Good morning. Interesting engineering has a new story which they got from a Chinese state paper. Researchers at Xinghua University in Beijing are building a training hospital that would be completely powered by artificial intelligence, AI. The headline is here, the AI doctors can treat 3,000 patients a day. Their AI can help 10,000 patients in a few days, which would take human doctors two years.
We’ll have two questions for the stories we do it. First question is, why is this the world’s first? There is a global race for AI, especially in medicine. So why wasn’t the first AI hospital built in Silicon Valley or in New York? The second question is, why did China build one? China has plenty of doctors and nurses. There is an imbalance between urban hospital systems here and the rural ones. There’s more medical services and higher quality in the big cities and towns in China compared to the countryside. So it’s possible they’re building to redress some of that, but maybe they have another objective also.
Artificial intelligence is going to revolutionize medicine. It will allow care to be personalized and targeted to individual patient needs. It’s already being used in Chinese labs to develop drugs. Virtual health assistance can be paired with robotics to do everything from administering drugs to helping patients exercise after major surgeries. There’s a huge broad spectrum of applications there. Medical imaging and diagnostics may be the biggest of all. Immediately, when AI can discover cancerous tumors and diagnose problems that show up on X rays better than humans can, and we’ll find out that we don’t need to have oncologists and radiologists and nuclear medicine people being paid half a million dollars a year. The patient just needs to get a quick scan. Let the AI power diagnostics run in, 30 minutes later, it’s done. That’s what’s coming. And that’s the answer to the first question, why it’s not being built in the United States. The American Medical Association and 50 state medical licensing boards, we’ll make sure that this technology is never, ever used in the United States. It just won’t happen that US patients can go around expensive doctors and hospitals and drug companies to get the care that we need. So come back to China then. The hospital for now is a training hospital for doctors. They’re going to be using AI tools to build a system of machine learning in the virtual world. And then in the real world, the large language model system is going to be conversational and learn as it goes and learn as it grows. And it’s already learning fast. The AI doctors are already at 93% accuracy on respiratory disease, for example, and that is already far higher than the accuracy in real world primary care settings.
In this study, researchers looked at COPD, chronic inflammatory lung disease. It’s a group of diseases which includes bronchitis and Zema. Researchers found that 13% of diagnoses are over diagnosed, more serious than actual, and 59% under diagnosed. Overall, the effective rate was just 78.2%. 14.5% of the tests were outside clinically acceptable bounds of accuracy over a large range of error on everything else.
In this study from Brazil, researchers looked at diagnoses for asthma, and they found the same problems there, lots of errors that lead to bad diagnosis. Only 59% of participants in the study got the correct diagnosis. Some of them had asthma, but they were told that they didn’t. 34% didn’t really have asthma, but they were told that they did. So these are good examples of what we hope AI can do for us in medical diagnostics, reducing false negatives and false positives. A false negative means that a patient is told that he doesn’t have it, but he does. A false positive means a patient is told he does have a disease, but in reality he doesn’t. And the Chinese data on the AI for respiratory disease was the only one cited in the article, the only one they shared publicly.
It’s hoped that this hospital will operate across the spectrum of healthcare. The intelligent agents can simulate diagnosis, treatment and follow up first on virtual patients and soon enough on live patients. So what else could these researchers here in China be hoping to gain? What is the objective of Xinghua University building an AI town where artificial intelligence basically runs the show? If you believe that the future of medicine is in personalized care, that the future involves fast, accurate diagnosis. And with that diagnosis, a company can fabricate a pharmaceutical compound and ship it to you overnight anywhere in the world.
In a nuts and bolts way, a patient can go to a clinic in his neighborhood or he can call a technician to come to his home or office, not a doctor, not somebody who’s gone to medical school, just a technician with a piece of equipment. The technician comes and takes a blood test and some DNA and then uploads the data to one of China’s AI hospital towns. Within two hours, the AI hospital has diagnosed the patient’s problem. Prescribe a compound of pharmaceuticals. They can print off the instructions and the patient’s native language and put it into an overnight bag for more comprehensive treatments. The patient could be invited to go to Hainan, which is a giant free trade zone here in China for medical tourism.
Now everything I’ve just described in that world is ready right now except for the testing of the AI on large populations of patients, simultaneously testing the technology at industrial scale, 1 million patients at a time, starting small but scaling up very fast. So now there are 10,000 patients in three days. Scale that up there at Xinghua and then build AI towns and other places, then they’ll be doing a million soon enough. And Chinese kids graduating Chinese medical schools, they won’t be graduating to help people merely here in China. They’ll be taking over medical care across the entire world, across entire medical fields. These grads at Xinghua might be for respiratory illnesses in South America. Another group will handle cancer in Eastern Europe and so on. That’s what the future of medicine is gonna look like if they can get this AI to work.
This is how we need to assess news stories like this when we see them. We shouldn’t just read the headline and look at the photo in the article and say the Chinese are using robots to listen to my heartbeat. No, they are taking over the world’s medical systems.
China already makes most of the drugs that we use. China already builds most of the diagnostic equipment that we use in medicine. And once Chinese companies are direct to patients, all of the middlemen in our markets, they go away. Multi billion dollar pharmaceutical companies basically are marketing companies for Chinese made medicines. They will go away. Lots of doctors, especially in oncology and radiology, respiratory illness and internal medicine, they will go away, too. That is why this is a no go in the United States, but everywhere else. This is the world we’re gonna see. This is Wang Shan, Yellow Mountain. It’s in Anhui province. Bigger.