-related causal relationships from longitudinal data produced by wearable sensors.

Methodology:

Applies Convergent Cross Mapping (CCM) grounded in Takens' Theorem to reconstruct system dynamics from single-variable time series and infers causality by predicting points between opposing time series.