Welcome to TechTalks’ AI book reviews, a series of posts that explore the latest literature on AI. This post is the first part of a two-part interview with Dr. Eric Topol about the impact of artificial intelligence on health care and medicine.
Since the first human civilizations took shape, doctors have been an ever-present member of human communities, curing diseases and caring for the ill. With advances in science and technology, their methods have evolved from praying to mystic gods to mixing herbal concoctions to using scientific methods and advanced medical equipment to diagnose, treat and prevent diseases.
Today, doctors can easily save their patients from diseases that killed millions of people in past centuries. But there is still room for so much more improvement.
The next revolution in health care and medicine might be tied to advances in deep learning, the branch of artificial intelligence that has become very popular in the past decade. Deep learning has already made several remarkable achievements in medicine, and there’s a lot of excitement about how artificial intelligence can change health care and there are many articles that explore how deep learning algorithms can help diagnose and treat complicated diseases.
But something that is less discussed is how learning might also change the way we interact with our doctors.
In his latest book Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again, Dr. Eric Topol, physician, scientist and researcher, sheds light on how AI can solve some of the biggest challenges in medicine and health care. This includes the challenges that the doctor-patient relationship is faced with.
In an interview with TechTalks, Dr. Topol discussed some of the promises of deep learning in improving human interactions between doctors and patients, as well as some of the key challenges that lie ahead.
Doctors have become data clerks
We often think of medicine in terms of activities such as reading vital signs, looking for complicated patterns in medical scans, doing surgery on patients and writing drug prescriptions. But in medicine, the human interaction between doctor and patient is just as important as all the scientific endeavors that take place.
Often, the warm and comforting tone of a doctor can have as much an effect on patient as the treatment itself.
Unfortunately, in today’s health care systems, the doctor-patient relationship has degraded a lot. Doctors spend less and less time communicating with their patients and more time on doing other things.
Dr. Topol expands on this point early in his book. “What’s wrong in healthcare today is that it’s missing care. That is, we generally, as doctors, don’t get to really care for patients enough. And patients don’t feel they are cared for,” he writes.
“Today, we’re seeing the erosion of the patient-doctor relationship, where we have burnout among physicians that’s the highest recorded in profession history,” Dr. Topol told TechTalks in a phone call. “We have depression and suicide, also at the highest recorded levels among clinicians.”
Ironically, part of the problem stems from technological advances. Compared with, say, a century ago, medicine has changed a lot and has become a digitized and data-based domain. There are many more ways to collect and analyze data, and much of the interactions that used to take place between doctors and patients are now replaced with data collection and examination tasks.
But those tasks still require a lot of human efforts to collect and analyze the data, which has all fallen on the shoulder of doctors. Doctors must spend a great deal of their time entering data in databases and stare at monitors and less time interacting with patients. A study published in the Annals of Internal Medicine found that on average, physicians only spend 49 percent of their of their time on filling electronic health records (EHR) and doing desk work, and only 27 percent of their total time goes to direct clinical face time with patients.
“The reason we have burnout that is so profound is because doctors are data clerks and have lost their morale,” Dr. Topol says. Dr. Topol warns that the diminishing spirits of doctors also leads to medical errors.
Another study by the University of California, San Francisco (UCSF), found that 82 percent of data entered in EHRs are copy-pasted or imported, and only 18 percent of the information is entered manually. This can cause clinical errors and lead to harmful treatment decisions.
AI can give back the gift of time
Fortunately, this is an area where AI shows a lot of promise. Artificial neural networks, the core technology underlying deep learning algorithms, is very good at finding relevant patterns and correlations in messy, unstructured data such as images, audio and text. That’s why it has made great inroads in fields such as computer vision, speech recognition and natural language processing.
In the field of medicine, AI algorithms can automate some of the tasks that previously required a lot of human labor. For instance, AI algorithms can relieve doctors of the pain of taking notes while visiting patients. There are already interesting efforts in the field, including projects by Microsoft and Google. Machine learning algorithms can extract meaningful information from the encounter between the doctor and patient and register them in the patient’s health record.
“AI-derived notes from natural language processing and machine learning are quite remarkable. And that’s already getting underway in places like the UK and China and pilot studies in the U.S. And that’s where this is headed,” Dr. Topol says.
There are many other fields where AI can improve the speed and accuracy of tasks doctors perform, such as analyzing medical scans and finding relevant information in medical records. Collectively, these technologies can free up much of the time doctors spend away from their patients.
“AI could be the greatest way to improve and restore the patient-doctor relationship, because of the gift of time,” Dr. Topol says. “And that is a product of many different things that AI can bring us. It includes elimination of keyboards, being able to process all the data of a patient to make lives easier for doctors and clinicians, to be able to do a lot of their pattern recognitions like scans and slides and other things that are very much used day to day by doctors, and to do it more accurately.”
In chapter 3 of Deep Medicine, Dr. Topol delves into many of the areas where deep learning algorithms are helping automate diagnosis tasks, including brain, heart and eye diseases.
Things can still go the wrong way
But Dr. Topol also warns that the increased speed and efficiency offered by AI can steer things in the wrong direction. “All this productivity, efficiency, workflow improvement, accuracy and speed could be used against doctors and patients, because it could lead to administrators and managers demanding more patients to be seen in any unit time, or more scans be read and more slides, and on and on,” he says.
How can we prevent that from happening? The industry will have to make its priority to employ advances in AI to restore the doctor-patient relationship. “It won’t happen by default. It’s going to require a lot of activism. It’s going to take intense effort to stop with the big business of medicine and start to get into the human connection,” Dr. Topol says.
This might sound like an odd suggestion, given that the general perception on AI is that it is here to replace humans and automate their skills. Some scientist have gone as far as suggesting doctors will be completely replaced by AI algorithms. But Dr. Topol believes that we should focus on the human element.
“It’s ironic, but this is one chance where we have technology to enhance humanity and it’s not direct, but it’s the outgrowth of all the different powerful aspects of AI for image recognition, speech recognition, all the things that it can do technically to make lives easier and more accurate for doctors and patients.”
The process, tools, practices and facilities of medicine have evolved in tandem with human societies and sciences. AI will bring more enhancements to the field. But one thing that has remained constant throughout history are human doctors. That isn’t likely to change anytime soon.