Printing tissue from the tip of a catheter without opening the body — computing backward to a gel that flows under shear and holds when still
The National Science Foundation awarded roughly $1.64 million to Harvard University to design hydrogel composites for in situ transcatheter bioprinting by combining machine learning with experiment. The materials must behave in opposing ways: flowing while under shear, and holding mechanical properties when at rest.
Grant overview (primary data)
- Award amount$1,643,000
- RecipientHarvard University (Massachusetts)
- ProgramOFFICE OF MULTIDISCIPLINARY AC, BIOMATERIALS PROGRAM
- Period2026-10-01 〜 2030-09-30
- FunderU.S. National Science Foundation (NSF) / NSF
Key points
- Designing hydrogel composites that flow under shear and hold mechanical properties at rest, by combining machine learning with experiment ($1,643,000).
- The application is in situ transcatheter bioprinting — filling patient-specific tissue defects without open surgery.
- The technical focus is unraveling the tightly coupled processing-structure-property relations.
- Tabular foundation models are used in a human-in-the-loop structure for prediction and for conditional inverse design with quantified uncertainty.
- The institution is Harvard University; Massachusetts accounts for 8 of the awards this site holds as of 2026-09-02.
1It has to flow, and then it has to hold
Delivering material into the body to fill a tissue defect demands opposing properties. While passing through a narrow tube it must flow; once placed, it must keep its shape. Hydrogels are candidates because they can behave differently under shear than at rest. This project aims to make that switch something that can be designed for deliberately.
2Processing, structure and property
- 1ProcessingHow it is synthesized and assembled
- 2StructureHow multiscale porosity arises and evolves
- 3PropertyWhat flow behavior, viscoelasticity and mechanics result
- 4Inverse designStarting from the application and solving for the composite that has the needed properties, with uncertainty quantified
In materials science, working out the relations among processing, structure and property is the basic task. The difficulty with this material is how tightly the three are coupled. The project deploys tabular foundation models in an iterative, human-in-the-loop structure, using them in both directions: prediction (what properties follow from this processing) and generation (what composition yields these properties).
3Repairing without surgery
The named application is in situ transcatheter bioprinting: filling patient-specific tissue defects without open surgical intervention. Material is delivered from the tip of a tube to repair deep tissue, and the demanding set of conditions on the material follows backward from that goal. It is fundamental materials research with a concretely specified application.
4People and outreach included
Another central activity named is creating the next generation of leaders in AI-driven materials design, discovery and translation. Education, outreach and workforce development span K-12, undergraduate and graduate students and the public, through museum exhibits, research experiences for undergraduates, Rising Stars symposia for graduate students and public lectures.
The 120 NSF awards this site holds as of 2026-09-02 come from 93 institutions, with Massachusetts accounting for 8 by state. Amounts are the obligated amount as of the check date and may change.
Why it matters
Fundamental materials research with a concretely specified application. Using AI not only for prediction but for inverse design, with the uncertainty quantified, is a pattern that carries across materials development.
FAQ
Why does it need opposing properties?
What is inverse design?
Sources (primary)
Source: NSF Award Search (U.S. National Science Foundation, public domain). Amounts are the obligated amount. For privacy, we do not handle principal investigator names.
- NSF Award (original, official)
- NSF Award ID: 2522538