AI-based medical care for deep space exploration
AI-based medical care for deep space exploration
HARNESSING ARTIFICIAL INTELLIGENCE FOR MEDICAL DIAGNOSIS AND TREATMENT DURING SPACE EXPLORATION MISSIONS
TOPICAL:
Enabling a Precision Health System for
Deep Space Exploration
Space is considered to be the most inhospitable environment known to man. A lack of oxygen, microgravity, extremes
of temperature, ionising radiation and the inability to grow food being only a few of the challenges that space
exploration may pose to those brave enough to travel there.(1) Consequently, astronauts encountered numerous
health risks primarily due to the effects of microgravity and ionising radiation as well as the psychological impacts
of isolation and connement.(2,3) Because of this, it is imperative that the health and wellbeing of astronauts be
monitored closely to ensure their safety. This is currently achieved via telemedicine, in which medical personnel on
earth communicate with those in space. However, this is not without limitations such as the inability to physically
examine those in space and communication delays that may be encountered due to the vast transmission distances.
In the case that communications become compromised or infeasible, astronauts may be left vulnerable to a wide
range of potential health complications. Therefore, a redundancy solution to monitor astronaut health alongside
direct astronaut-to-earth communication is required. This could be achieved via the use of articial intelligence (AI).
AI allows for the health of astronauts to be automatically monitored and provides an effective solution to some of
the biological and psychological issues that astronauts encounter. This essay explores the role of AI in a selection of
health issues encountered by astronauts.
TELEMEDICINE IN SPACE
Almost every physiological system in the human
body can be affected by space travel, often resulting
in hypotension, tachycardia, uid redistribution, optic
disc oedema, dysrhythmias, cardiac remodelling,
kidney stones, muscle atrophy, bone resorption, carotid
intima thickening and long-term risks of cancer.(1,3,4) In
addition to this, astronauts also experience signicant
psychological stress due to isolation, connement and
a complex work environment.(2,5) Currently, astronauts
are monitored from Earth using telemedicine which
grants medical personnel the ability to monitor and
manage astronaut health from a distance. This
generally involves monitoring vital signs such as heart
rate, respiratory rate, blood pressure, temperature,
and ECG ndings alongside diagnostic imaging in the
form of ultrasound which are all transmitted to Earth
for interpretation. However, as previously mentioned,
telemedicine is limited by its inability to perform
physical examinations or intervention. A further issue
with telemedicine relates to the difculties encountered
with deep space communication. Communication
between astronauts and Earth generally requires
transmissions to be sent over very large distances.
While transmissions generally travel at the speed of
light (~300,000km/s), communication latency becomes
a signicant challenge as the distance between
transmitters and receivers increases. For example,
the average distance between Earth and the Moon is
approximately 384,000km which results in minimal
transmission latency of ~1.3 seconds. However, this
becomes far more signicant when considering that the
distance between Earth and Mars is anywhere between
55,000,000 - 378,000,000km resulting in a transmission
latency between 3 - 21 minutes.(6) Furthermore, there
is also risk of damage to transmission relay satellites
due to space debris or cosmic radiation. Such
communication challenges highlight the need for a
redundancy measure in place to monitor the health of
astronauts if direct communication became unavailable
or ineffective. This is particularly relevant in the current
day with the intention to send astronauts to Mars and
beyond.
THE ROLE OF ARTIFICIAL
INTELLIGENCE IN SPACE MEDICINE
Journal of the Australasian Society of Aerospace
Medicine (JASAM) No.13 (2024) 1-5
10.2478/asam-2024-0001
© 2024 Connor Greatbatch. This is an open access article licensed
under the Creative Commons Attribution-NonCommercial-NoDerivs
License (http://creativecommons.org/licenses/by-nc-nd/3.0/).
