From e-visits to e-billing, technology is changing the face of healthcare. Automating everything from providing services, prescribing medications, and filing and paying insurance claims may seem like a time saver. But who does it really serve? Did it do the job it was supposed to do?
As interest in ChatGPT and other language modeling programs surges, many AI evangelists are touting their use as healthcare diagnostic tools, even broadly claiming that it is the future of medicine. As with many high-tech innovations, such claims tend to obscure the small but critical applications we immediately grasp.
personal story
A company for which I worked as a programmer won a contract to automate medical procedures at a large US hospital. When the project went live, we received an error report from a very angry doctor complaining that the system would not allow him to prescribe a sufficient dose of a particular drug. After our investigation, we found that his script contained a few milligrams, while the safe maximum enforced by our software was 20 micrograms, or about 100 times less. It turned out that he had been doing this for decades, and no one had ever had the courage to correct him. The nurses just silently corrected his erroneous paper prescriptions. Sounds good, but e-prescribing has uncovered potentially dangerous practices.
on the other hand
When you add artificial intelligence into the mix, you can face some complex challenges.
Just like humans make mistakes, software also makes mistakes. There are many reports of electronic patient record (EPR) systems “forgetting” to send important patient notifications, obscuring critical treatment information, and causing harm and even death to the people we are trying to care for. The problem, as UK health leader Joe McDonald describes it, is that with paper “when you make mistakes, you make them one mistake at a time”, but EPR “unfortunately gives you the chance Make the same mistake thousands of times.
Dr. Chat GPT
ChatGPT recently passed the U.S. medical licensing exam, but using it for real-world medical diagnosis could quickly prove fatal. As Dr. Josh Tamayo-Sarver, who practices clinical practice in a local community emergency department, found out, ChatGPT works quite well as a diagnostic assistant, but only if you feed it perfect information, and the actual patient has a classic case of rarely There is such a situation. By providing 35-40 patients with detailed medical histories and detailed medical narratives of the symptoms that brought them to the emergency room, he found that for about half of the patients, ChatGPT suggested six possible diagnoses, as well as the “correct” diagnosis, or At least that’s what he believed to be a diagnosis – believed to be correct after completing evaluation and testing – and was one of six suggested by ChatGPT.
Doesn’t sound that bad. But a 50% success rate in an emergency room setting isn’t good either. His experiments illustrate how most medical encounters are about finding the right patient narrative. For example, if someone comes into the emergency room saying they have a broken wrist, but it’s not due to any recent accident, it could be a psychosomatic reaction to the patient’s grandson’s fall, or it could be due to a sexually transmitted disease, or something else entirely. The art of medicine is extracting all the necessary information needed to create the right narrative. It takes one person.
Private issues
Another consideration is patient privacy. Currently, anything entered into ChatGPT is basically published online. To be HIPAA compliant, any factor that establishes a link to a specific patient must be removed or changed.
personalized care
Personalized care refers to the development of a customized treatment plan for individual patients. In practice, this is increasingly done through technology and data. The most advanced application is genomics, where artificial intelligence is used to analyze a patient’s DNA to diagnose and treat disease and create personalized medicines for specific populations, down to the molecular level (sometimes called precision medicine). Many researchers believe that a personalized approach to health care can lead to better patient outcomes and more efficient use of medical resources, and will play an increasingly important role in addressing future health care challenges.
IoT-driven virtual hospital and telemedicine 2.0
This trend includes telemedicine and wearable devices connected to the Internet of Things (IoT) global network. By using connected devices to remotely monitor patients and provide communication channels for healthcare professionals, more care can be delivered remotely. We call it “telemedicine 2.0” because it goes beyond simple telecare (such as teleconsultation) to a holistic approach to remote patient care and treatment. Virtual hospital wards are an example of this trend in 2024 – where a central location will act as a hub for monitoring multiple patients at home.
Will artificial intelligence serve health care or kill it?
The answer seems to be both. Technology may seem like a panacea—and in some cases it is—but it would be foolish to assume that computers are infallible. For now, it’s wise to keep humans informed and consider simple, low-tech strategies first. Maybe we shouldn’t rush to centralized robotic logging solutions and stick with some manual offline methods for now. After all, that’s how we handle elections, and the worst outcome of a voting machine error is a stupid outcome—we can check that out with a good old newspaper.
About the Author
douglas squirrel He has been engaged in programming for 40 years and has led software teams for 20 years. He harnesses the power of conversation to create massive productivity gains for technology organizations of all sizes. His experience includes serving as CTO, growing software teams at startups in fintech, biotech, music and everything in between; working for over 200 organizations across the UK, US, Australia, Africa and Europe Provides product improvement consulting; and coaches a diverse range of leaders to improve their conversations, align business goals, and create productive conflict. He lives in Frogholt, England, in a wooden cottage built in 1450. . Learn more at douglassquirrel and squirrelsquadron.
