Kevin Plaxco, PhD, received an email with the subject line, “Rats 3, postdoc 0.”

Because some days the rats win.

Plaxco has spent two decades trying to build a biosensor that works for longer than a day in the human body, measuring any molecule in real time. Preliminary results and shorter runs — including an in vivo success in 2017 — suggested that he was getting close.

Lab rats were the final hurdle. “They’ve got these prehensile thumbs,” Plaxco said. “They’re very clever at figuring out how to remove the device.”

Plaxco’s team attached tiny cages to the rats’ backs, protecting the sensors resting just beneath the skin. But the rats figured out how to remove the cages too. The postdoc leading the effort kept sending Plaxco emails tallying the rats’ victories.

Finally, after several iterations, a three-dimensional printed housing worked. The biosensors lasted for 7 days, accurately monitoring levels of the antibiotic tobramycin.

“We showed we could push the measurement duration to a week, making a measurement every 12 seconds,” Plaxco said. “Working for longer than a day is a major breakthrough, because a device that needs to be replaced every few days is way better than one that needs to be replaced several times a day.”

Creating reliable biosensors has been quite the quest, and not just for Plaxco. Long-lasting biosensors could help manage everything from drug dosage to chronic disease. Scientists are developing biosensors to measure inflammation, antiretroviral drugs, NT-proBNP (indicating heart failure), and cortisol, for instance.

Plaxco’s week-long results, detailed in the Journal of the American Chemical Society in March, came on the heels of another milestone. His biosensor recorded vancomycin concentration in six healthy humans in 5-minute intervals over 24 hours — the first-ever clinical trial of a biosensor to continuously monitor drug levels. Vancomycin is a notoriously difficult-to-dose antibiotic that can treat serious infections, such as sepsis, but can have severe side effects, including kidney damage.

“Real-time information on how much of that drug is in the patient’s body is going to save lives,” Plaxco said.

The ‘Holy Grail’ of Biosensors

There’s only one commercially successful biosensor: the continuous glucose monitor (CGM), which became widely available in 2004. Two CGMs were recalled this year after being linked to seven deaths and more than 800 serious injuries — a fair indication of how difficult it is to make a biosensor work, let alone be reliable. CGMs use a naturally occurring enzyme to chemically convert glucose into a different species, producing a measurable electric current. That limits their applicability.

“The glucose sensor is critically reliant on the chemical reactivity of its targets, and that’s what prevents it from being generalized to very many targets,” Plaxco said. “Our technology is completely agnostic to the chemical reactivity of the target. We don’t have to transform it to anything else.”

A molecular biophysicist by training, Plaxco grew so convinced of biosensors’ potential that he became an engineer in the late 1990s so that he could build the technology himself. Now a Distinguished Professor of Chemistry at the University of California, Santa Barbara, he first demonstrated an electrochemical aptamer-based (EAB) sensor in 2005. The generalizable platform is designed to be able to detect any molecule in serum. Plaxco and others have used it to measure more than two dozen different molecules in live rats. 

EAB biosensors use DNA aptamers (lab-generated DNA or RNA molecules) that act like antibodies and bind to specific targets. Some labs focus on developing aptamers to target certain molecules. But the key to Plaxco’s technology is something called an aptamer switch, a reengineered aptamer that changes shape when it binds to a target. Plaxco developed an electrochemical method of “reading” that shape change, which produces a signal for the sensor.

It’s all inspired by the human body.

“Evolution invented ways of measuring molecules in your body a long time ago, and it works great,” Plaxco said. “That’s been the holy grail of the biosensor field, to also be able to measure arbitrary molecules in the body in real time.”

But because aptamers are DNA, nucleases in the body will degrade them after about a day. Scientists call the resulting loss or change in electrical signal “drift.” Plaxco and H. Tom Soh, PhD, Stanford University, Stanford, California, teamed up in 2017 to figure out how to correct the drift and recover accurate measurements. It worked well for many hours, but the drift eventually depleted the signal to unusable levels.

Plaxco tried something else. Rather than a traditional DNA aptamer, he used a nonnatural analog to DNA, “a modified ribose sugar that doesn’t occur in nature but has more or less the same shape as ribose, so you get more or less the same function,” he said.

The nonnatural backbone is much more resistant to nucleases and slowes down degradation, paving the way for the 1-week in vivo success. His team stopped their experiment to adhere to animal ethics permissions, but “the signal noise was still excellent,” Plaxco said. “I’m quite convinced it would go for 2 weeks.” 

