Predicting Spaceflight Operational Performance From Sensorimotor Assessment Tasks

INTRODUCTION
Exploration class missions will require a new level of crew autonomy to self-assess and treat their sensorimotor dysfunction around periods of gravitational transition, where sensorimotor disturbances are at their highest. A sensorimotor assessment task set was developed to assess performance capabilities prior to early extravehicular activities (EVAs) based on lessons learned from Apollo and through feedback from subject matter experts. The task set included the following requirements: 1) mimic body maneuvers that would be performed during an EVA such that crew can self-assess their potential ability to complete operational tasks; 2) provide opportunities to develop strategies to recover from off-nominal body positions; and 3) aid in progressive adaptation to the novel gravitational environment. The purpose of this study was to validate the task set using a sensorimotor spaceflight analog to induce varying levels of disruption including the Sensorimotor Disorientation Analog (SDA) [1].

METHODS
Forty-one healthy non-astronaut subjects (21 males, 20 females; Mean ± SD, Age: 34.0 ± 7.8 years) participated. Subjects attended two testing sessions: 1) familiarization to the tasks and baseline data collection, and 2) tasks completed under two levels of sensorimotor disorientation mimicking astronaut postflight performance at R+0 and R+1. Subjects performed a series of seven sensorimotor assessment tasks followed by two operationally relevant EVA analog tasks. The sensorimotor assessments tasks, ordered from least to most provocative of the sensorimotor system, included: postural stability, step test, prone & supine recovery from fall (RFF), four-square-step-test, obstacle negotiation with turns (OTT), kneel-and-reach. Operational analog tasks included a top-hatch capsule egress and a 5-minute EVA traversal. Disorientation was applied using galvanic vestibular stimulation and weighted trunk, wrists, and ankles.

RESULTS
Pearsons’ correlation r-values were calculated between 1) all assessment tasks and 2) each assessment and operational analog task, across individuals and SDA levels (n=120). Strong correlations were present between each assessment task (r ≥ 0.59). The two assessment tasks with the strongest correlations to the operational analog tasks were the RFF (r ≥ 0.45) and OTT (r ≥ 0.60). To assess the ability of the sensorimotor tasks to predict operational analog task performance and aid in task down-selection, leave-one-out cross-validation (LOOCV) was conducted with a linear model of assessment tasks selected using stepwise regression. The operational analog predictors were the time to complete the top-hatch capsule egress and the number of bag transfers complete during the 5-minute EVA traversal. For capsule egress, only the OTT was included in the model with a mean absolute error (MAE) of 12.8 sec between predicted and observed values (Q2 = 0.35). For the EVA traversal, the RFF, OTT, and FSST were included in the model with an MAE of 6.5 bags (Q2 = 0.53).

DISCUSSION
These results provide early evidence for the down-selection of assessment tasks that were best able to predict performance in operationally relevant tasks for use in upcoming Artemis missions.

ACKNOWLEDGEMENTS
This project is funded by the NASA Human Research Program.