American healthcare is in trouble, but luckily AI is here to help. Here are 7 ways artificial intelligence will revolutionize healthcare.
- Electronic medical record assistance
- Diagnosis of health conditions
- Patient messaging triage and response
- Patient results triage and interpretation
- Cancer research and drug discovery
- Personalized medicine
- Patient monitoring
This post may contain affiliate links.
With the explosion of AI (artificial intelligence) technology in recent months, it feels like we’re entering a brave new world. But rather than a dystopian future where AI controls our lives through nefarious ways, I envision a future where AI can be harnessed for the betterment of humanity. When it comes to healthcare, I personally believe that the influence of AI is going to be both profound and overwhelmingly positive.
And it couldn’t come at a better time.
Healthcare needs a savior
The Covid-19 pandemic has taken a troubled industry and brought it to the brink. Even prior to the pandemic, we were were calling the problem of physician burnout an epidemic. And now a few years later, physician burnout is worse than ever. Layer on top of this a looming physician shortage and the Great Resignation, and you’ve got an industry that is in dire need of innovation to help our healthcare providers.
Everything I’ve noted here for physicians can also be said about nurses, pharmacists, and technicians, by the way. It’s imperative that we do all we can to decrease the burden on our healthcare workers. We are lucky that artificial intelligence is entering a golden age of innovation, ready to help our healthcare workforce.
Read more:
- The Epidemic of Physician Burnout
- Doctors are More Burnt Out. Is it Being Employed or Just Covid?
- The Physician Shortage: A Threat to the American Healthcare System
- Will Healthcare Survive The Great Resignation?
Artificial intelligence will help save healthcare
In its highest and best form, artificial intelligence augments the brilliance of mankind. It makes us smarter, faster, and more productive. For example, ChatGPT is a chatbot launched by OpenAI in late 2022. It’s a remarkable tool that was trained via deep learning on terabytes of data and fine tuned by researchers. The end product is a flexible tool that can generate remarkably natural responses to questions. (It even helped create the outline for this post.)
Now for the first time, the general public is starting to see the implications of AI in our society. As a urologic surgeon, blogger, and amateur futurist, I’d like to put forth the argument that AI has the power to save healthcare. Investors clearly agree with this potential, based on the billions of dollars they are investing into the sector.
When I think about all of the pain points in the day to day lives of physicians and nurses, there are so many ways that AI can help. Here are just seven areas where I think AI will revolutionize healthcare by increasing efficiency, diagnostic accuracy, and finding new treatments:
- Electronic medical record assistance
- Diagnosis of health conditions
- Patient messaging triage and response
- Patient results triage and interpretation
- Cancer research and drug discovery
- Personalized medicine
- Patient monitoring
(You can click the links above to go directly to the corresponding section, or just scroll down.)
Electronic medical record assistance
I can speak from personal experience in regards to electronic medical records (EMR). Since the mid 2000s when I started my medical career, I’ve been intimately familiar with the blessing and curse of electronic health records systems. The EMR was widely touted as a way to decrease human error and share data between different health systems. By creating a way for patient data to be preserved neatly in electronic form, it has certainly achieved these goals. But by combining the EMR with billing and patient messaging systems, it’s also become an incredible source of frustration for medical professionals.
Let’s take my own experience. In my urology practice in SoCal, I had about 15 minutes to spend with each patient. Considering a few minutes for record review and walking to and from my office to the exam room, I was lucky to have five minutes to speak to each patient. The rest of the time I needed to type their medical information into the EMR system and enter orders. By being efficient with my workflow, I generally only needed an hour after my clinic day to wrap up notes before I could go home. I know better than to complain about this, because I have some good friends in family medicine who routinely spent hours every evening at home finishing up their notes from the day.
Artificial intelligence to the rescue
There are early signs that AI will soon be able to help with the pain of the EMR. Voice to text transcription technology has vastly improved in recent years, with the quality of speech recognition software improving every year. But these systems still require a physician to sit down in front of the computer and dictate into a microphone, fix the inevitable transcription errors, and then format the text into a readable product. In my own experience, I found that this software was helpful in reducing repetitive injury to my wrists, but didn’t actually save me that much time.
