The value of a spaceflight clinical decision support system for earth-independent medical operations

As NASA prepares for crewed lunar missions over the next several years, plans are also underway to journey farther into deep space. Deep space exploration will require a paradigm shift in astronaut medical support toward progressively earth-independent medical operations (EIMO). The Exploration Medical Capability (ExMC) element of NASA’s Human Research Program (HRP) is investigating the feasibility and value of advanced capabilities to promote and enhance EIMO. Currently, astronauts rely on real-time communication with ground-based medical providers. However, as the distance from Earth increases, so do communication delays and disruptions. Moreover, resupply and evacuation will become increasingly complex, if not impossible, on deep space missions. In contrast to today’s missions in low earth orbit (LEO), where most medical expertise and decision-making are ground-based, an exploration crew will need to autonomously detect, diagnose, treat, and prevent medical events. Due to the sheer amount of pre-mission training required to execute a human spaceflight mission, there is often little time to devote exclusively to medical training. One potential solution is to augment the long duration exploration crew’s knowledge, skills, and abilities with a clinical decision support system (CDSS). An analysis of preliminary data indicates the potential benefits of a CDSS to mission outcomes when augmenting cognitive and procedural performance of an autonomous crew performing medical operations, and we provide an illustrative scenario of how such a CDSS might function.

Condition outcomes affect mission specific metrics including Task Time Lost (TTL) by the affected crew member and does not include the CMO's time, Loss of Crew Life (LOCL), and the probability that a patient will need to return to definitive care (RTDC) i.e., evacuation. Each condition was identified by literature review for both the treated and untreated states. Capabilities were adapted by subject matter experts (SMEs) from terrestrial practice standards and assigned a SoP value by consensus among at least 3 clinicians expert in the specific capability. The SoP was guided by established terrestrial training curricula outlined in Error! Reference source not found. in the main text . (e.g. a Primary Assessment is SoP 1 while management decisions for refractory Sepsis are SoP 5). SoP classification values outlined in Error! Reference source not found. of the main text are based on well established, curricular requirements for terrestrial providers including national registry emergency medical technician, national registry paramedic, certified emergency registered nurse, first year medical resident, and attending physician. Capabilities were also categorized as procedure execution skills (e.g., performing an ultrasound or inserting a chest tube) or cognitive skills (e.g. interpreting ultrasound images or performing a differential diagnosis). The method made a conservative simplification and assumes a condition can only be treated if all its associated capabilities can be performed by a provider of equal or greater SoP level.
Each outcome metric (e.g. LOCL) was summed across all conditions based upon their treated or untreated values to estimate the outcome's total mission risk. Both the on-board Cognitive SoP (C-SoP) and Procedural SoP (P-SoP) and each condition requires multiple tasks of varying SoP level. Both C-SoP and P-SoP had to equal or exceed the maximum value required by the condition to be considered as 'treated'. If one or both of the on-board SoP scores were less than the conditions' maximum value required SoPs, the untreated values were used. A table was generated for each C-SoP and P-SoP combination for each outcome, and relative risk reduction was calculated using the following equation. Supplementary Equation 1 applies to TTL, RTDC and RTDC. The following assumptions and limitations were used to simplify this analysis: 1. The IMPACT-MD was assumed to represent a Mars LDEM even though it was designed for a Lunar LDEM 2. The mission lasts 26 months, with 7 month transits and 12 months on the Martian surface 3. The mission has 3 male and 3 female crew members 4. All conditions can occur to any crew member at any point during the mission except: a. Extravehicular Activity (EVA): Each crew member performs 1 EVA per week on the destination surface for a total of 624 EVAs. EVA conditions can only occur while on EVA. b. Space Adaptation Syndrome conditions can only occur during the first 5 days after launch from either Earth or the destination. c. Gravity Well Adaptation Syndrome conditions only occur within 5 days of landing at the destination. d. Surface Operation Conditions only occur during the 12 month stay on the destination surface e. Male/Female specific conditions can only occur to 3 of the 6 crew. 5. Mean probability of a condition occurrence is constant for the duration of mission. 6. Capabilities required to treat a condition can only be performed by a provider with equal to or greater SoP level than the maximum SoP capability required. CDSS assistance can be included to increase SoP level. 7. All capabilities must be performed to treat a condition. 8. Treatment resources are unlimited. 9. Task impairments sum linearly and are evenly distributed across the crew. 10. Task impairment only applies to the affected crewmember. It does not include the time lost by the treating CMO.
