Continuous Glucose Monitoring Outside the Traditional Areas, ATTD 2026 Conference Catch-Up

You know that feeling of mild panic when the battery icon on your phone suddenly turns red? Oh yeah, the immediate scramble for a charger. Right. Or like the absolute certainty you feel when your car's speedometer reads exactly 65. I mean, we are entirely wired to trust dashboards, you know? A number on a screen just inherently feels like the absolute undeniable truth. Exactly. But what happens when you take that deep seated desire for a quantifiable control? And you just strap a dashboard directly to your own biology. Welcome to your deep dive. I'm your host. And today our mission is to unpack a really fascinating stack of medical conference transcripts. And as the resident expert here, I'm excited to get into this. We are looking at the realities, the hype and well, the hidden science of continuous glucose monitors. Or CGMs, as you usually hear them called. Yeah, because if you spend like any time on social media, you have almost certainly seen the ads for these things. Oh, they are everywhere. They really are. And they're pitched so aggressively to healthy people, you know, people without diabetes, promising effortless weight loss, energy optimization, and this whole idea that controlling your blood sugars like the ultimate key to hacking your metabolism. Right. The whole biohacker dream. But the goal today is to cut straight through that noise for you. We're going to separate the intense marketing hype from the actual clinical reality, which we definitely need to do. Absolutely. So we'll look at how these devices perform when healthy people use them for wellness diets, how they act as medical sleuths for rare conditions, the physics of why certain hardware just totally fails. And finally, how they're completely upending the rigorous world of clinical trials. Okay, let's unpack this starting with the wellness hype. Because like we said, the whole glucose goddess trend is everywhere right now. It really is the most public facing use of CGMs today. You have healthy adults tracking their blood sugar 24 hours a day, seven days a week. Right. And the central premise they're sold on is glucose guided eating or like hunger training. The pitch is basically, if you keep your blood sugar perfectly flat, your body stops storing fat, your cravings magically disappear, and you just naturally lose weight. Yeah. Well, the clinical reality drawn directly from the data presented by leading researchers like Nicolas Gas is that there is virtually zero evidence that using a CGM leads to weight loss for a non diabetic. Wait, zero. Like none at all. Basically none. The biological data fundamentally contradicts that entire premise for healthy individuals. I mean, for people without diabetes, a temporary spike in glucose after a meal isn't, it's not the primary isolated driver of fat storage. Oh, wow. Right. Nor is it a causal risk factor for most of the chronic diseases they're trying so hard to prevent. If healthy adults want to improve their long-term outcomes, the clinical goal should actually be lowering LDL cholesterol and managing blood pressure. Which I'm assuming CGMs do not measure. Exactly. They do not measure either of those metrics. They give you this intense hyper focus on one single variable that just doesn't dictate your overall metabolic fate. But logically, I mean, shouldn't having real-time biofeedback make someone eat healthier? Like, if I see a massive red spike on my phone after I eat a donut, I'm probably going to put the second donut down, right? You would think so, yeah. But that assumption relies on the idea that the user will simply eat less overall. Which they don't do. No. What the clinical studies actually show is a substitution effect. When non-diabetic UCGMs, they absolutely tend to eat fewer carbohydrates, sure, because carbs are what cause that line on the screen to go up. Okay, that makes sense. But they don't just stop eating. They replace those mis-incarbitidrates with foods that are really heavy and fat and protein. So, that result is that their total daily calorie intake remains exactly the same. Oh, I see. Yeah. No overall calorie reduction means there is no weight loss. So, wait, let me make this straight. If I eat a banana, which is packed with potassium and great for my blood pressure, my CGM might show a really sharp glucose spike, right? Yeah, you're likely. Yes. And the dashboard implicitly flashes red and tells me, hey, you made a bad choice. But if I sit down and eat a stick of pure butter, which is catastrophic for my cardiovascular risk and my LDL cholesterol, the CGM shows a perfectly flat line. Yep, a flat green line. It essentially pats me on the back and says, great job. I mean, isn't this device essentially a broken compass for healthy eating? That is a perfect way to describe it. What's fascinating here is that short-term glucose feedback can actually actively encourage highly incoherent, even dangerous dietary choices. Because you're gamifying the exact wrong metrics. Exactly. Your butter and banana example is perfectly aligned