The next generation of AI systems for note-taking is called “ambient clinical intelligence,” or ACI. This technology aims to completely essentially act as an automated medical scribe by listening in the background and transcribing normal conversation into organized notes. Deepscribe is an early breakout in this field and says it’s “The world’s most widely adopted AI-powered medical scribe.” Their demo video is impressive, and I look forward to this technology becoming more widespread.
Beyond medical note-taking, I can see the use of artificial intelligence in bringing efficiency to order entry. It takes a comical number of clicks to do simple things, like order a CT scan or discharge a patient from the hospital. By applying voice commands and prediction to order entry, clinical workflows can become much more efficient. Just like improving the process of note-taking, this will free up clinicians’ time and decrease burnout.
Read this excellent summary on LinkedIn for more information: AI in Medical Documentation
Diagnosis of health conditions
Another way that artificial intelligence will revolutionize healthcare is in the diagnosis of illness. The diagnosis of everything from lymphoma to a broken hip can be facilitated by AI. As summarized in this paper, radiology is a great example of how AI is already offering benefits to the healthcare sector. Via software like AIDoc, radiologists can already use AI to help diagnose serious conditions like pneumothorax, brain hemorrhage, and aortic dissection.
As an image analysis tool, AI will undoubtedly help speed up the diagnosis of various radiologic conditions, leading to decreased healthcare cost and improved patient satisfaction. While there will always be the concern that AI will replace the need for radiologists, I don’t think this is a true concern. There’s no replacement for the well trained human eye (and our legal system still needs a responsible party). So in clinical practice, there is always going to be a radiologist needed to double check the work of the AI software.
A second opinion is always welcome
I’ve spoken to many radiologists who are burnt out from the constant deluge of mammograms, chest x-rays, and CT scans. Healthcare technologies like AI that increase the probability of accurate diagnoses while saving time will offer a much needed lifeline to radiologists everywhere. It will also likely lead to earlier detection of dangerous conditions like cancer and stroke, leading to improved patient outcomes.
But the benefit of AI in diagnosis won’t just be limited to radiology. By operating in the background of medical records, AI will be able to diagnose conditions alongside physicians. By integrating patient complaints with diagnostic data like vitals, lab tests, and radiographic studies, machine learning AI systems can augment human intelligence by offering a number of probable diagnoses. Rather than replacing a physician’s diagnosis, I see artificial intelligence as acting as an educated second opinion from a fastidious colleague.
Depending on the physician’s confirmation of the likely diagnosis, the AI can then offer evidence based treatment options, speeding care.
Over time, I expect the medical field to benefit from fewer missed diagnoses, improved clinical decision support, and better health outcomes. Proper training and oversight of the AI software will be key to making this a reality.
Patient messaging triage and response
You might be surprised that I’m including the topic of patient messaging in a post about how artificial intelligence will “revolutionize” healthcare. Patient messaging sounds very boring. But patient messaging is one of the single biggest pain points for clinicians!
Modern health care is all about increased access of patients to their doctors and nurses. From the patient perspective, this is a good thing. It allows patients to discuss minor aspects of their medical issues via email, rather than having to come into the office. This is a great convenience factor.
The dark side of this convenience is a never-ending deluge of messages for doctors.
My experience with patient messages
In my 14 person urology group in SoCal, we’d routinely get hundreds of patient messages every single day (even weekends). It took an army of medical assistants and appointment clerks to sift through the messages. The majority were for non-urgent matters, like appointment requests, while some were for important matters like fevers or complications from surgery. Since there was only rudimentary automation to the message sorting, we were routinely far behind on our message handling despite our best efforts. This led to a lot of angry patients and occasionally worse health outcomes.
