Latest / Common Sense Health / This Is Why You Accept False Health Information Online
Transcript
- Stephen Thomas BSc. Hons: Hello. Have you ever noticed that when you search for information on the Internet, you tend to simply believe the first answer you get, unless it's something you actually know a lot about and then you can spot some inconsistencies in the answers? I had a fascinating conversation today, by the way, not with my doctor, not with a metabolic health practitioner, and not even with a patient who has reversed their type 2 diabetes and now terrifies their GP by having normal blood sugar without medication. Now, I had a conversation with AI and I have to say it was very interesting because at first it did exactly what you'd expect. It came out with the standard institutional script. You know the one LDL cholesterol is terrifying. Statins are wonderful. Plant sterols are marvelous. Guidelines are sacred. And if your blood results don't fit neatly into the NHS, computer system, then obviously the problem is you, not the system. So I thought, right, let's take this digital parrot to metabolic school. And to be fair, once I started giving it actual clinical context, poor thing began to wobble a bit, which is amusing really, because we're told AI is the extraordinary intelligence, this magical brain of the future, this thing that's going to transform medicine and save the planet and run our lives and probably decide whether the fridge is being emotionally supported enough for you. But I'm increasingly convinced AI doesn't really stand for artificial intelligence. It stands for artificial information because it doesn't really know anything. It gathers, it predicts, it rearranges and repeats a bit like a junior doctor with Wi-Fi and no lunch break. the 47 milligram deciliter emergency that wasn't. So I gave the AI a scenario I see quite commonly. Someone following a ketogenic or carnivore way of eating checks their blood glucose and the meter says 47 milligrams per deciliter. Now, in a standard hospital ward, that number would trigger panic. The nurse would be reaching for the orange juice and the doctor would be looking very serious. Someone would probably appear with a sandwich and a dietitian somewhere in the building would sense a disturbance in the force because in a carbohydrate dependent person, yes, a glucose of 47 milligrams per deciliter could be very concerning. But here's the part that gets missed. Context. The person's ketones were five millimoles per liter, and this completely changes the picture. The brain doesn't have an absolute requirement for glucose. It has a requirement for energy substrate. And when someone is properly fat adapted, ketones become the major fuel source. They cross the blood brain barrier, they supply energy and suddenly that apparently terrifying glucose number looks very different. This person isn't necessarily collapsing into a hypoglycemic coma. They may be sitting there perfectly calm, mentally sharp, not hungry. and wondering why everyone else is eating every two hours like a toddler at a birthday party. The AI, to its credit, eventually understood this. It realized that standard reference ranges are often based on the average population. And the problem with that is obvious. If the average population is metabolically unhealthy, overfed, undermuscled, underslept, stressed, medicated, inflamed and eating cereal with a halo, then normal might not actually mean optimal. It may just mean broken. why LDL can raise when someone gets healthier? Then we moved on to LDL, the granddaddy of metabolic panic. Now, if someone cuts out bread, pasta, cereal, potatoes, biscuits, sugary snacks and all the usual beige carbohydrates, something interesting can happen. They lose weight, they have blood. pressure normalises, their triglycerides usually fall, their HDL often rises, their energy improves, their hunger reduces, their waist shrinks, their type 2 diabetes may even go into remission. And then one number on that blood panel goes up, LDL. And suddenly the entire medical system throws itself onto the floor like a Victorian widow. no, the LDL. Never mind the fact that the person has just reversed half the metabolic disaster that got them in trouble in the first place. Never mind the fact that their insulin has improved. Never mind the fact their triglycerides have dropped. Never mind the fact that their blood pressure has normalized. Never mind they can now walk up the stairs without sounding like a broken accordion. The LDL went up. So apparently we must all gather around the sacred calculator and begin the pharmaceutical ceremony. But here's the issue. LDL is not floating around the bloodstream looking for innocent arteries to mug. Fat is not water soluble. So if you become fat adapted, your body needs to transport fat based energy around the body, deliver packages, triglycerides into particles, sends them out. Tissues take up the energy and those particles then can remodel. This is physiology. That is the transport. That is the energy logistics. is not automatically disease. In a lean, metabolically healthy, fat adapted person with low triglycerides, good glycemic control, low-fasted insulin, good inflammatory markers, good blood pressure and perhaps a zero calcium score, the context is completely different. But standard medicine often doesn't do context. It does boxes. And if your LDL box is red, the computer says And once the computer says no, common sense is asked to leave the building. protocol versus the patient. This was probably the most human part of the conversation, because when we then talked about the hostility many patients face when their results improve in ways that doesn't match the expected script. I mean, I've seen count Countless times this happened. Someone changes their diet. They cut the refined carbs. They start eating proper food and the glucose normalizes. The blood pressure improves. They come off medication. They lose weight. They feel 10 years younger. And instead of the doctor saying, well done, this is fantastic. What exactly have you been doing? The response is sometimes you shouldn't have done that, which is a bit like telling someone off for escaping a burning building because they didn't use the official exit. The reality is many doctors are trapped. And I do have sympathy for that. They work under pressure, have protocols, time limits, targets, legal risks, prescribing pathways and guideline algorithms staring at them from every