CDSS subgroup project

AWG Projects AIML AWG Open Projects
41

Thank you for letting us know, and thanks as well for adding your ideas to the page—we really appreciate it. Please take care, and we look forward to catching up with you soon.

Best wishes.

42

Hello everyone, thank you so much!

This is a quick reminder that our meeting will begin in nearly 30 minutes.

Today, we’ll be discussing the goals we’ve developed so far for the subgroup. We will also begin dividing the work among interested members.

Looking forward to meeting everyone in the session!

Meeting Link: https://meet.google.com/sxb-qkcp-wmv

1 Like
43

hello, i wont be able to attend due to the fact that i am in school right now! please let me know what work is available. i will definitely get the work done!

2 Likes
44

Hi CDSS Crew! :rocket:

I wanted to highlight some NASA documents that are key for building our space-based CDSS prototype. These are critical for understanding the challenges and helping us fill in gaps in the current research.

1. CDSS Framework & Concepts

  • EXMC (Exploration Medical Capability) CDSS Concept of Operations

    • The EXMC ConOps lays out the core framework for a space-based CDSS, focusing on how to handle communication delays and health monitoring in deep space.

    • It’s super valuable for understanding the operational goals and the system architecture of a space CDSS.

    • You can check it out here: EXMC Concept of Operations.

Why it’s critical:

  • This document gives us a solid foundation for our space healthcare system and helps guide the design of our system. It also shows how to deal with space-specific challenges like limited communication and the need for autonomous decision-making.

Examples of Key Scenarios:

  • Kidney Stone Prediction

  • Bone Health Monitoring

  • Emergent Scenarios

3 Likes
45

2. Technical Reports on Medical Monitoring

  • NASA Health Monitoring Systems

  • Telemedicine in Space

These documents are critical for our agentic model development and architecture. To build the architecture, we need to understand these reports carefully so that we can have a holistic view of how to monitor astronaut health in space.

NASA technical reports are an excellent source of information. A great paper by our leaders @lauren.sanders and @rtscott2001 (link below) gives valuable insights that can help us refine the system.

Check it out here: NASA Health Monitoring Systems - NTRS

Key Insight from the Report:

  • Proposes an AI system that is pretrained on Earth clinical knowledge and refined during deployment via active learning from sensor streams, astronaut data, and human-in-the-loop feedback.

  • Emphasizes distribution shift in spaceflight data, edge computing, and dimensionality reduction.

  • Synthetic data streams to mimic long-duration mission inputs are also recommended.

Another important paper related to medical monitoring is:

Supporting Crew Medical Decisions on Deep Space Missions: A Real-Time Performance Monitoring Capability

Highlights:

  • Discusses the importance of continuous monitoring of subtle performance decrements.

  • Provides a taxonomy of CDSS functions (vitals, lab/imaging interpretation, medication-related support).

  • Highlights classic CDSS failure modes like:

    • Disruptive alerts

    • Excessive false alarms

    • Workflow disruption

    • Poor data quality

    • User distrust/over-trust

    • Poor usability

4 Likes
46

3. Biomedical and Health Data from Space Missions

To feed our system with data, we’ll focus on human data, realistic data, and space-based data. To start, I found two key datasets: Inspiration4 and the NASA Twin Study.

As we progress, we can explore generated data, Earth-space analog data, and multi-species data to further enhance our model.

Inspiration4 Datasets

All data is publicly available on OSDR:
Inspiration4 Datasets on OSDR

Key Data Types:

  • Blood Samples:

    • OSD-569 — Whole Blood

    • OSD-570 — Peripheral Blood Mononuclear Cells (PBMC)

    • OSD-571 — Blood Plasma

    • OSD-575 — Blood Serum

  • Swab Samples:

    • OSD-572 — Skin, Oral, and Nasal Swabs

    • OSD-573 — Environmental (Dragon Capsule)

  • Biopsies:

    • OSD-574 — Skin Biopsies

  • Excreta:

    • OSD-630 — Stool

    • OSD-656 — Urine

  • Cells:

    • OSD-687 — T-cells

NASA Twin Study

NASA Twin Study Data
This dataset is available upon proposal and request, offering comparative health data from identical twins—one in space and one on Earth.

4 Likes
47

Thank you. @AliReza-H The twin study has some interesting high risks that could be difficult to manage reactively via CDSS , sans, cognitive decline , vascular / GI changes that may indicate an effective CDSS may benefit from some more proactive components to reduce risk as well as the clearer interventions to be used once damage is flagged.

4 Likes
48

Thanks, @Wilester2025 .

I agree that we need a more proactive care approach—one that allows us to monitor, predict, and ultimately prevent occurrences, rather than only reacting once issues appear. Also, each person’s health panel is unique; for example, someone may naturally live with lower hemoglobin or oxygen saturation levels that are healthy for them but would be considered abnormal for the average person.

Because of that, the datasets we use should ideally be longitudinal, covering health data before, during, and after flight. For example, in Inspiration4, the dataset appears to span roughly 92 days before flight to about 194 days after return, which provides a useful temporal profile.