Connor Greatbatch
University of Tasmania
BMedSci (Hons)
MBBS (Year 5)
connorgbatch@gmail.com
JASAM VOLUME 13, 2024 | 2
ARTIFICIAL INTELLIGENCE
The utilisation of AI provides an effective and innovative
way to overcome the challenges associated with
telemedicine in space. This powerful tool could allow
for the effective monitoring of astronauts, in addition
to acting as a redundancy measure in the case that
communications become compromised. AI refers to
mathematical algorithms that mimic human intelligence
with capabilities such as learning, adaptation,
reasoning, and sensory interaction.(7) It functions by
analysing large datasets, extracting important features
and using those features to extrapolate new data and
formulate accurate predictions or classications.
The introduction of this technology has resulted in a
paradigm shift in the way we analyse complex data
and has proven to be a powerful problem-solving tool.
(8,9) Deep learning is one of the most rapidly growing
subelds of AI. This utilises algorithms called neural
networks that are designed to mimic the architecture
and function of the brain.(10) In the past, AI was limited
to working with simple, tabular data, however, with
the rise of deep learning, AI is now able to learn from
complex, natural data such as images and sequences.
In the eld of medicine, there are already numerous
clinically approved AI systems with successful
application in stroke diagnosis, diabetic retinopathy
screening, ECG analysis and MRI interpretation.(11)
Given the complexity of aerospace medicine and
the signicant challenges in providing healthcare to
astronauts, there is an opportunity for AI to play a
signicant role in delivering healthcare to astronauts as
they venture further and further from the earth.
MONITORING VITAL SIGNS
The effects of microgravity and space travel result in
physiological dysfunction across almost every body
system. Therefore, it is crucial to monitor these systems
to identify abnormalities, manage them accordingly
and prevent complications that could compromise
the mission or even be life-threatening. The most
fundamental and critical vital signs to monitor are
heart rate, blood pressure, respiratory rate, oxygen
saturation and temperature. These can be measured
via the use of blood pressure cuffs, pulse oximetry and
electrocardiograms (ECG) or with innovative wearable
technology that allows vital signs to be measured
via a wearable device or a piece of clothing with
integrated monitoring.(12,13) This continuous stream
of numerical data provides an excellent opportunity
for AI-integrated monitoring of astronaut health. It is
well documented that microgravity causes signicant
cardiovascular dysfunction resulting in hypotension,
tachycardia and risk of arrhythmias; all of which be
identied with the aforementioned monitoring.(14) A
simple computational program would be effective at
identifying when numerical values become abnormal,
however, a more complex system is required to
analyse ECG waveform morphology to identify cardiac
dysrhythmias. Microgravity infers an increased risk of
cardiac arrhythmias which has presented in previous
studies as premature ventricular contractions, atrial
brillation, non-sustained ventricular tachycardia and
QTc prolongation.(15–17) Continuous ECG monitoring
by an AI could be highly effective at identifying and
classifying cardiac arrhythmias. Numerous studies have
developed AI systems that can classify arrhythmias
with exceptional accuracy.(18–22) This illustrates the
potential effectiveness of implementing AI systems for
the continuous monitoring of astronaut vital signs and
ECG interpretation during space-travel.
ULTRASOUND DIAGNOSTICS
Currently, ultrasound (US) is the only feasible
modality of diagnostic imaging in space due to its
portable nature compared to computed tomography
(CT) or magnetic resonance imaging (MRI). US is
particularly useful in space due to issues related to
microgravity. Microgravity has a profound effect on
the cardiovascular system causing a decrease in
circulating blood volume leading to uid redistribution,
hypotension and subsequent cardiac remodelling.(23)
This remodelling can result in a signicant reduction
in physical capacity upon returning to partial gravity
(i.e. the Moon or Mars) or full gravity when returning
to Earth.(24) AI can allow for monitoring of such cardiac
dysfunction through the automated interpretation of
ultrasound echocardiography. Previous studies have
utilised AI models to correctly identify and interpret
cardiac dysfunction on echocardiography such as
abnormal left ventricular volumes, ejection fraction and
myocardial wall motion with high accuracy.(25–29) This
identication could allow for appropriate measures
to be taken to optimise the physical capabilities of
astronauts who experience this issue.