direction. Modern medicine does not reward a doctor for saying this patient in front of me is clearly metabolically healthier and I'm going to interpret their blood results in that context. It rewards a doctor for following the protocol. If they follow the protocol and the patient slowly deteriorates, they are protected. If they use clinical judgment and step outside the algorithm, even for good reasons, they can be exposed. So what happens? The rule book becomes more important than the patient. The chart becomes more important than the person and the reference range becomes more important than the story behind the result. That is a problem because a blood test is not a moral judgment. It's a snapshot and snapshots need context, especially when someone is eating away. that changes fuel metabolism completely. So interpreting bloods in the context of a ketogenic or carnival way of eating is absolutely crucial. Otherwise, you end up treating physiology as pathology. And that is how healthy people can get frightened back into being unwell. AI, data centers and the comedy of modern priorities. Now, here's the part that makes me laugh. As the conversation went on, the AI started asking me questions, lots of questions. What markers do you value most? How do you prepare patients for difficult GP appointments? What timelines do you see clinically? What changes do you observe first? And I thought, well, hang on. This thing supposedly has access to mountains of medical literature and endless studies and institutional guidelines, pharmaceutical data and probably several million PDFs that nobody's ever read, by the way. And yet it still wanted the real world experience. It wanted the ground truth. It wanted the messy human clinical reality. And there's something very funny about that because We are building enormous data centers all over the place, huge warehouses of servers, massive energy demands, water consumption through the roof. Apparently we all need to save the planet, turn the heating down, eat bugs, stop using plastic straws and feel personally guilty for owning a kettle. But if a giant tech company wants to build a data center the size of a small principality so an AI can generate a picture of a cat wearing a waistcoat. Well. suddenly the environmental concern becomes very nuanced, doesn't it? Very complicated. Lots of stakeholders, very important innovation. after all that electricity, all that water, all that computing power, all those servers humming away like a mechanical beehive, the AI still has to come back and ask a human being, what are you actually seeing in real patients? And that should tell you something, because human beings are not just data processors. We notice patterns. We understand context. We see the person behind the chart. We remember the woman who came off blood pressure medication after 30 years on it. We remember the man who type 2 diabetes has been reversed. We remember the patient who was terrified of their LDL, but had never been told to look at insulin, triglycerides, inflammation, blood pressure, calcium score or the wider metabolic picture. AI has information, humans have meaning and medicine desperately needs more meaning. the bigger lesson. So what did I learn from talking to AI and taking it to metabolic school? I learned that when you feed it the standard information, it gives you the standard answer. No surprise there. But when you give it context, mean, real clinical experience and a framework based on physiology, it does start to shift. And that is the lesson for patients as well. Do not be bullied by isolated numbers. Do not be frightened by reference range without context. Do not assume that normal means healthy and do not assume that outside the range automatically means diseased. A glucose of 47 milligrams per deciliter means one thing in a carb dependent person, shaking, sweating and confused. It may mean something very different in a calm, fat adapted person with ketones at five millimoles per litre. An LDL of 250 means one thing in someone with high triglycerides, poor glucose control, high insulin, high blood pressure, inflammation, fatty liver and poor metabolic health. But it may mean something very different in a lean, insulin sensitive, fat adapted person with low triglycerides, strong HDL, good blood pressure, low inflammatory markers and even possibly a calcium score of zero. Context is not optional. Context is really the whole point. Closing call to action. So if you are following a low carb ketogenic or carnival way of eating and your blood results don't fit the standard mold, don't panic, just get educated, ask better questions, look at the whole picture. And if your doctor tries to treat you like a walking LDL particle with a troublesome personality, take a breath. You are allowed to ask. Well, what does this mean in the context of my overall metabolic health? What are my triglycerides? What is my fasting insulin? What is my blood pressure? What is my HbA1c? What are my inflammatory markers? What is my calcium score? What is the actual outcome we are trying to prevent? Because the goal is not to win an argument with your doctor. The goal is to understand your own health well enough that you're not frightened into bad decisions by a number on the screen. And if even AI with all its servers and data centers, water usage and planet saving corporate press releases still needs to ask humans what is actually happening in the real world. Well, maybe there's hope for common sense after all. Anyway, until next time, keep cutting the carbs, keep checking the context. Remember, AI may have artificial information, but you still have a brain. Use it before someone tries to prescribe you something for it. And with that, I hope you enjoyed it. If you did, please consider liking, sharing and subscribing. And if you didn't like it you got to the end, then please drop a comment because I'd like to know what you didn't like about it. We do keep in mind that artificial information is not the be or an end all. And that was the whole point of this really, that if you go on to Google or you use Gemini or perplexity or whatever, don't always accept the first answer. Just make sure that you interrogate it, try and think of other additional questions that you can ask, because the more you dig deep, the more surprised you'll be that the artificial information will try to listen to you and work out the actual answer to what your problem is. Anyway, thanks for watching.