My colleagues would routinely cite the volume of patient messages as a major problem. Since our clinic day was quite busy with patient appointments, the only time to deal with these were after hours. So after we would finish our notes, we’d then have to spend more time to respond to patient messages. By that time, I’d often be so hungry and tired that I’d be typing responses while half slumped over in my office chair. And if some messages had urgent matters, I had to figure out how to deal with them after hours.
Researchers at Vanderbilt completed a recent study where they used AI to sort patient messages by complexity. Although their machine learning algorithm wasn’t ready for prime time, it was an important first step in creating a message triage system.
I envision a smart AI system that can efficiently triage patient messages and even respond appropriately to non-medical concerns like appointment requests. A system like this could provide incredible cost savings while at the same time improving the quality of life for both patients and physicians.
Patient results triage and interpretation
A related but different way that artificial intelligence can improve healthcare is with patient results. On top of the daily deluge of patient messages, physicians also have an avalanche of patient results: lab tests, radiology tests, and pathology results. Each of these (in my prior healthcare system) required a physician to review the result and take action.
The more tools we have to diagnose illness, the more results there are to interpret. And these days, we have so many tools that clinicians now are often overwhelmed with results during a patients’ workup. While many results come back normal, the abnormal results are the ones that require a lot of time and attention.
Each abnormal result requires complex decision making to decide if the patient should:
- repeat the test immediately
- repeat the test in a few months
- come in for a discussion
- have an intervention like a biopsy or new medication
Many times, I’d consult my colleagues to determine the best course of action or review literature to decide the best intervention for my patients. So while I could resolve many results in seconds, some results took five or ten minutes each to deal with. When I had many dozens of results coming in a day, this became quite a time consuming part of my workday.
While no AI system will be able to replace the value of collaboration, I can envision a powerful AI driven tool that does the following:
- categorizes results by acuity
- suggests the next step
- provides links to supporting information and research
- assists in the next step (like ordering follow up studies and notifying the patient)
This last function is key. Because results are not time consuming only because of the decisions they require. Enacting the next steps is quite time consuming. Simply to order a few labs and a CT scan might take a hundred mouse clicks in your average EMR.
A predictive AI software could save incredible amounts of time and improve the patient experience as a result.
Cancer research and drug discovery
Yet another interesting application of artificial intelligence in healthcare is in the arena of cancer research. From breast cancer to lung cancer, new and innovative treatments are often targeted very specifically to a patient’s specific type of cancer. This type of approach, called immunotherapy, is different from traditional chemotherapy in many ways.
Yet identifying new immune therapies for specific cancers can be very difficult. New AI models like the one powered by Deep Mind, a Google owned artificial intelligence company, have shown promise in predicting and modeling the complex ways that proteins fold and interact.
One day soon, AI-biotech companies will be able to offer cancer patients new hope by speeding the development of patient specific cancer drugs. The economic feasibility of this approach remains to be seen, but this is a problem for the pharmaceutical industry, insurance companies, and the government to argue about in the future.
Personalized medicine
In the same way that individual patients can have targeted cancer treatments powered by AI, I foresee a new age of personalized medicine in our future. Although this is an area fraught with concerns over medical ethics, it has long been the case that there are certain genetic profiles that predispose some people towards things like cancer or diabetes.
Asian Americans, for example, are 40% more likely to be diagnosed with diabetes than non-Hispanic whites. Why is this the case? Some of this is likely due to cultural and dietary differences, but certainly some of this phenomenon is also due to genetic differences that predominate in Asians.
In the near future, artificial intelligence will usher in a new age of personalized medicine. In this future, there will be new opportunities for the prevention and treatment of diseases based on AI’s analysis of your personalized information. These data sets will include your personal medical history and genetic information to create a bespoke treatment that is optimized for your care.
The first step in creating personalized medicine will be to gather more data about genetic differences in different ethnicities, as pointed out by others. After that step when the technological hurdles are worked out, the main hurdle to a new age of completely personalized medicine will be cost.
