“Artificial intelligence may provide more effective treatment than any psychiatrist in the world. Combining diagnostic AI with therapeutic AI will create a super-psychiatrist,” says professor Marek Krzystanek, pioneer of telemedicine in Poland, in an interview with Monika Redzisz
Monika Redzisz: We are facing a pandemic. Physicians are only seeing people in critical condition, standard appointments are being cancelled. What is a patient in long-term psychiatric care supposed to do if he cannot see his psychiatrist?
Marek Krzystanek*: He should make an online appointment. I am convinced that telemedicine, including telepsychiatry, can save health care right now. When outpatient care is severely limited, clinics should switch to consultations by video and by phone. I would encourage all physicians to provide medical services remotely. As the chairman of the Scientific Department of Telepsychiatry of the Polish Psychiatric Association I have drafted recommendations for physicians and I have also published several manifestos and articles on this subject in “Psychiatria Polska” and “Psychiatria” (“Polish Psychiatry” and “Psychiatry”).
When did you, as a precursor of telemedicine in Poland, come into contact with it for the first time?
In 2012, a businessman friend of mine told me: “Marek, there’s this new thing called ‘telemedicine’. How about we do something with it together?”. He had no idea that I had already practiced telemedicine 10 years before that. In 2003 I ran the psychiatry department for an online medical guide; it was like an educational website. I offered to provide advice by e-mail.
Where did the idea come from?
I had already worked as a physician at the time and I knew how long patients had to wait to see a psychiatrist, especially at the National Health Fund clinics. Secondly, I thought that using electronic mail is an excellent idea for expanding the website’s potential. I worked that way for one year and I summarized my experiences and insights in an article that I sent to “Psychiatria Polska”. I used the word “telepsychiatry” in the title. Reviewers wrote me back, saying that there is no such word in Polish. “Obviously,” I wrote them back. “I’m the one who’s coined it, but I think that in a few years you will be proud to be the first journal to have used it.”
What were your impressions then? Did you think that contact by e-mail is sufficient and that it is a real form of help?
I had the impression that for many people it was the easier form of contact. The people who wrote me frequently said that they would be embarrassed to see a psychiatrist at a clinic. Furthermore, many people found such conversation by e-mail completely sufficient. They received answers to their questions and they didn’t need to see a psychiatrist in person.
So when in 2012 your friend mentioned telemedicine to you…
…I told him that I had practically come up with the idea of “telepsychiatry” by myself. During this first meeting I came up with the general outline of our therapeutic phone application. We named it Moneo after the Emperor Leto’s aide from Frank Herbert’s “God Emperor of Dune”. We created an interdisciplinary team.
Whom was this application designed for?
It was a general-purpose therapeutic application, but we conducted our clinical trials on patients with schizophrenia. It is one of the most difficult disorders in psychiatry. We wanted to see if it could be treated with telemedicine: if it would be safe, effective, and acceptable to the patients. The medical community was very skeptical. It was commonly believed that nothing could replace traditional treatment and that our system would most likely lead to the deterioration in the patients’ condition.
We live in interesting times – we are going back to the renaissance. Scientists can no longer just sit behind the desk and create; they need to be versatile once more
Obviously, I only accepted patients in relapse who took their medicine, but who were stable and showed no symptoms. The trial lasted for one year; two hundred patients received phones with our application installed, while a further one hundred patients acted as a reference group – they were using the application, too, but with severely limited features. More than fifty physicians were also involved in the trials.
I understand that there were no artificial intelligence algorithms involved at the time.
Back then, between 2013 and 2015, we knew nothing about artificial intelligence. Those were completely different times. Our idea was primarily based on videoconferences, or regular video meetings with the physician whom the patient was also seeing once every three months in a clinic. In addition, the application also displayed notifications concerning the medicine that the patients should take. “Do you remember that you need to take your medicine in one hour?” And after one hour the application would query, “Have you taken your medicine yet?” This is one of the greatest problems in psychiatry: half of the patients do not take their medicine, half of the doses are skipped. The longer the disorder lasts, the worse this problem becomes.