with the clinical observations. You can consume a heavy, high fat, high calorie meal and receive positive biofeedback from the device just because your glucose from air is stable. It gives you a neurological dopamine hit for a behavior that actively harms your long-term health. That is wild. Are there hard numbers on this like studies testing this exact thing? Oh, definitely. Researchers point to the highly publicized DOE study. DOE is this massive, well-funded, personalized nutrition platform. Right. I've heard of them. Yeah, they ran a huge trial that made a major splash in medical journals. It featured a highly involved intervention. Participants wore CGMs. They logged their meals, used a dedicated app, and received this personalized algorithmic feedback on how to eat based on their unique glucose responses. So basically, a completely optimized, incredibly tech-driven diet. Exactly. Now, on the other side, the control group was just given a standard generic health leaflet about eating well. No apps, no monitors, just a piece of paper. Okay, so high tech versus a piece of paper, what happened? At the end of the intervention, the high tech CGM wearing group lost about 1.8 kilograms more than the group reading the leaflet. Wait, less than two kilos. Yep, less than two kilos. After sticking a needle in their arm, tracking every morsel of food and stressing over every single spike on a screen, the difference was barely negligible compared to just basic generalized advice. That's pretty shocking, right? It entirely fractures the prevailing myth of personalized nutrition through continuous glucose tracking. And the failure of that approach largely comes down to what biologists call intra-person variability. Intreperson variability, meaning like, my own body doesn't even react the same way to the same food twice. Precisely. It goes far beyond the food itself. Your glucose response to the exact same meal on Tuesday will look wildly different on Thursday based on all these hidden factors. Like what kind of factors? Well, how many hours did you sleep? Are you under a tight deadline at work and producing a bunch of stress hormones? Then there is the second meal effect. What's that? It's this biological quirk where your blood sugar response to lunch is heavily dictated by the macronutrients you consumed at breakfast. Your baseline literally shifts throughout the day. Oh, wow. I know idea. Yeah, and exercise plays a massive role, too. If you complete a brutal high-intensity spin class and need a banana right after, your liver is releasing stored glycogen to help your muscles recover. That results in a huge glucose spike on your monitor. But if I don't know the biology behind that, I look at the app and I just think, oh no, bananas are suddenly toxic to me. Exactly. But it's really just your liver doing exactly what it's supposed to do after workout. You misinterpret a totally healthy biological recovery phase as a dietary failure. Man. So if this tech is essentially a broken, anxiety-inducing compass for the average healthy person, it really begs the question, who is it actually built for outside of standard diabetes management? Like, where does this technology actually shine? Well, this brings us to the realm of rare medical mysteries, specifically a condition called non-diabetic hypoglycemia. Okay, what is that exactly? These are individuals who do not have diabetes, but their blood sugar just randomly and dangerously crashes. And for decades, diagnosing this was an absolute nightmare for clinicians. Really? Why was it so hard? Because the traditional gold standard for diagnosis was a grueling process. You had to catch the low blood sugar event exactly as it was happening using venous blood tests. Which means you obviously can't just do that in a doctor's office during a quick 15-minute checkup? No, not at all. You had to admit the patient to a hospital, confine them to a bed, and subject them to a medically supervised 72-hour fast. Wait, they starved them for three days? Up to three days, yeah. Just waiting for their blood sugar to crash so they could draw blood at the precise moment that they became symptomatic. That sounds miserable and wildly inefficient. So in this scenario, the CGM essentially acts like a stakeout camera. That's a great analogy. Right. Because instead of walking someone in a hospital room for 72 hours, hoping a biological crime happens, you just stick a monitor on their arm, send them back out into their normal life, and record the data in the wild to see exactly when and how the drops occur. It's absolutely revolutionary for these specific patients. The case studies presented by researchers like Robert Andrews really highlight how life altering this is. Take a 62-year-old patient named John. Hi. He was waking up repeatedly with intense nocturnal sweating and heart palpitations, and he had zero history of diabetes. Which most people would probably write off as, I don't know, severe anxiety or maybe a cardiac issue, right? Exactly. But clinicians put a CGM on him and set him home. And the monitor captured severe, repeating morning lows. Armed