Another complication of microgravity is the
predisposition to kidney stones due to bone breakdown
and urine calcium supersaturation.(30) This can
escalate to a medical emergency if left untreated
due to signicant pain and infection risk with a very
challenging prospect of evacuation from space. Luckily,
many of these events occur shortly after returning to
Earth, however, in 1982 a Russian cosmonaut narrowly
missed being medically evacuated during a mission
due to kidney stones.(31,32) Modern AI systems have the
capability of kidney stone diagnosis from US images
as well as determining stone composition (i.e. calcium,
urate); however, management in space remains a
limitation with the optimal treatment being prevention.
(33–35)
The risk of trauma during space travel is another risk
that is practically impossible to completely remove
and may require urgent diagnostics to ascertain if
emergency management is required. US is highly
effective at identifying abdominal trauma with the FAST
(Focused Assessment with Sonography for Trauma)
scan which is utilised around hospitals internationally
to identify any bleeding or uid in the abdomen.
Further utilisation of AI may include the automated
interpretation of FAST scans in space. Previous clinical
trials have already illustrated that a deep learning
GREATBATCH | THE ROLE OF ARTIFICIAL INTELLIGENCE IN SPACE MEDICINE
JASAM VOLUME 13, 2024 | 3
algorithm can rapidly interpret FAST scans (in both
adult and paediatric populations) with sensitivity and
specicity between 90 - 100%.(36,37) There has been
a previous case report of ocular trauma on the ISS
requiring ocular ultrasound for diagnostic purposes.
(38) This is yet another potential role of AI as Chen
and colleagues developed a deep learning model to
automatically diagnose abnormalities on ophthalmic
ultrasound.(39) In future years, there will likely be many
imaging modalities available such as portable MRI
machines or entirely new imaging techniques that can
be applied in the domain of space travel and would
allow even further expansion of the capabilities of AI-
guided image diagnostics.
PSYCHOLOGICAL STRESS
Astronauts experience signicant psychological
stress due to prolonged isolation with a small group
of individuals, connement, and the risk of little to no
possibility of rescue in an emergency. This is further
complicated by circadian dysfunction and poor sleep
quality due to microgravity and 90-minute periods
between sunrise and sunset on the ISS.(1) This can
manifest as psychiatric disorders such as depression,
anxiety and adjustment disorders which can impair
an astronaut’s technical and cognitive performance.
There is strong evidence that AI can be utilised to both
identify and manage psychological disorders such as
depression and anxiety. For example, previous studies
have utilised neural networks to identify emotional
states and even diagnose depression on facial features
alone.(40,41) This approach also been adopted and
expanded to diagnose depression based on speech
waveforms alone.(42,43) Similar approaches have also
been adopted for the identication of other psychiatric
issues experienced by astronauts such as anxiety and
stress-related disorders.(44) Not only is AI applicable to
the diagnosis of psychiatric disease, but it can also play
a role in management. There have been several clinical
studies of AI chatbots created to deliver psychotherapy
(eg. Cognitive behavioural therapy) for the management
of depression and anxiety. For example, Fitzpatrick and
colleagues developed Woebot, a fully conversational
agent designed to deliver cognitive behavioural therapy
which has been shown to improve symptoms of
depression in a randomised controlled trial.(45) Similar
systems have also been trialled which have shown an
increase in positive mood and a reduction in anxiety
levels.(46,47) In fact, this concept has already been trialled
on the ISS with an AI system called CIMON (Crew
Interactive Mobile Companion) as a method of relieving
psychological stress experienced by astronauts.(48,49)
CONCLUSION
In summary, outer space is an incredibly hostile
environment that poses signicant health risks to
astronauts. These risks are currently managed via the
use of telemedicine; however, AI provides a potential
alternative and redundancy measure in cases where
standard telemedicine becomes unavailable or
compromised. AI-based systems have the capability
of monitoring astronaut vital signs, analysing ECG
waveforms, performing US diagnostics, and even
providing psychological support to those who need
it. Whilst we are still a long way from generalised AI
as seen in science-ction such as HAL 9000 from
2001: A Space Odyssey or TARS from Interstellar,
with technological advancements, astronauts may be
sharing space shuttles and exploring outer space with
AI to help reduce the psychological and psychological
burden of space travel.