But the promise of personalized medicine is an AI driven model that suggests the treatment for your specific genetic makeup and real time expression. This personalized plan will have the best chance of offering you a healthy, disease free life.
Patient monitoring
The final way that I foresee that artificial intelligence will revolutionize healthcare is in patient monitoring. While at first glance this seems like an unexciting application of the power of artificial intelligence, that’s exactly the point.
In the hospital right now, there are trained professionals staring at monitors. From blood pressure monitors in the ICU to cardiac monitors on the telemetry floors, someone is being paid good money to sit and monitor patients for dangerous changes.
Artificial intelligence in healthcare offers the power to be another set of eyes on patients in the hospital, monitoring them carefully for signs of decompensation. Rather than replacing that cardiac nurse, AI can augment their ability to predict and react to dangerous changes in their patients.
But I think AI can offer even more than just monitoring. I think it will one day help predict which patients are going to do poorly.
My experience with patient monitoring
As a urologic surgeon, one of the most puzzling (and dangerous) situations I deal with are infected kidney stones. The combination of a febrile urinary tract infection with an obstructing kidney stone is dangerous and requires antibiotics, diversion of the urine with a stent, and close monitoring. But it’s very difficult to predict which patients will actually become septic, which is that dangerous combination of severe infection and low blood pressure that can be quickly fatal.
When I have these patients in the hospital, I will repeat labs such as white blood cell counts and lactates to track their progress either towards improvement or sepsis. I’ll also closely monitor their vitals with the help of my nursing colleagues. But I’m sometimes surprised at which patients seem to suddenly decompensate (and which patients do just fine).
Through deep learning and integration of real time vitals, medical history, and labs, I believe AI will one day offer better tools that will predict exactly which patients are at the highest risk to decompensate. This will improve lessen the burden on physicians and nurses and help target medical care to the patients that need it the most.
Conclusion
For many years, the benefits of artificial intelligence in healthcare have been “right around the corner.” Now more than ever, we are starting to see the first glimpses of the powerful ways that AI will revolutionize the delivery of healthcare in our country. It’s about time, because American healthcare is suffering.
From a rampant burnout epidemic to a healthcare worker shortage that continues to cripple our hospitals, we need all the help we can get. By improving efficiencies and offering solutions for both the diagnosis, treatment, and monitoring of our patients, I believe AI will help bring healthcare into a brave new world of improved care.
— The Darwinian Doctor
Do you agree, or am I being overly optimistic? Let me have it in the comments below, and make sure to subscribe to my free newsletter below.
Want to support the blog?
- Join our investor club at Cereus Real Estate
- Visit my Recommendations page
- Check out my wife’s food blog: Eat Dessert First
- Stay at our luxury short term rentals
- Check out my TikTok channel
- Follow me on Instagram
- Follow me on YouTube
- Contact me with questions
Perhaps you’re more of a Facebook type?
Are you a physician, spouse, or professional and you’re interested in using Real Estate to gain financial freedom? Join us in our Facebook group and accelerate your journey!
Great review on this subject, and I agree that AI could be a game-changer for us in medicine–especially by freeing us from the inefficiencies of the EHR that we have all become enslaved to.
The integration of our human touch and senses along with the power of these AI tools will make the future medical world better.
Absolutely agree! Hopefully there will be some innovation ahead!
Would you consider a non-clinical career at a medical AI company?
I think it could be very interesting!
You’ve masterfully captured the essence of AI’s transformative potential in healthcare, emphasizing its timely arrival during an industry crisis. Your post not only provides hope but also tangible solutions through artificial intelligence that could alleviate the immense pressure on healthcare professionals. Your firsthand insights into the struggles with EMR systems and the meticulous detailing of AI applications are enlightening. It’s particularly impressive how you balance your roles as a surgeon, blogger, and futurist, offering a well-rounded perspective on the future of medicine.
Thanks For Sharing Such beautiful information with us
Thanks for sharing such valuable information about AI in Healthcare. Keep it up