I don’t currently have access to the NASA Twin Study dataset. Compared with Inspiration4—which was a 3-day mission—the Twin Study involved a 340-day mission, so it could capture more realistic or longer-term health events. However, I’m not sure whether it includes a comprehensive pre- and post-flight health panel.

It might be worthwhile for us to apply for access to the Twin Study data as well.
@lauren.sanders — would that be possible?

2 Likes
49

Agreed. Also an up-to-date copy of the Integrated Medical Model with the 100 possible medical emergencies . Old copy review V3 2015 https://ntrs.nasa.gov/api/citations/20150002713/downloads/20150002713.pdf

3 Likes
50

And perhaps if there was an uptodate version of the human risk frame work from the risk management section.

I do so love DAGS - so much thought involved

2 Likes
51

Not my strength in any way ever but DAG to something like python ?

2 Likes
52

Thank you @Wilester2025 for sharing this. It is wonderful, and I think it is fundamental to our work. It builds on theHealth System Risk Board’s work and is also related to the Causal Inference Subgroup’s work, which we need for explainable AI in our prototype.

Regarding translating DAGs into Python, yes, we can do that. Nowadays, with large language models, we can essentially translate from almost any language into another. The authors included DAGitty-based code in Appendix C for all 29 scenarios.

I spent a few hours translating all of them and uploaded the results to my GitHub repository ( GitHub - Alirezahayatimedtech/CDSS_Prototype · GitHub )

You can simply click on “Open Appendix C DAG Viewer” on the README page and use it without cloning any code. It is not perfect, but at first glance it looks useful and reasonably well documented.

If the group finds it helpful, this will soon be integrated into the main AWG GitHub repository.

Thank you again for your help. We can discuss this further in our next CDSS meeting.

image
image1368×1764 314 KB

4 Likes
53

Hi everyone,

Thank you all for the enthusiastic response and support. I’m truly grateful for the collaboration and willingness everyone has shown.

Members who expressed interest in working on problem research are welcome to start contributing their research in the shared Word document.

For additional context and resources, please find the following materials:

I sincerely appreciate the effort and teamwork from all members contributing to this subgroup. Looking forward to our continued collaboration.

Best regards.

3 Likes
54

Hi everyone,

In our last meeting, many showed interest in contributing to the problem definition and research. Please find the attached Word document. I kindly request everyone to add your name and include your contributions in the document at your convenience.

I will also update the Google Drive link shortly.

Thank you all for your valuable contributions.

Word document link: Problem Research - Google Docs

Link to all the other resources: CDSS subgroup project - #53 by Ritika_Saha

@vaishnavi.nagesh @luis.depombopuerta @zhooper @ccnaney @Dayeon @robertn01 @dr.aslinger

3 Likes
55

Reading through the Read me work you did. as well as each person having their own health panel perhaps there is a need for a mission health panel as well as 1 or more people with a health issue may impact the viability off the mission as a whole.

1 Like
56

Perhaps the mission health panel could also manage the proactive processes. 2 weeks to mars start reactivating musculature system, day 30 on voyage review psychological state and interactions, unexpected period of increased radiation review each person in 3 days etc.

1 Like
57

Hi @alavia . If this is some of your work it sounds exciting and perhaps complementary

1 Like
58

Hello everyone,

I have researched a few key datasets that are relevant to our ultimate goal of building the CDSS framework. These include RNA/multi-omics datasets for tracking gene expression and immune response, metagenomics datasets for understanding microbiome dynamics and pathogen detection, environmental sensor data for monitoring habitat conditions, plant image datasets for crop health analysis, and radiation biology datasets for modeling DNA damage and long-term risks.

The document provides a structured overview of these datasets, their relevance, and how they are integrated into the ML pipeline through preprocessing, feature engineering, and modeling. It also explains how different agents like Human Health, Agriculture, Habitat, and Radiation will utilize these datasets to generate meaningful insights.

Additionally, the document highlights the problem relevance, emphasizing that current space health and habitat monitoring systems are fragmented, reactive, and heavily dependent on Earth-based intervention. It points out the lack of integration across domains such as human health, environment, and microbial systems, which limits early anomaly detection and proactive decision-making.

In this context, the datasets play a critical role in enabling a unified, multi-agent CDSS that can continuously monitor, predict risks, detect anomalies early, and support autonomous, real-time decision-making for long-duration space missions.

CDSS Research Doc 1 20 March'26
CDSS Research Doc 1 20 March'261414×2000 535 KB

CDSS Research Doc 1 20 March’26 (1).pdf (5.2 MB)

3 Likes
59

Hi everyone,

I have updated the Problem Research Documentation with relevant research papers and references on CDSS for space missions, as shared by everyone.

Anyone who would like to contribute or add additional studies can update the document as well.
Thank you!

Problem Research Word document: Problem Research - Google Docs

3 Likes
60

@Ritika_Saha Thanks, Ritika! This is great. I’ll add some contents too!

3 Likes