Google’s Smart Contact Lens Initiative: What Went Wrong?

In early 2014, Google kicked off an ambitious project that had the potential to make a significant impact in healthcare, especially in Silicon Valley. The company revealed they were developing a prototype for a smart contact lens capable of measuring glucose levels in tears, utilizing a miniaturized sensor embedded within the lens layers. This innovative technology was said to offer glucose readings once per second.

If successful, this lens could have dramatically simplified glucose monitoring for the approximately 830 million people with diabetes worldwide. It also had the potential to elevate Google in the realm of consumer wearables, demonstrating that a tech giant could address a key biomedical issue. It would have positioned Google as a serious contender against traditional medical device companies, making a solid mark in healthcare.

However, the project didn’t pan out as envisioned.

Despite the initial excitement, reports two years later revealed a trail of difficulties. A 2016 investigation highlighted a significant hurdle: tears proved to be an unreliable fluid for gauging blood glucose levels. This project also underscored broader challenges encountered when merging consumer technology with healthcare needs. Just miniaturizing hardware isn’t always sufficient. The complexities of human biology mean that medical devices demand highly precise and reliable readings. For instance, while a device that tracks daily steps can afford some leeway, it’s a different story for something that measures blood glucose in individuals with diabetes.

Today, research continues on a highly sought-after goal in wearables: noninvasive glucose monitoring. The aim is to create devices that avoid contact with bodily fluids and instead identify glucose’s unique molecular signature through the skin, allowing for indirect estimation of blood glucose levels. Remarkably, progress is being made in addressing the intricate physics, chemistry, and materials needed to achieve this objective.

The Challenge of Biological Measurement

The challenge with Google’s smart contact lenses boiled down to a fundamental biological problem: the human body contains a myriad of components, and glucose is merely a tiny part of that mix.

When someone measures glucose by pricking a finger, they apply a drop of blood to a test strip. These strips utilize an enzyme that specifically reacts with glucose, resulting in an electrical current that can be measured to determine glucose concentration. Continuous glucose monitors take this further by measuring glucose present in interstitial fluid just beneath the skin.

The effectiveness of these devices lies in their highly controlled chemistry. The enzymes react almost exclusively with glucose, and glucose levels in interstitial fluid mirror those in the blood. In contrast, tears present a convoluted picture. Their glucose concentration is lower, fluctuating in ways that don’t always align with blood glucose measurements. When you add in the complexity of avoiding bodily fluids altogether, the task becomes even more daunting.

Noninvasive monitoring isn’t a new idea. Devices like smartwatches measure heart rate by detecting changes in blood volume in small vessels near the skin. They shine light into the skin, where blood absorbs more light than surrounding tissue. As heart rates fluctuate, these variations are captured and translated into beats per minute. This technology has made heart rate measurements incredibly accurate.

But glucose measurement doesn’t function in the same way. Blood is a blend of various elements, and finding glucose amidst these complexities poses a significant challenge. Judith Su, an associate professor of optical sciences and biomedical engineering, points out that glucose produces only a minimal signal, especially when compared to water, which can overshadow its presence in measurements.

The complexity of human biology necessitates a tool capable of isolating glucose’s unique attributes, which led researchers to explore Raman spectroscopy.

Raman Spectroscopy: A Potential Solution

Raman spectroscopy involves shining a laser onto a sample and analyzing how a small fraction of that light changes after interacting with certain molecules. A significant portion of the light merely bounces back unchanged. Nonetheless, a minuscule amount interacts with molecules, causing them to vibrate and reflect light differently. This response creates a distinctive Raman spectrum for different molecules, including glucose.

In ideal conditions, a sensitive sensor can discern this weak Raman-shifted light from the stronger original laser light. A computer then compares the measured spectrum against known reference patterns to identify molecules, allowing for quantifying glucose levels based on the intensity of the distinctive peaks. While the concept appears straightforward, executing it has proven exceptionally challenging.

“Raman is elite among noninvasive optical techniques because of its specificity,” says Arianna Bresci, an optical engineer at MIT’s Laser Biomedical Research Center. “However, the downside is that the Raman signal is very weak—out of every million photons that enter, only one is a Raman photon.”

Advancing Towards a Noninvasive Glucose Monitor

In the quest to develop noninvasive glucose monitoring, the MIT team led by Jeon Woong Kang has made strides. Back in 2020, they demonstrated the ability to accurately measure glucose Raman signals directly from skin. This breakthrough stemmed from their ability to filter out unwanted noise by adjusting the angle of the near-infrared light used.

Although their initial device was quite large—comparable to a desktop printer—they have since reduced its size considerably. Recently, the team showcased a functioning device roughly the size of a shoebox and compared it against conventional glucose monitors.

The dream is to one day integrate such a device into something as compact as an Apple Watch or an Oura ring, but significant hurdles remain. Capturing weak Raman signals demands high-quality optical components. For glucose detection, the challenge intensifies because glucose signals in the skin are minute compared to surrounding substances. A smaller device typically collects less light, complicating the task of distinguishing glucose from background noise.

Ultimately, accuracy is non-negotiable when it comes to blood glucose readings. The standards set by current finger-prick methods must be matched or exceeded.

Looking ahead, the MIT team aims to refine their prototype further and conduct clinical trials to ensure it can compete with existing methods. They have partnered with a startup called Apollon to help bring the project closer to market viability.

“Our goal is to launch in 2029 or 2030, pending FDA clearance,” Kang mentions, emphasizing the pathway still ahead for noninvasive glucose monitoring technology.

The journey of noninvasive glucose monitoring will depend on whether researchers can condense an entire lab’s worth of sophisticated optics into a wearable format.