Tiny magnetic millirobots could shorten heart infection treatment to hours

Endocarditis is hard to treat because antibiotics given through a vein get diluted on the long trip to the heart. By the time the drug reaches the infected valve or tissue, there often isn’t enough of it to wipe out the bacteria.

That leaves doctors with two uncomfortable choices. They can raise the dose, which risks toxic side effects, or they can send the patient for open-heart surgery to clean out or replace damaged tissue. Neither is something a patient wants, and both mean long hospital stays.

The idea behind the new robots is simple: if the drug can be placed directly on the infection, a small amount should do the job, and the rest of the body is spared.

The system uses two robots that work as a team rather than one bigger device. Each is roughly as wide as the tip of a pencil or the head of a match, and about as long as a grain of rice.

• The needle module goes first. It grips the heart wall near the infection with a tiny barbed needle and stays put, acting as a parking spot.
• The drug module carries the antibiotic. It travels to the needle, snaps onto it with magnets and slowly lets the medicine out over the infected area.

Because the anchor stays in place, an empty drug capsule can undock and be pulled back, and a fresh one can be sent to the same spot for another dose. Nobody has to re-anchor a new robot each time.

Splitting the job in two also keeps each robot small, which helps them squeeze into tight corners of the body that a single larger device couldn’t reach. First author Anuruddha Bhattacharjee, a former Johns Hopkins postdoc who now teaches at the University of Alabama at Birmingham, has described the anchored robot as a miniature docking station.

This is an incredible form by @AIatMeta pic.twitter.com/fTcUKkvnqu

— Digital Trends (@DigitalTrends) October 12, 2026

Neither robot carries a power source. Both contain small magnets, and a set of electromagnetic coils placed around the body steers them wirelessly, using the same kind of magnetic fields and gradients found in an MRI scanner.

Senior author Axel Krieger, a medical roboticist known for robots that can operate without a human at the controls, says the appeal is that the robot itself never needs power. Instead of making pictures, the magnetic fields push and pull the devices into place.

Getting in is meant to be minimally invasive. Krieger describes the entry as “a little pinprick”, with no infection or blood loss, and says a surgeon would not need to steer the device by hand. Bhattacharjee talks about injecting the robot into a vein and letting it travel to the target.

So far the robots have been tested only in the lab, not in animals or people. The team ran them through artificial heart models made of gel that mimic a human heart, including pulsing fluid to simulate blood flow, and the robots moved on their own against that flow.

In those experiments the two modules docked successfully on every attempt, and undocking worked in the large majority of tries. The team says its barbed needle design took far less force to push into tissue-like material than a smooth one, which matters when you are trying to anchor something gently on a heart.

The researchers then used pig heart tissue outside the body to check two key steps: that the needle robot could pierce real heart tissue, and that the drug module could undock afterward. Both worked.

If everything goes to plan, Bhattacharjee says one delivery might be done in roughly 10 minutes, and repeat doses within about an hour. That is where the headline claim comes from: turning four to six weeks of treatment into one day or less.

For patients, the payoff would be huge. A day or less of targeted treatment instead of more than a month of IV drips could mean shorter hospital stays, fewer side effects from high antibiotic doses and, for some people, avoiding open-heart surgery entirely.

The catch is timing. The researchers themselves say they are at an early research stage and that much more work is needed before the robots can be used in people. A gel model and a piece of pig heart on a lab bench are very different from a living heart that never stops beating while blood rushes past.

Moving tissue is the next big hurdle. A beating heart could throw off how the robots steer, anchor and dock. The team also wants to track the robots with medical imaging such as ultrasound or X-rays, rather than watching them with a camera as they did in the lab.

The next step is testing in more realistic phantom hearts that contract and beat like real ones, with simulated blood flow. Only after that would tests in living systems come into view.

The team also sees uses beyond infections. The same docking idea could deliver cancer drugs to tumors where they are spreading, or help with other minimally invasive procedures in hard-to-reach parts of the body. Krieger calls targeted, steerable robots like these the frontier of surgical robotics.

If you or someone you know is being treated for endocarditis today, nothing changes yet: IV antibiotics and, when needed, surgery remain the standard. This is research to watch, not a treatment you can ask for.