I believe @jgong has a meeting of this subgroup project tomorrow Friday, at 10am PT
You should see it on your calendar
1 Like
Thank you and great to have you joining, Aditya. If you prefer email calendars, feel free to PM me your email address and I will add you (and anyone else) to our calendar invite. We will talk about projects and ideas during the meeting.
Jian.
2 Likes
Thanks for sharing the roadmap, @Jgong. I’d be happy to contribute on the benchmark and evaluation side. One area I can help with is designing a standardized annotation/evaluation framework (input schema, expert labels, confidence, uncertainty, and evaluation metrics) along with a literature review of existing SANS/CDSS approaches so our benchmark aligns with current research. If that would be useful, I can start drafting a document for the team.
Here’s my mail hitaeshi25sehgal@gmail.com. Looking forward to collaborate!
1 Like
Dear all -
I will not be able to host a digital twin meeting tomorrow due to travel plans. No meeting tomorrow. The next week I will have to see as well. Enjoy your summer!
Jian.
2 Likes
There is no meeting today. Sorry all. I could not make time. We will resume meeting next Friday. Thank you!
3 Likes
Hi All !
I was reading some papers and essays on Digital Twin advances , sharing them with you here 
https://www.nature.com/articles/s41587-026-03121-4
https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.124.070526
https://www.nature.com/articles/s41746-025-01679-y
3 Likes
New member here - interested if there is still work to contribute. I will try and peruse all these links/github forks from the early 2024 post onward…
Matthew Gebre
'12 B.Sc. Engineering, Biomechanical at Stanford University
'19 M.D. at Stony Brook University
2 Likes
Hi everyone & @jgong quick update on the OSDR ChatBot work.
The chatbot is still an early work in progress, but it started as a simple command-based way to search and explore OSDR studies. Since the earlier demo, I have been building on that foundation and added an AWG comparison workflow, so we can compare two OSDR studies side by side and organize useful details before moving into tomography, imaging, or omics work.
A few things I have added so far:
- AWG study comparison for organism, tissue, assay, platform, experimental conditions, and study compatibility
- Clearer separation between repository metadata, reported findings, interpretation, and follow-up ideas
- Study-comparison visuals and visual abstracts grounded in OSDR metadata
- Gemini-family model support for the chatbot and media workflows, including image generation for study visuals
- Early Veo support for short scientific-motion and relatable clips, with provenance and fallback handling
- More detailed provenance so generated outputs can be traced back to the study pair, model/provider, and evidence framing
- Early work on a controlled Computer Use/controller-style feature
For the digital-twin work, I see the chatbot mainly as an upstream discovery tool: it can help identify and compare relevant eye, retina, optic-nerve, spaceflight-analog, and omics studies before bringing selected datasets into the tomography or future spatial-transcriptomics workflows.
It is nowhere near finished, but I wanted to share the current direction and get feedback early.
- Current production build from my session this week: OSDR ChatBot
- Short walkthrough: Video Walkthrough
- Code(forked) repository: OSDR_Chatbot on GitHub
Would it be useful for me to open a small PR once we identify an integration point between the chatbot output and the tomography/visualization work? I would also appreciate suggestions on which eye, retina, or spatial-transcriptomics datasets should be prioritized next.
Thanks.
-Gaston D.
3 Likes
Hi Gaston -
We would love to have you do a small demo/informal talk in one of our Friday digital twin meetings. Would it be possible? We can record it for more people to see (or we can wait and not to record it until later when you feel comfortable). If the 10am PST time not working out for you, happy to find an alternative time. I am excited about this work! Thank you for sharing it!
Jian.
2 Likes
Hi Jian,
Glad you like’d it, was working on this specifically all week, and got further than I expected with it too.
Yeah, I’ll try to make it this friday.
Thanks again, I would’ve done more configuring & integration for my chatbot updates with this specific tool GeneLab AWG Tomography you shared earlier but I saw that after I got done. 
Maybe I can pick your brain & everyone’s on integrating it as a /awg command for the chatbot so it pops up inside it.
Either way pretty cool, & I had a pretty good time building it out this week!
Thanks again, I appreciate the opportunity.
1 Like
Interesting paper for our world-model / digital-twin discussions: RoboPhys-3D
I came across this very recent paper, “RoboPhys-3D: A Comprehensive Embodied World Model Evaluation via 3D Reconstruction” (Aug. 2026), and thought it could be particularly relevant to our discussions around world models, synthetic data, digital twins, and trustworthy AI for long-duration spaceflight.
The key idea is that evaluating a world model based on whether its generated video looks realistic is not enough. RoboPhys-3D evaluates generated rollouts at several levels — 3D geometry, persistent scene state, and ultimately task/execution success — using the same 3D reconstruction pipeline for generated and ground-truth data.
One point I found especially interesting for our context is that perceptual/VLM-based evaluation can miss failures that become apparent when the model is evaluated against the underlying physical state and whether an action would actually succeed.
That raises an interesting question for our NASA work:
Could similar 3D/state-grounded evaluation principles be applied to spaceflight digital twins or multimodal clinical/operational world models — particularly when synthetic data and simulation are being used to predict astronaut states or support decision-making?
The paper also seems useful as a reference point for thinking about what an open, reproducible evaluation framework for NASA world models might look like.
Paper: RoboPhys-3D — arXiv
Would be interested in hearing whether others see a connection to our current digital-twin / synthetic-data / CDSS work.
2 Likes
Hello @jgong, I am new to the AWG and I found this project really interesting. I would love to join and contribute.
2 Likes
Hi all -
I am sorry that I am out in the field today and the limited network is preventing me to host the digital twin meeting. Thanks for all the interests and I will message individually to give you feedback. Congratulations to people who have abstracts accepted for presentation at ASGSR!
Jian.
1 Like
Hi there,
Just saw this and no worries!
If you’d like to go over my /awg in-command build for the OSDR_chatbot, you can easily test it out and find it here in the mini-hub I made for it: https://awg-osdr-chatbot-show-2026.ai.studio
You can find the production and repo(forked) links on this page that I planned to cover today, as well as your specific work that I plugged in, too @jgong I’ll be happy to go over it again for the next meeting.
Also, the walkthrough is available here: https://youtu.be/zQ2LPgYweps
Enjoy the day/night & weekend ahead!
-Gaston D.
1 Like