Complementary Approaches

Combining aspects of different sensors could result in a more effective one, Plaxco said. For example, Soh and others — including Netzahualcóyotl Arroyo-Currás, PhD, and Jason C. Heikenfeld, PhD — have each used different coatings to reduce drift. The membranes reduce the rate at which proteins that can damage the sensor, such as nucleases, reach its surface. The membranes also have pitfalls: they can complicate sensor fabrication and insertion into the body and may slow down sensor response times. 

“Using a nonnatural backbone reduces the damage caused by the proteins, so if you could simultaneously reduce the rate with which proteins get there and reduce the damage that they cause when they get there, those two things are complementary,” Plaxco said. 

A 2025 paper showed that Soh’s biosensor, SENSBIT, tracked antibiotic concentration in live rats, retaining more than 60% of its signal after 1 week. It uses a human gut mucosa-inspired membrane to protect against drift.

“There are remaining hurdles that have to be overcome before it’s translated to people. Physical stability is one,” Soh said. “But we’ve shown that; here’s proof of principle. This is possible.”

While it works well for small molecules, Soh said, SENSBIT is unlikely to effectively measure larger molecules, like proteins.

Shana Kelley, PhD, Neena B. Schwartz Professor of chemistry and biomedical engineering at Northwestern University in Evanston, Illinois, has been focused on biosensors that can continuously measure cytokines — proteins associated with inflammation. In 2024 proof-of-concept experiments, Kelley’s “molecular pendulum” biosensor measured inflammation biomarkers in the interstitial fluid of anesthetized diabetic rats for about 6 hours. The sensor uses electrical currents to both detect when the protein binds and shakes off, generating measurable changes.

“It’s these two ways of using electric fields and currents to both detect and then reset so that we can see concentrations going up and down,” said Kelley, who is also the president of bioengineering at Biohub, Chicago, a biomedical research center.

The mechanism behind the “reset” of Kelley’s sensor has perplexed some researchers. 

“I don’t think we understand that process very well — the regeneration of the receptors and, therefore, the regeneration of the signal when you’re trying to detect a protein with one of their sensors,” said Arroyo-Currás, associate professor of chemistry at the University of North Carolina at Chapel Hill and the lead author of the 2017 study demonstrating Plaxco’s technology in rats for several hours. “If you look at some of the data, it doesn’t seem like it fully recovers.”

On the Cusp

Scientists are in a sprint to bring their products to market. Plaxco is working with the Australia-based company Nutromics to gain FDA approval for a biosensor that would measure vancomycin. Adaptyx Biosciences, a spinoff of Soh’s lab, recently presented a cortisol sensor that worked for multiple days in humans. But researchers say there are still many questions and unknowns that could get in the way of commercialization.

“There is no area of startups that is a bigger graveyard that I’m aware of than biosensors,” said Heikenfeld, professor of electrical engineering and biomedical engineering at the University of Cincinnati, Cincinnati, focused on developing EAB sensors that can help manage chronic disease. “The impact is huge, but it is incredibly hard.” 

Kelley’s molecular pendulum EAB sensor, for instance, depends on the “pristine gold surface” of the electrode, she said. Her lab scaled a method of cleaning each electrode individually, using electrochemical etching to form a layer of gold oxide on the surface, and then stripping the oxide off.

“The chemistry of the surface is what sets the stage for everything that happens afterwards,” Kelley said.

She’s also trying to develop a sensor that can “look at multiple markers at one time in parallel,” such as inflammation and other markers of diabetes. Her June 2026 research shows how it could work by adding progesterone-detecting transcription factors to molecular pendulum sensors.

Arroyo-Currás’ lab is focused on answering fundamental open questions about biosensor technology. For instance, he said, “Is it even realistically feasible to monitor proteins on the skin in a manner that correlates with physiologic status? I don’t think that’s known. I think there’s a lot of hype saying, yes, you can, but I don’t think that it has been conclusively demonstrated.”

His lab is trying to demonstrate that different molecules transport into the skin of rodents and that molecules transport with different time constants. The investigations may give scientists more clues to the feasibility of subcutaneous biosensors for monitoring proteins.

In the meantime, Arroyo-Currás is optimistic that an aptamer-based continuous biosensor will become available within 5 years.

“There’s companies out there that are testing it in humans already,” he said. “This is not science fiction anymore. It’s going to happen. The question is, how?”

Soh reported having a financial interest in Adaptyx Biosciences, CHIPSENSE, and Range Biotechnologies (spin-off companies from his lab). Shana Kelley reported being a founder and equity holder in Arma Biosciences, a startup that is commercializing pendulum technology, and reported holding patents related to her continuous monitoring work. Heikenfeld is co-founder of Kilele Health. Plaxco leads the Biosensor Advisory Board ofNutromics, a company focused on the clinical translation of EAB sensors. Arroyo-Currás reported having no conflicts. Disclosure information for study authors is available in the original study publications.