REFERENCES:
1. Thirsk R, Kuipers A, Mukai C, Williams D. The space-
ight environment: the International Space Station
and beyond. CMAJ. 2009 Jun 9;180(12):1216–20.
2. Morphew E. Psychological and Human Factors in
Long Duration Spaceight [Internet]. Vol. 6, McGill
Journal of Medicine. 2001. Available from: http://
dx.doi.org/10.26443/mjm.v6i1.555
3. Hodkinson PD, Anderton RA, Posselt BN, Fong KJ.
An overview of space medicine. Br J Anaesth. 2017
Dec 1;119(suppl_1):i143–53.
4. Stepanek J, Blue RS, Parazynski S. Space Medicine
in the Era of Civilian Spaceight. N Engl J Med.
2019 Mar 14;380(11):1053–60.
5. Kanas N. Space Psychology and Psychiatry. Wertz
JR, editor. Microcosm Press, Springer; 2008.
6. Communication Delay [Internet]. Australian Space
Academy. [cited 2021 May 4]. Available from:
https://www.spaceacademy.net.au/spacelink/
commdly.htm
7. Amisha, Malik P, Pathania M, Rathaur VK. Overview
of articial intelligence in medicine. J Family Med
Prim Care. 2019 Jul;8(7):2328–31.
8. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, et al.
Articial intelligence in healthcare: past, present
and future. Stroke Vasc Neurol. 2017 Dec;2(4):230–
43.
9. Bohr A, Memarzadeh K. The rise of articial
intelligence in healthcare applications. In: Bohr A,
Memarzadeh K, editors. Articial Intelligence in
Healthcare. Academic Press; 2020. p. 25–60.
10. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature.
2015 May 28;521(7553):436–44.
11. Benjamens S, Dhunnoo P, Meskó B. The state of
articial intelligence-based FDA-approved medical
devices and algorithms: an online database. NPJ
Digit Med. 2020 Sep 11;3:118.
GREATBATCH | THE ROLE OF ARTIFICIAL INTELLIGENCE IN SPACE MEDICINE
JASAM VOLUME 13, 2024 | 4
12. Kumar A. Evaluation of the Accuracy of Astroskin
as a Behavioral Health Self-Monitoring System
for Spaceight [Internet]. Annual Summer STEM
Internship Symposium; 2015 Aug 22; Salinas,
CA. Available from: https://ntrs.nasa.gov/api/
citations/20150021842/downloads/20150021842.
pdf
13. Jonas Dino BD. LifeGuard: Wireless Physiological
Monitor [Internet]. NASA. 2008 [cited 2021 May
5]. Available from: https://www.nasa.gov/centers/
ames/research/technology-onepagers/life-guard.
html
14. Hughson RL, Helm A, Durante M. Heart in space:
effect of the extraterrestrial environment on the
cardiovascular system. Nat Rev Cardiol. 2018
Mar;15(3):167–80.
15. D’Aunno DS, Dougherty AH, DeBlock HF, Meck JV.
Effect of short- and long-duration spaceight on QTc
intervals in healthy astronauts. Am J Cardiol. 2003
Feb 15;91(4):494–7.