with the exact timing from that data, doctors knew precisely when to bring him in for targeted blood work. And what did they find? They discovered an insulinoma, which is a very rare tumor in the pancreas that overproduces insulin. They removed the tumor surgically and he was entirely cured. Wow. That is incredible. And it isn't just older patients. The research also highlighted a 33-year-old marathon runner named Matt. He suddenly started passing out at his desk at work, and he was suffering extreme fatigue during his normal runs. "Well, let me guess the CGM caught the same thing." Yep. The CGM caught the exact same pattern of morning hypoglycemia, leading straight to the discovery of another insulinoma. "That's amazing." But perhaps the most striking case is a 48-year-old woman named Mary. She underwent bariatric gastric bypass surgery, which obviously altered her digestive anatomy. Following the surgery, she began experiencing profound blood sugar crashes. "How bad were they?" It became so debilitating that she was suffering 46 severe hypoglycemic episodes a week. "Wait, 46 crashes a week, you can't function like that, you can't drive, you definitely can't work." She actually did lose her job because of it, it was devastating. But by using a CGM, her medical team could monitor how her body reacted to different interventions in real-time. So they basically ran experiments on her diet? Essentially, yes, and they eventually figured out a way to feed her directly through her remnant stomach, which completely bypassed the need for a highly risky reversal of her original bypass surgery. The CGM gave them the precise data needed to engineer a physical work around. "And did it work?" It did. She was able to return to work and completely reclaim her life. "See, that is where the technology feels like absolute magic." It does. But the clinicians in the research were also very careful to temper expectations here because technically medical guidelines still do not officially approve CGMs as a standalone diagnostic tool for these rare conditions. "Wait, why not? If it works so well for John and Mary?" The math behind why they aren't approved yet is pretty crucial to understand. It comes down to two statistical concepts. Sensitivity and specificity. "Okay, break that down for me." Well, CGMs have an incredibly high specificity at low ranges, about 96%. This means if your blood sugar is completely normal, the machine is 96% accurate at recognizing that normality and staying quiet. It rarely cries wolf when nothing is wrong. Okay, so high specificity means it's really great at ignoring normal behavior. Got it. But they have very low sensitivity at those extreme low ranges, around 40%. "Wait." Yeah, sensitivity is the machine's ability to actually detect a real event. If your blood sugar legitimately crashes, the CGM will simply fail to notice it or register it accurately 60% of the time. "Let me make sure I have this straight. It misses more than half of the actual crashes." Yes, it does. Because it's measuring interstitial fluid, which is the fluid between yourselves, not the blood directly. So it really struggles at the extreme bottom of the scale. Oh, that makes sense. Think of the CGM at these low ranges, like a highly specific but incredibly blurry security camera. If it manages to capture a clear picture of a burglar, you definitely have a burglar. But because the lens is so blurry, it completely misses 60% of the people who break in. Okay, so if it does beep and say you are low, how do you know if it's a real crash or just a, I don't know, a mathematical glitch? That is called the positive predictive value. And it relies entirely on prior probability. The clinical analogy often used is a pregnancy test. How so? Well, if a biological whale takes a pregnancy test and it reads positive, the test is mathematically flawed. Because the prior probability of that patient being pregnant is zero. The context renders the data totally useless. Right, obviously. So if a perfectly healthy 25-year-old with no symptoms gets an alert on their CGM saying their blood sugar is dangerously low, the prior probability of a rear tumor is near zero. Meaning it is almost certainly a false positive from a blurry camera. Oh, I see. But if Naria patient with altered gastric anatomy who is passing out daily shows a reading of 2.5 on her monitor, the prior probability is incredibly high. So the clinician uses that clue to act. Exactly. The machine doesn't make the diagnosis. The doctor makes the diagnosis using the machine's blurry clues combined with the patient's history. That makes a lot of sense. You know, having seen how transformative this data is when used correctly, the natural question just sort of leaps out to me, why do these devices still require a physical needle inserted into the arm? The hardware question. Yeah. Because we live in an age where a smartwatch can shine a light through my wrist and tell me my blood oxygen level or take an electric cardiogram. Why can't we just wear a watch to track this simple sugar molecule? You have just touched on the literal holy grail of