I’m introducing a new project idea for the AI/ML AWG focused on mapping molecular signaling across disparate tissues. While most current spaceflight transcriptomic analyses are siloed to single organs, this project aims to leverage multi-modal machine learning to identify systemic “cross-talk” patterns.
Summary Spaceflight induces complex physiological adaptations that are fundamentally interconnected. The OSD-914 (RR-8) dataset provides a unique opportunity for systemic modeling, as it contains small RNA data across 13 different organs for the same subjects. This project will utilize this high-dimensional dataset to determine if molecular stress signatures in peripheral tissues (e.g., Liver) can serve as predictive biomarkers for adaptations in the Central Nervous System (Brain).
Research Question: Can a multi-modal neural network architecture identify predictive transcriptomic signaling links between hepatic metabolic stress and neuro-inflammatory responses in spaceflight models? The hypothesis is that systemic signaling creates identifiable correlations across the OSD-914 organ-wide profiles that are missed by traditional single-tissue differential expression analysis.
Deliverables: The expected deliverables include:
A standardized cross-tissue data matrix derived from OSD-914.
An open-source AI pipeline (Python/PyTorch) for multi-organ predictive modeling.
Explainability maps (SHAP/LIME) identifying the specific miRNA families driving inter-organ correlations.
A manuscript submission to a peer-reviewed journal (e.g., Nature Scientific Data or Life).
I am looking for collaborators with expertise in Systems Biology, Transcriptomics, or ML Optimization to help with biological validation and feature selection.
If you’re interested in collaborating on this systemic approach, please reply here or reach out directly!
Best regards,
Pratheek Mukkavilli
NASA AI/ML AWG Member @AIMLawg
Hello Pratheek, thank you for sharing this project idea. I found it very interesting, especially because it approaches spaceflight biology from a systemic perspective rather than limiting the analysis to isolated tissues.
I would be very interested in collaborating on this work. My interest is particularly in the biological interpretation of cross-tissue signaling, the development of systemic hypotheses, and the translational relevance of molecular stress responses across organs, especially in relation to neurobiological adaptation, inflammation, and physiological regulation.
Although I am still expanding my technical depth in transcriptomics and machine learning pipelines, I believe I can contribute meaningfully to the conceptual and biological validation side of the project, including literature integration, hypothesis refinement, and interpretation of liver-brain or peripheral-central signaling patterns.
This is a very promising direction, and I would be glad to contribute and learn alongside the team if there is an opportunity to participate.
Hi @Pratheek – make sure to raise this with @vaishnavi.nagesh & @lauren.sanders to make sure that you have a chance to pitch this proposed project at the May AI/ML AWG mtg
In this model, gravity (Δg) is the primary driver, and β (mechanotransduction sensitivity) dictates how each tissue responds.
Your project is interesting to me because it directly tests a key implication of this framework: that different tissues (liver vs. brain) will have distinct β values, yet show correlated dynamics under the same Δg stress.
I would be interested to see if your multi-organ data supports this prediction. If you are open to collaboration, I would be happy to help with the theoretical interpretation of your results.