16. Caiani EG, Martin-Yebra A, Landreani F, Bolea J,
Laguna P, Vaïda P. Weightlessness and cardiac
rhythm disorders: Current knowledge from space
ight and bed-rest studies. Front Astron Space
Sci [Internet]. 2016 Aug 23;3. Available from:
http://journal.frontiersin.org/Article/10.3389/
fspas.2016.00027/abstract
17. Vernice NA, Meydan C, Afshinnekoo E, Mason
CE. Long-term spaceight and the cardiovascular
system. Precis Clin Med. 2020 Dec;3(4):284–91.
18. Smith SW, Rapin J, Li J, Fleureau Y, Fennell W,
Walsh BM, et al. A deep neural network for 12-lead
electrocardiogram interpretation outperforms a
conventional algorithm, and its physician overread,
in the diagnosis of atrial brillation. Int J Cardiol
Heart Vasc. 2019 Dec;25:100423.
19. Ribeiro AH, Ribeiro MH, Paixão GMM, Oliveira DM,
Gomes PR, Canazart JA, et al. Automatic diagnosis
of the 12-lead ECG using a deep neural network. Nat
Commun. 2020 Apr 9;11(1):1760.
20. Alfaras M, Soriano MC, Ortín S. A Fast Machine
Learning Model for ECG-Based Heartbeat
Classication and Arrhythmia Detection. Frontiers in
Physics. 2019;7:103.
21. Zhu H, Cheng C, Yin H, Li X, Zuo P, Ding J, et al.
Automatic multilabel electrocardiogram diagnosis of
heart rhythm or conduction abnormalities with deep
learning: a cohort study. Lancet Digit Health. 2020
Jul;2(7):e348–57.
22. Chen T-M, Huang C-H, Shih ESC, Hu Y-F, Hwang M-J.
Detection and Classication of Cardiac Arrhythmias
by a Challenge-Best Deep Learning Neural Network
Model. iScience. 2020 Mar 27;23(3):100886.
23. Summers RL, Martin DS, Meck JV, Coleman TG.
Mechanism of spaceight-induced changes in
left ventricular mass. Am J Cardiol. 2005 May
1;95(9):1128–30.
24. Gallo C, Ridol L, Scarsoglio S. Cardiovascular
deconditioning during long-term spaceight through
multiscale modeling. NPJ Microgravity. 2020 Oct
1;6:27.
25. Ghorbani A, Ouyang D, Abid A, He B, Chen JH,
Harrington RA, et al. Deep learning interpretation of
echocardiograms. NPJ Digit Med. 2020 Jan 24;3:10.
26. Omar HA, Domingos JS, Patra A, Upton R, Leeson P,
Noble JA. Quantication of cardiac bull’s-eye map
based on principal strain analysis for myocardial
wall motion assessment in stress echocardiography.
In: 2018 IEEE 15th International Symposium on
Biomedical Imaging (ISBI 2018). 2018. p. 1195–8.
27. Madani A, Arnaout R, Mofrad M, Arnaout R. Fast and
accurate view classication of echocardiograms
using deep learning. NPJ Digit Med [Internet]. 2018
Mar 21;1. Available from: http://dx.doi.org/10.1038/
s41746-017-0013-1
28. Sanchez-Martinez S, Duchateau N, Erdei T, Fraser AG,
Bijnens BH, Piella G. Characterization of myocardial
motion patterns by unsupervised multiple kernel
learning. Med Image Anal. 2017 Jan;35:70–82.
29. Narula S, Shameer K, Salem Omar AM, Dudley
JT, Sengupta PP. Machine-Learning Algorithms
to Automate Morphological and Functional
Assessments in 2D Echocardiography. J Am Coll
Cardiol. 2016 Nov 29;68(21):2287–95.
30. 30. Zerwekh JE. Nutrition and renal stone disease in
space. Nutrition. 2002 Oct;18(10):857–63.
