Validating Causal Diagrams of Human Health Risks for Spaceflight: An Example Using Bone Data from Rodents

Area of Science:

• Spaceflight research
• Systems engineering
• Risk management

Background:

• NASA utilizes causal diagrams (directed, acyclic graphs or DAGs) to model human system risks associated with spaceflight.
• Current methods lack empirical validation for these complex causal models.
• This gap hinders robust risk assessment and mitigation strategies.

Purpose of the Study:

• To develop and demonstrate a procedure for empirically testing the veracity of NASA's human system risk DAGs.
• To validate a specific DAG related to bone fracture risk from spaceflight exposure.
• To assess the utility of existing rodent datasets for this validation process.

Main Methods:

• Utilized a DAG focused on spaceflight-induced bone fracture risk.
• Applied statistical tests for expected marginal correlations and conditional independencies derived from the DAG.
• Analyzed four rodent datasets previously collected for bone fracture research.

Main Results:

• The rodent data generally aligned with the proposed DAG structure, supporting its validity.
• One dataset showed incongruencies, likely due to insufficient representation of a critical variable.
• Demonstrated the feasibility of using empirical data to test DAGs.

Conclusions:

• The proposed procedures offer a method for empirically validating NASA's risk management DAGs.
• Rodent data can serve as a proxy for validating spaceflight risk models.
• Future work should integrate human and multiomics data for enhanced validation.