Hi,
Today we’re welcoming @robert.j.reynolds-2 a pioneer in DAGs (Directed Acyclic Graphs) for space health.
Here’s a summary from the note-taker. Thanks, @jgong for sharing.
3 Likes
Thanks @alireza-h, very useful starting point. I’m a new AI/ML AWG member 
, my background is medicine and AI, bioinformatics. Could you please tell me which areas of the project would be most useful to contribute to right now?
4 Likes
Hi @dnkmd , happy to see you in AI/ML 
To reach a workable CDSS, we are currently trying to solve these problems, and I think your skills could really help if it is of interest to you:
-
A user-centered (crew member) and generalizable clinical decision support system framework
-
A real dataset to validate the framework, with the Inspiration4 dataset available at OSDR (data exploration , cleaning , restructuring )
-
Searching for and developing possible solutions to integrate with CDSS to heighten reliability and performance, including Directed Acyclic Graphs, knowledge graphs, and LLMs such as Gemini and Grok
-
Articulating our solutions and targeting journals that may have the same interests and needs
Best wishes,
Alireza
2 Likes
Also please check the current repo and prototypes
1 Like
Hi everyone, I am also interested in contributing to the group work project, please let me know what type of work is still available to do.
2 Likes
Hi, join the meeting Friday, we’ll talk about it. See you
1 Like
CMO-DA by NASA & Google :
advanced Natural Language Processing (NLP) for input (voice , text) ,
trained on Space medicine literature ,
validated using the Objective Structured Clinical Examination ,
Tested using space concepts of operation , emergency scenarios and got 80% accuracy
Link : Google AI becomes NASA’s first doctor for deep space missions
2 Likes
Ejenta:
In collaboration with Google and NASA is going to connect Crew Medical Officer Digital Assistance (CMO-DA),
How?
they solve an important need and point of care : integrate inputs from Ultrasonography Probe into the AI assistance using multi-agent technology
Link: Intelligent Systems to Support Crew Health and Performance During Earth-Independent Medical Operations — Ejenta
2 Likes
Hey, I don’t know if I have the sufficient expertise to do this, since I have only been working on basic data analytics and machine learning stuff, but this seems really interesting, and I would love to learn and contribute in any shape or form! Please contact me at p.prathmeshsoni5330@gmail.com, Thanks!
3 Likes
1 Like
2 Likes
Hi , sharing today’s meeting summary :
( Thanks @jgong for providing this )
22 May 2026
## Summary
### Spaceflight CDSS Framework Development
The subgroup meeting focused on developing a clinical decision support system (CDSS) for long-term spaceflight missions. The co-leads, Ritika and Alireza, outlined the goal to create a novel framework that integrates advanced AI with existing CDSS concepts, addressing challenges like data privacy, security, and implementation in space environments. Participants discussed key components such as data collection methods, individualized patient databases, federated learning, and the use of retrieval augmented generation (RAG) to handle data security. The group agreed to first develop a comprehensive system design architecture that incorporates these ideas before seeking real data from NASA and affiliated universities for validation and prototype development.
### CDSS Development Strategy Discussion
Rob advised focusing on a small, specific medical condition for the CDSS development rather than trying to address all medical issues at once. He shared a document with his thoughts on how the CDSS should be structured, which Alireza can share with others. Jian explained that while CDSS focuses on practical patient-facing applications, digital twin work is more scientifically driven and designed for knowledge discovery and hypothesis testing, though both initiatives share similar goals. Jian emphasized that the meetings are intended to foster open community discussions where diverse perspectives can lead to better ideas and outcomes.
### Collaboration
* All subgroup members: Consider and draft abstracts (individually or in teams) for submission to the ASGSR conference, focusing on CDSS, digital twin, federated learning, or related novel frameworks, by the end of the month. (Link : https://asgsr.org/ )
* Ritika & Jian: Plan and lead a future session focused on tools and processes for UI/UX design, including implementation details, for the subgroup.
* All subgroup members: Review Rob Reynolds’ shared CDSS framework document and consider how to integrate its recommendations into the subgroup’s framework development.
(Rob’s Docs link : DM me or @Ritika_Saha if you don’t have it )
* Ritika & Alireza: Lead the development of a novel CDSS framework incorporating discussed elements (DAGs, RAG, federated learning, privacy, etc.), and prepare for data requests from NASA/universities after framework completion.
2 Likes
Thanks @AliReza-H @Ritika_Saha for the summary and the leadership of the group. The summary is great , then I don’t need to worry about the 6 spam emails with links to otter, Ai, etc saying read the minutes here 
. For the short period of time the group has been going it has made some major forward steps. I agree that Robs idea of focussing on one area especially to help with development will probably speed up the process further. It’s such a big undertaking.
1 Like
I look forward to attending our next meeting. My background is in AI governance, technical documentation, and human-centered AI systems, with experience across some Big Tech projects and IEEE SA, and AI policy/research initiatives. Interested in contributing to CDSS architecture, RAG/agentic AI approaches, human factors, and secure knowledge integration for long-duration missions. Excited to collaborate and learn from the group. I also saw the amazing paper collection so far — especially around autonomous medical operations, SANS, cognition monitoring, and multimodal physiology. I’ve been thinking about complementary areas around trustworthy agentic AI systems for CDSS environments: evaluation frameworks, uncertainty handling, RAG reliability, long-context memory architectures, and human-AI teaming in high-latency/high-risk settings. Happy to share a few papers/resources around AI governance, multi-agent orchestration, and operational safety frameworks that may help as the system architecture evolves.
3 Likes
Not certain if the is on the readings NASA HRP AI
https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://ntrs.nasa.gov/api/citations/20250006312/downloads/ArtemisBoards_27Jun2025.pdf&ved=2ahUKEwi_7PmR1NWUAxXuV2wGHRQLMt4QFnoECE0QAQ&usg=AOvVaw1tr_mzjQIdA7PyRHn8pwS6
Hi @Ayse
I think you’ve got some awesome skills and experiences that could really make a difference on this project.
I’d love it if you shared your thoughts in this forum so everyone can contribute.
Also, Robert Reynolds wrote this super helpful piece that lays out our goals and needs, so you can figure out the best way to jump in.
I love this presentation,
For me, the best idea I got from this is what’s written at the end.
I see this as an exploration, a curiosity-driven adventure to find use cases for classical thought on space travel in an era of powerful AI technology.
*we should build many Git repos 
“”"
AI sprints are designed to show potentials and understand what might be useful
• This is not an effort [or the mandate] to build flight CDS
• Learning as we go, and ‘de-mystifying’ AI and LLMs. Working closely with flight surgeon
“”"
1 Like
Thank you, always happy to help, just asked @robert.j.reynolds-2 to grant edit/access rights…
2 Likes
Hi all !
Here is a great , relevant event by Organization for space medicine , engineering and design (OSMED) which i really love their efforts !
Topic : Special Lecture: Autonomous Medical Officer Support Software Technology & the ISS
I suggest everyone in this subgroup register for this free , online event
6 Likes
Hi @AliReza-H . AliReza Tavakkoli at UNR has developed an AI virtual process for testing for SANS . Tested on ISS April 2026. Redirect Notice
Could be a current source of SANS data
1 Like