In October 2011, Paul Zygielbaum was trapped in a business meeting when he started feeling unwell. Zygielbaum was then CEO of a startup called C8 Medisensors, which was developing a smart-phone connected blood sugar monitor, and he has type 2 diabetes. He suspected that his blood sugar was dropping. Stuck in the meeting and surrounded by pastries and Jolly Ranchers, he popped one of the hard candies into his mouth and pulled out his phone. It displayed readings sent wirelessly from a C8 monitor strapped to his abdomen. The phone confirmed that his blood glucose was below normal. As the sugar in the candy took effect, “I was watching my glucose go back up on the phone,” Zygielbaum remembers.

The other people at the meeting took notice. “They weren’t worried about me,” he says. “They were stunned by the technology.” One of them, Wade Randlett, an investor, watched as Zygielbaum showed the results to another attendee and was “impressed enough to end up writing him a check,” Randlett recalls.

C8 Medisensors was tackling a long-standing and tricky problem: creating a convenient and discreet way for people with diabetes to monitor their blood sugar continuously, without having to draw blood.

It was a technological challenge that some 70-odd companies had already tried and failed to overcome, Zygielbaum says, but by the time of his impromptu conference room demonstration, C8 seemed to have hit on the solution. The company had more than $60 million in funding from investors including GE Capital, GE Healthcare, and private individuals, and a manufacturing facility in San Jose was on board. A year later, in October 2012, European regulators approved the sale of its system. Seemingly on the verge of commercialization, the company began taking reservations through its website.

But before the end of that winter, C8’s amazing run would end. The technology was good but needed to be a bit better, and manufacturing was inconsistent. The company needed more money, perhaps $15 million, to fix those issues, says Zygielbaum. C8 couldn’t raise it.

Without that cash, C8 quickly failed, closing its doors in February 2013. It was an illustration of how many things—technological, financial, and human—must come together perfectly for a startup to make it in mobile health.

This was not the coda C8’s founders had imagined 10 years earlier, when they formed the company after testing ideas in a laser lab built in cofounder Robert McNamara’s backyard. McNamara and Jan Lipson, two of Zygielbaum’s old Caltech roommates, contacted him to see if he would help. Zygielbaum, a mechanical engineer and MBA, was undergoing cancer treatment at the time, but their idea for a noninvasive way to monitor glucose was so compelling he signed on as a cofounder.

An ideal glucose monitor would avoid breaking the skin, to prevent pain and risk of infections. It would also provide nearly continuous, automatic measurements, to make tracking blood sugar less intrusive and to better meet the needs of insulin-dependent diabetics, who need to keep tight tabs on their blood glucose.

Current options for checking blood sugar are inconvenient and uncomfortable. The most common involves pricking a finger to apply a small drop of blood to a testing strip, which is then read by a glucose meter. While highly accurate, these measurements provide only intermittent peeks into blood glucose levels that change throughout the day. Patients can also wear a continuous glucose monitor that has a needle-like sensor inserted under the skin and usually attached to a larger device taped to the body. Such monitors provide readings every few minutes, and they help people avoid the devastating effects of blood sugar lows or highs, but they’re less accurate than a blood test. Both require blood-drop readings multiple times a day for calibration.

C8’s team thought it could use a measurement technique called Raman spectroscopy. Its device would be strapped around the abdomen and would shine a beam of light into the skin to measure glucose levels in the fluid that bathes skin cells. It could send readings wirelessly to a smartphone for a near-continuous record of glucose changes throughout the day.

The potential market was huge. Today, 382 million people around the world suffer from diabetes, according to the International Diabetes Federation. Market research firm GlobalData estimates that the worldwide market for blood glucose monitoring was $8.9 billion in 2010 and could grow to $12.2 billion by 2017.

To compete in that market, C8 first had to overcome a number of technical challenges. “Glucose is a small molecule in and of itself,” explains Ishan Barman, a bioengineer at Johns Hopkins University who is also trying to develop a noninvasive glucose tracker based on Raman spectroscopy. “There are lots of other [compounds] in the bloodstream which are found at a much higher concentration,” he says. Because noninvasive systems don’t directly test blood glucose, the changes in their readings can lag behind direct readings of the blood, he says. Noninvasive measurements do have advantages, however. Because they can be taken nearly continuously, they can be used to predict whether future glucose levels will move higher or lower. Continuous measurements also provide immediate feedback about the body’s response to treatments.

Developing software to translate the spectrographic readings into glucose levels is tricky, says Barman. Because the light used in Raman spectroscopy doesn’t penetrate very deeply, the system calculates blood glucose levels on the basis of glucose levels in the skin, and that calculation can vary from person to person, he says.

That variability from user to user, and even from spot to spot on the same user, proved to be C8’s greatest challenge. Some problems were driven by manufacturing variability, says Zygielbaum. The way the device was mounted on the skin could change the readings as well.

Compounding those technical challenges were a series of incredible personal misfortunes. One of the founders, Jan Lipson, died in a biking accident in July 2010, just as the first prototypes of the final, miniaturized version of the device were being tested by volunteers. The progression of Zygielbaum’s cancer slowed but later picked up again, which was probably what caused him to develop diabetes. Zygielbaum had to step down in the middle of the last round of fund–raising for the $15 million, and his replacement died in an ice-hockey accident just five weeks after.

“That took the heart out of the investors,” says Zygielbaum.

Even under the best of circumstances, funding the development of a new medical device is difficult, says Fred Toney, chief financial officer at C8 during its final two and a half years. “One reason the venture capital community has increasingly been stepping back from funding private medical-device companies is they are hard and take a lot of capital,” he says.

After C8 shut down, the technology was sold to a small private equity group that continues to work on it, says Toney. “Whether used for glucose or other things, it’s a technology that will certainly come to market,” he says.

Some diabetics had been closely watching C8’s progress. “If it had worked with sufficient accuracy, I believe many would have jumped at the chance of buying C8’s product,” says Mike Kendall, who was diagnosed with type 1 diabetes in 1991 and writes a well-read blog about living with the disease. “I know I would have.”

Others continue to chase the dream of a better glucose monitor. An Israeli company called Integrity Applications expects to begin selling its GlucoTrack system, which detects glucose using ultrasonic, electromagnetic, and thermal measurements through an ear clip, in Europe this year. Barman’s team is starting to test its system as well. Despite C8’s demise, he thinks the field has advanced impressively. “I see this as a much more hopeful time than 10 years back,” he says.