536
Artificial Intelligence? Can we really build minds from silicon and code? Well, yes. Machine learning has revolutionized many industries — computer programming, transportation, finance, engineering, and even creative pursuits like interior design and graphic design.
ChatGPT— an advanced AI chatbot made available to the public in late 2022 — passed the U.S. Medical Licensing Exam. AI will not just help doctors; it will aid doctors become doctors, revolutionizing education.
With the World Health Organization estimating a shortage of 10 million healthcare workers by 2030, AI is poised to transform the industry at a time when we need it most. AI could save the U.S. healthcare industry up to $360 billion annually and help finally achieve the goal of universal healthcare coverage within the next five years.
Why the healthcare scarcity? Many modern problems are the result of prosperity. More people are living longer, which, when combined with lifestyle choices, results in the increased prevalence of chronic disease. Thus, more patients are emerging just as the bulk of healthcare providers are entering retirement age – the perfect ingredients for a healthcare shortage. Of course, the pandemic did little to help, leading to widespread burnout in the medical field.
Digital Options
Though the pandemic assisted with solidifying the tele-health industry, spurring services like Doctor on Demand for patients, digital interfaces are now being launched to help triage patients, too. According to a 2024 report by the World Economic Forum: “People who are generally healthy can use self-monitoring devices to optimize their mental and physical health, while those with health issues have access to a wide range of digital solutions.” This could potentially decrease readmission rates by 30% and greatly alleviate the workload of healthcare providers. Using ChatGPT at home to interpret complex MyChart test results also helps patients prepare meaningful questions ahead of their visit.
On the provider side of things, chat-based interfaces like ChatRWD (Chat Real-World Data) are now available for healthcare professionals, demonstrating a high accuracy rate in answering clinical questions. This type of Retrieval-Augmented Generation (RAG) incorporates external context beyond the given prompt — like insurance claims, lab results, wearable device data, doctor’s notes and even patient reported outcomes. Current treatment protocols and real-world clinical registries are also referenced, leading to a return of useful diagnostic answers to more than 58% of questions (compared with 2%-10% for the Language Learning Models, or LLMs, which perform based on prior training).
AI in the Doctor’s Office
Even during day-to-day office visits, doctors use AI for notetaking purposes, asking if patients are okay with having AI open during their chat. This allows them to focus on the patient instead of a screen, and patients are reporting higher satisfaction with the experience. These face-to-face interactions also aid retention, and doctors are more likely to engage with the patient in meaningful ways. Using AI for note taking during surgeries is providing doctors with a higher level of detail than ever before. With AI recording every step of the process in an organized way and ensuring all procedures are followed, physicians have more assurance that nothing is overlooked.
Another highly valuable contribution made by AI tools is in medical imaging. Algorithms trained on vast datasets can detect anomalies like tumors, fractures or internal bleeding, often more quickly and accurately than human radiologists. These tools analyze X-rays, MRIs and CT scans with remarkable speed and precision.
Machine learning algorithms and deep neural networks are also providing AI-powered EKG/ECG interpretation that is as accurate as that of an expert cardiologist with 30 years of experience. AI can predict arrhythmias, heart defects, sudden cardiac death, stroke, and other cardiovascular abnormalities faster and more accurately than traditional methods, potentially leading to quicker diagnoses and treatments. Of course, these tools should only be used to augment physician diagnoses, as patients are hesitant to accept purely autonomous diagnoses.
Trusting AI?
A recent study in the UK found that “just 29% of people would trust AI to provide basic health advice.” But overworked healthcare professionals are becoming increasingly reliant on AI to tackle everyday tasks, so regulation of these tools remains vital. At the end of the day, the legal responsibility falls on the clinician, not the robot.
Despite skepticism, there is no doubt that AI technology is saving lives. Everyday wearable devices like Fitbits are seeing major AI upgrades. There are countless stories of people sitting at home relaxing after a long walk with the dog, a moderate hike, maybe even just sleeping soundly in their bed at night – and suddenly their Fitbit alarm goes off. They view a heart rate reading of 180 bpm. Feeling completely fine, they suspect a Fitbit error but begrudgingly head to the hospital — just to be safe. Come to find out it was rapid atrial fibrillation — and their wearable device probably just prevented them from having a stroke, or worse.
When time is of the essence, AI tools can help guide life or death decisions, like whether a patient needs an ambulance or even emergency surgery. By objectively monitoring factors such as mobility, pulse, blood oxygen levels and chest pain – AI has been correct more than 80% of the time when predicting whether a patient needed to be transferred to a hospital by ambulance.
According to an AI model developed by researchers at Imperial College London, a new AI software is “twice as accurate” as professionals at examining the brain scans of stroke patients. The software is also able to identify the timeframe within which the stroke happened – which can help doctors decide if a patient is eligible for surgical treatment.
Robot Surgery
Deciding whether to do surgery is one thing, and AI-assisted machines are helping surgeons perform certain surgeries. Surgery outcomes are better if patients spend less time on the table under anesthetic, so a robot that can perform surgery faster is highly desirable. Microsurgery, hair restoration and cardiac surgery are examples where human-guided machines help doctors perform surgeries more accurately and efficiently.
AI can perform simple tasks through the robot, such as closing a port site and tying a suture or a knot. AI can even make real-time recommendations. During a colonoscopy, for example, AI software will be able to identify a potential polyp. Predictive AI can see 20-30 seconds in the future and give warnings, such as, “Hey, you are about to cut the common bile duct. Do you really want to do that?”
We may not be able to cheat death, but AI is a promising cheat-code that can possibly give us a few extra lives!
Text by Emily Alberts
Freelance writer Emily K. Alberts has been using em dashes since before ChatGPT was even born, thank you very much!