medical hardware development, the non-invasive glucose monitor. And if you look at the industry history presented by experts like Loutineman, it is a masterclass in technological hubris. Because if you go online right now, you will see a flood of devices. I'm talking smartwatches and rings selling for 50 bucks, openly claiming they can measure your glucose non-invasively using light or electromagnetic waves. And the harsh, unvarnished truth from engineers who have spent their entire lives on this is staggering. After 28 years of rigorous research, millions upon millions of dollars in funding, countless clinical trials, and thousands of press releases promising that non-invasive technology is just one to four years away. Let me guess it's not here yet. There is still not a single non-invasive glucose monitor legally on the market in the US or Europe. The field is widely referred to as a graveyard of failed attempts. But why is glucose so elusive? Like we can track heart rate, we can track oxygen. And why does this one molecule defeat thousands of the smartest engineers on the planet? Think of it like trying to listen to a specific quiet whisper at a heavy metal concert. Okay. Heart rate and blood oxygen are loud drums. They have massive, strong optical signatures that are super easy for a sensor's light to bounce off and read. Glucose, on the other hand, is the whisper. So it's just a much weaker signal. Much weaker. When a smartwatch shines a light into your skin, that light scatters everywhere. It hits sweat. It reacts to minor temperature changes. It bounces off varying thicknesses of fat tissue and melanin. Isolating the specific optical signature of glucose through all that biological noise is mathematically excruciating. The signal just gets completely drowned out by the environment. Exactly. The interference is so bad that in tightly controlled laboratory settings, scientists have actually tested these non-invasive machines on phantom solutions. What are phantom solutions? They're liquids that contain absolutely zero glucose. And the machines still regularly generate false, fluctuating glucose readings, simply based on background noise and temperature shifts. They're essentially reading ghosts. Wow. Here's where it gets really interesting though. Because the optical physics aren't advancing fast enough to pass regulatory hurdles, the entire industry seems to be executing this massive pivot. Yeah, they really are. Because the FDA standards for diabetes management are incredibly strict. And they have to be a false low reading could cause a patient to withhold insulin and a false high could cause them to inject a fatal overdose. The accuracy has to be nearly flawless. Right. So instead of trying to solve the physics and pass those life or death medical hurdles, these hardware companies are just abandoning clinical accuracy entirely. Yeah. They are pivoting their marketing directly to the unregulated wellness space, which really loops us right back to our first segment. They are taking tech that they know reads ghosts and they're selling it to healthy biohackers who don't actually need it just to secure a return on their venture capital investment. It's a really fascinating and slightly disturbing economic reality. And it leads perfectly into the final major arena we need to explore today, the absolute minefield of clinical trials. Let's get into it. Because if non-invasive watches are a pipe dream for now, the medical community still has to rely on current needle-based sensors to test every new diabetes drug, diet, and therapeutic intervention. Right. And according to the research, just the act of wearing a sensor fundamentally skews the scientific data you were trying to collect. If we connect this to the bigger picture, CGMs have become the ultimate outcome measure in clinical trials. They are the judge and jury for whether a new pharmaceutical actually works. But researchers have to account for a massive psychological obstacle. The observer effect. Oh man. It's the difference between unblinded and blinded sensors. If you put a giant unblinded monitor on my arm that buzzes every time I eat a carbohydrate, it is literally like having my mother standing over my shoulder watching every bite I take. Exactly. Of course I'm going to alter my behavior. I'm going to subconsciously eat better because the machine is judging me. Psychological burden of being observed actually changes your biology. Clinical data shows that it takes a minimum of two to four full weeks of wearing a sensor for a patient's novelty to wear off. Two to four weeks. Yeah, it takes a whole month for them to stop trying to beat the game and revert back to their normal baseline dietary habits. Which means if you run a trial with an unblinded sensor, your first month of data is essentially garbage. You aren't measuring the drug. You're measuring the patient's reaction to the dashboard. Precisely why blinded CGMs are the gold standard for clinical trials today. These are devices that record the continuous blood sugar data but they keep the screen entirely blank. So the patient can't see