31. Simon JC, Dunmire B, Bailey MR, Sorensen
MD. DEVELOPING COMPLETE ULTRASONIC
MANAGEMENT OF KIDNEY STONES FOR
SPACEFLIGHT. J Space Saf Eng. 2016 Sep;3(2):50–
7.
32. Buckey JC. Space Physiology. Oxford University
Press; 2006. 283 p.
33. Sudharson S, Kokil P. An ensemble of deep neural
networks for kidney ultrasound image classication.
Comput Methods Programs Biomed. 2020
Dec;197:105709.
34. Selvarani S, Rajendran P. Detection of Renal Calculi
in Ultrasound Image Using Meta-Heuristic Support
Vector Machine. J Med Syst. 2019 Jul 31;43(9):300.
35. Black KM, Law H, Aldoukhi A, Deng J, Ghani KR.
Deep learning computer vision algorithm for
detecting kidney stone composition. BJU Int. 2020
Jun;125(6):920–4.
GREATBATCH | THE ROLE OF ARTIFICIAL INTELLIGENCE IN SPACE MEDICINE
4
| 5
36. Sjogren AR, Leo MM, Feldman J, Gwin JT. Image
Segmentation and Machine Learning for Detection
of Abdominal Free Fluid in Focused A
ssessment
With Sonography for Trauma Examinations: A Pilot
Study. J Ultrasound Med. 2016 Nov;35(11):2501–9.
37. Kornblith AE, Addo N, Dong R, Rogers R,
Grupp-Phelan J, Butte A, et al. Development
and Validation of a Deep Learning Model for
Automated View Classication of Pediatric
Focused Assessment with Sonography for
Trauma (FAST). medRxiv [Internet]. 2020;
Available from: https://www.medrxiv.org/
content/10.1101/2020.10.14.20206607v1.abstract
38. Chiao L, Sharipov S, Sargsyan AE, Melton S,
Hamilton DR, McFarlin K, et al. Ocular examination
for trauma; clinical ultrasound aboard the
International Space Station. J Trauma. 2005
May;58(5):885–9.
39. Chen D, Yu Y, Zhou Y, Peng B, Wang Y, Hu S, et al.
A Deep Learning Model for Screening Multiple
Abnormal Findings in Ophthalmic Ultrasonography
(With Video). Transl Vis Sci Technol. 2021 Apr
1;10(4):22–22.
40. Zhu Y, Shang Y, Shao Z, Guo G. Automated
Depression Diagnosis Based on Deep Networks
to Encode Facial Appearance and Dynamics.
IEEE Transactions on Affective Computing. 2018
Oct;9(4):578–84.
41. Melinte DO, Vladareanu L. Facial Expressions
Recognition for Human-Robot Interaction Using
Deep Convolutional Neural Networks with Rectied
Adam Optimizer. Sensors [Internet]. 2020 Apr
23;20(8). Available from: http://dx.doi.org/10.3390/
s20082393
42. He L, Cao C. Automated depression analysis using
convolutional neural networks from speech. J
Biomed Inform. 2018 Jul;83:103–11.
43. Chlasta K, Wołk K, Krejtz I. Automated speech-
based screening of depression using deep
convolutional neural networks. Procedia Comput
Sci. 2019 Jan 1;164:618–28.
44. Yashaswini DK, Bhat SS, Sahana YS, ShamaAdiga
MS, Dhanya SG. Stress Detection using Deep
Learning and IoT. International Journal of Research
in Engineering, Science and Management [Internet].
2. Available from: https://www.ijresm.com/
Vol.2_2019/Vol2_Iss8_August19/IJRESM_V2_
I8_14.pdf
45. Fitzpatrick KK, Darcy A, Vierhile M. Delivering
Cognitive Behavior Therapy to Young Adults With
Symptoms of Depression and Anxiety Using a
Fully Automated Conversational Agent (Woebot):
A Randomized Controlled Trial. JMIR Ment Health.
- Get link
- X
- Other Apps
Comments
Post a Comment