Supplementary Supplementary Table 1 and 2 show the results of a computational simulation using preliminary data from NASA's IMPACT project. IMPACT calculates 3 types of outcomes; Task Time Lost (TTL), Return to Definitive Care (RTDC) i.e., evacuation, and Loss of Crew Life (LOCL). The tables show how these mission outcomes vary by on board cognitive and procedure execution influenced by SoP levels. Supplementary Table 1 shows the absolute risk from the simulation reported in hours of time lost from illness for TTL and the number of occurrences for RTDC and LOCL while Supplementary Table 2 displays the relative risk reduction for these same metrics compared to C-SoP and P-SoP equal to zero. Results are averaged across 200 simulation runs.
There are 2 numbers reported for each scenario which represent a low end and high end estimation for mission risk. The high end risk was calculated using an absolute method. This method assumed that if the on board SoP was lower than the highest SoP required for a given condition no diagnosis or treatment could be given and hence no treatment benefit would occur. The low end method assumed that for a given condition all diagnostic and treatments could be given up to the maximum on board SoP. This means that if the on board SoP is 4 and a condition requires 10 capabilities to diagnose and treat with 9 being SoP 3 and 1 being SoP 5, only the SoP 5 capability could not be performed. In other words 9 out of 10 capabilities could be performed and the condition would receive a partial treatment benefit from them. While this is difficult to interpret clinically, it provides a reasonable best case treatment scenario on which to base a low end risk estimate.
For example, for the LOCL section of Supplementary Table 1, the C-SoP 0/P-SoP 0 scenario represents a hypothetical scenario in which no treatment can be rendered for any of the 120 conditions included in the simulated mission. In this case both the absolute treatment method and the partial treatment method demonstrate that 2.41 deaths occurred in the simulation. When the on board SoP is increased to a C-SoP of 1 and P-SoP of 1 representing the terrestrial EMT-B training the low end risk drops to 0.78 deaths while the high end remains at 2.41. SoP-1 is roughly equivalent to the ISS CMO training today. For the optimal situation in which on board C-SoP/P-SoP both equal 5 all conditions are fully treated and both absolute and partial treatment estimations drop to 0.04 deaths. This represents a relative risk reduction for the C-SoP/P-SoP 1 scenario between 0 and 68% and for the C-SoP/P-SoP 5 scenario of more than 98% over the fully untreated scenario. This means that if CDSS can increase an SoP 1 provider to SoP 5 the relative mission risk reduction likely lies between 34% and 98%.
This study does not specify the increase in SoP when a CDSS is used. However it can be seen that the results illustrate increasing SoP reduces the risk of LOCL, RTDC and TTL, i.e., all mission criteria improve in a beneficial direction. The results show SoP levels for cognitive skills have a greater effect on outcomes than procedural skills. The results of this analysis is for illustrative purposes showing the maximum benefit relative to a crew member with no medical training; however, note that it is likely that SoP will be greater than this in real missions. Results show that TTL can be improved by 97% if a CDSS improves both cognitive and procedural SoP to a level of 5. An interesting result from this analysis shows that Procedural SoP had little effect on outcome and Cognitive SoP needs to be greater than 3 to improve mission parameters. Cognitive SoP = 5 will most likely require a CDSS given restrictions on training and the wide medical coverage of likely conditions. Although this method cannot predict actual mission performance, it can provide an estimation of risk and the differences between SoP levels suggests that a high SoP