anything? The patient cannot see their numbers at all. It is the only reliable way to gather pure, untainted data on how a drug is actually performing in the body. Okay, so once they have that blinded data, how much of it do they actually need to prove a new drug is safe and effective? It depends entirely on the specific endpoint the trial is measuring. If researchers want a solid read on time and range, which is the overall percentage of the day, a patient's blood sugar stays in a healthy middle zone, they need 14 days of continuous data with at least a 70% capture rate. 14 days seems reasonable. It is. But if the goal of the drug is to prevent rare episodic events like sudden hypoglycemic crashes, 14 days isn't nearly enough. You need a full four weeks of pristine data to accurately track those random events. Wait, this brings up something genuinely confused me in the trial guidelines. What actually defines a low blood sugar event? Is an event triggered when the math on the sensor drops below a certain threshold? Or is it an event when the human being actually feels dizzy, starts sweating, and experiences symptoms? You know, this raises an incredibly important question, and it is arguably the single most heated debate in the field of continuous monitoring right now. The definitions are profoundly messy. Really? Why? Because the overlap between sensor detected hypoglycemia, which is purely mathematical, and patient-reported symptomatic hypoglycemia, how the human actually feels, is shockingly small. How small of an overlap are we talking about here? One major clinical study, and astounding 65% of the low readings detected by the sensors, were accompanied by absolutely zero physical symptoms for the patient. You are kidding! So the machine is internally screaming that there is a severe medical emergency, and the patient is just like sitting on the couch reading a book, feeling completely normal. Exactly. The mathematical threshold was breached, but the biological reality was completely fine. And it gets exponentially more complicated when pharmaceutical companies write the rules for their own trials. Oh, I bet. Yeah, like, how long does the math have to stay low to count as a real event? Is it dropped to three millimoles per liter for five minutes an official event? Or does the line need to stay down there for 30 minutes to count? Because if a drug company gets to define what an event is, they could theoretically tweak the timeline to make their new drug look better. Bingo. Depending on whether they define an event as five minutes or 30 minutes, they can radically alter the published safety profile of their drug. They can make a medication look incredibly safe and effective, or highly dangerous, simply by shifting the arbitrary definition of a few minutes of mathematical data. That is wild. The raw data might be objective, but the interpretation is highly subjective. It really highlights how messy and human the supposedly clean world of hard medical data actually is. Wow. We've covered incredible ground today. Let's pull all these threads together for everyone listening. We've learned that continuous glucose monitors are a modern medical miracle when they act as blinded stakeout cameras for rare diseases like insolentomas. They are an absolutely vital, if mathematically and psychologically complex, tool for running the clinical trials that bring us life-saving drugs. Definitely. But for the average healthy person looking to optimize their wellness or drop a few pounds, they are mostly just an expensive anxiety-inducing distraction. They are a broken compass that rewards eating butter over bananas. I think the core takeaway for you listening should be this. Knowledge is only valuable when it is deeply understood and applied in the correct context. We live in an era of absolute information overload. Just because we possess the technology to measure a single biometric variable 24 hours a day, seven days a week does not automatically mean we should. Especially if we lack the specific clinical framework needed to understand what those blurry, fluctuating numbers actually represent. Context, as always, is everything. So what does this all mean? I want to leave you with one final slightly philosophical thought to chew on today. We talked heavily about the observer effect in clinical trials. We know that simply seeing our own biometric data instantly changes our behavior. The unblinded dashboard alters our biological reality. If the mere act of observing our own bodies fundamentally alters how those bodies operate, is it ever truly possible to get an objective baseline of how we naturally live? Every dashboard we look at, every screen we trust for an absolute truth might actually be changing a very nature of the thing it is trying to measure. It really is the quantum observer effect. Take it out of the physics lab and apply directly to human biology. It's a brilliant paradox to keep in mind the next time you glance down at a piece of wearable tech. Absolutely. Keep questioning the beta, beware the illusion of the dashboard, and we will catch you on the next deep dive.