Human System Risk Management and Knowledge Graphs for Human Spaceflight

Human spaceflight is a challenging domain that requires a broad set of expertise to operate successfully and safely. Although we have over 60 years of spaceflight experience, the number of people who have flown in space is still small. In the early years of human spaceflight, the risk from engineering failures was the predominant form of risk encountered. As humans have stayed longer in the spaceflight environment, we have learned more about how human physiology responds to extended exposure to that environment. We are also continuing to learn how to best design and engineer the vehicles, habitats, and spacesuits that enable us to live and work in space. With the renewal of interest in lunar and Martian missions and the beginning of the Artemis program, the process of understanding how risk changes as spaceflight missions change is central to ensuring that human spaceflight can be done safely.

Graphs and risk management

Increasing system complexity is a central challenge to human spaceflight. Concepts that span from engineering to life sciences to medical and psychological and more must be woven together to enable and advance human spaceflight. Risk management approaches are needed that can address that complexity and translate scientific evidence to enable risk-informed decision making. The NASA Human System Risk Board (HSRB) (https://www.nasa.gov/hhp/hsrb)  has piloted projects using Directed Acyclic Graphs to map known and suspected causal connections from the hazards of spaceflight through key contributing factors, to undesired mission level outcomes that spaceflight providers seek to avoid. It has historically been difficult to communicate how certain cellular or physiologic changes in humans can flow through the crew-vehicle system to affect risk to the mission or the life of the astronauts. By using a graphical approach, experts in different fields can come to agreement on a shared conceptual model of risk that enables improved discussion regarding the relative value of contributing factors or countermeasures in risk reduction.

Evaluating evidence

Simply knowing that a physiologic process or countermeasure may influence system risk is not sufficient for applied problems. The evidence available to predict which problems will become clinically or operationally significant is limited in part due to the small number of humans that have flown in space as well as significant biologic variability among astronauts.  As evidence grows and knowledge graphs improve over time, the information they capture can help narrow uncertainty and enable improved prioritization of investments and resources in systems design. This is a critical need in spaceflight where the allocations for mass, volume, power consumption, and data storage and bandwidth are severely restricted compared to most terrestrial exploration. Significant effort can be spent on minor issues if there is no estimate of relative magnitudes of contribution to overall risk at the mission level.

Graphical approaches may offer a path for both quantitative and systematic evaluation of the magnitude of risk contribution for a specific risk and for the interconnected system of risk. To exploit this, it is necessary to understand the relationships between a contributing factor of interest and other factors in the complex system to the extent we are able. The DAGs in this collection start with consensus expert knowledge as a starting point – we draw the map of what we think we know. We are then able to leverage the flexibility of native graph databases to relate available evidence and either validate or falsify proposed connections between nodes. While NASA created the starting point for this process, it is expected that as we learn, these knowledge graphs will evolve with a growing evidence base. A collection such as this provides a forum for peer-reviewed submissions to build on and improve knowledge graphs that improve the art and science of human spaceflight.

Guidelines for DAG creation

NASA first published guidelines for graph creation and maintenance in 2022. These original documents are publicly available from the NASA Technical Reports Server. The technical report Directed Acyclic Graph Guidance Documentation (https://ntrs.nasa.gov/citations/20220006812) provides fundamental information about how these graphs were initially created and guidelines for standardizing graph structure and harmonizing terminology among graphs.  Questions about how and why we structured graphs in a certain way can be found here. A second technical publication, Directed Acyclic Graphs: A Tool for Understanding the NASA Human Spaceflight System Risks – Human System Risk Board (https://ntrs.nasa.gov/citations/20220015709) was published that same year with the HSRB approved DAGs for each of the official human system risks, code for their recreation, and common definitions for nodes used within the different DAGs. 

Applications for science and risk management

Graphical approaches to risk management provide a powerful new pathway for understanding and quantifying risk. The ability to build up knowledge graphs and eventually rapidly analyze a complex but understandable system is crucial to the spaceflight industry as they design and redesign vehicles, habitats, space suits and spaceflight systems that require effective human system integration. In the coming decades we will learn more about how the human body responds to the various spaceflight environments, how biologic variability affects systems risk posture, and how various countermeasures affect the resilience of the crew-vehicle systems in spaceflight. Capturing that information in knowledge graphs provides critical information and lessons learned that can help drive forward human system integration in spaceflight and systematically improve safety as we explore farther from Earth.