The application also included exercises that helped to exercise the patients’ cognitive functions, because schizophrenia can lead to attention deficits, impaired planning ability etc. Medicine cannot improve those functions, so they need to be exercised. And finally, patients also had access to a library of educational videos. I recorded 53 short educational films: what is schizophrenia, can it be treated, what is the acting mechanism of antipsychotic medicine, when can you stop taking your medicine, what is the difference between akathisia and extrapyramidal side-effects, and so on. Simple questions, simple answers, simple language.
Could the physicians check the patients’ activity in the application?
They could monitor it, but most importantly just before the face to face meeting they could check what the patient did over the course of the last three months. Previously, the psychiatrist had had to rely on what the patient was telling him, and such accounts are not always reliable. It happens that people simply lie, just like children lie to their parents. “Of course I was taking my medicine!” Meanwhile, the medicine was left untouched in the drawer. Here, the psychiatrist checked the application and saw how things stood: “A-ha, you stopped taking your medicine!” Over the course of one year we collected a huge amount of accurate and precise data concerning the patients’ mood and wellbeing.
What did you learn after that one year?
It turned out that the results are not only comparable to traditional outpatient care, but they are actually better. The patients’ mood and wellbeing was roughly 6-8 per cent better than in the reference group. The application was perceived very positively by both the patients and the physicians. I heard statements like, “Finally we got something like this for psychiatry!” or “I’ve been waiting for a system like this.”
Didn’t the patients get bored with the application?
Our applications – that one and the ones that followed – are tools that aid the physician. They have been medically tested and registered as a medical device. It is not up to the patient to decide whether he or she will be using it or not. The doctor decides, just like with what medicine to prescribe.
Is the application being used?
Unfortunately not. The investor who wanted to commercialize it wasn’t up to the task, so the project was suspended. I regret it a lot, but it has also taught me much. I started my own company and my own foundation, and I began designing and developing more tools that support physicians. The company is called “Science for Life” and this name is also our mission statement. I work with a few people whom I consider to be geniuses in their respective fields: a software engineer, a project manager, a computer art designer, and a musician. The five of us are capable of doing everything, and much more cheaply than in the case of companies that have 150 employees. Over the last three years we completed a dozen or so implementation grants.
Can you give me an example?
We have already finished several applications and we have been working on a few new ones. In Poland, access to cognitive behavioral therapy and psychotherapy in general is limited. I want to create tools that will improve the situation.
Right now, for example, we are working on a bot that will provide therapy for agoraphobia by using virtual reality. Patients with agoraphobia are afraid of open spaces, of being in crowds, or of going too far from home. In virtual reality a patient can select the environment that he or she is afraid of the most and set the difficulty level like in a video game. On the scale from 0 to 100 the patient defines the level of his or her fear and during sessions that follow he or she tries to become accustomed to this environment. The bot will select the next environment of exposure and increase the difficulty level.
Another bot will act as an assistant to people with obesity. It will guide the patient through a supermarket, through the kitchen and give tips on what to pick, and also stimulate the patient to be physically active. We will equip it with both artificial intelligence and virtual reality.
We have also applied for a grant to the Silesian Center for Entrepreneurship in order to create the Cyber Doctor platform. It would be a tool to replace the doctor’s office. A physician could sit by a swimming pool in Portugal and see patients in Poland, for example.
Artificial intelligence will change the world. We have two paths: wait and watch it happen or be the designers of those changes
Another virtual reality-based application that we created is for people with addictions: alcohol, drugs, gambling, or cybersex. Even those who have been sober for years can get tempted and the thirst can reawaken inside them in the blink of an eye. Our application aims to replace these stimulants and “feed” the patient in order to keep him or her safe from relapse.
How?
This is an area still subject to trade secret. Right now, there is a race in the market to see who creates a better tool. And all of it is based on proprietary solutions.
In order to work well an application needs to have many users that supply data. But it will have many users only when it works well. How to avoid this sort of catch?
I always apply for a subsidy that will be sufficiently high to ensure that the finished application is already perfect.
Are you using artificial intelligence algorithms now?
Yes! I believe that in 10 or 15 years artificial intelligence will make psychiatrists – or even physicians as we know them today in general – obsolete. Surely, AI algorithms in a diagnostic application are able to produce better diagnoses than any doctor. No human doctor is capable of holding as much data in his or her internal memory as neural networks do, nor is he or she capable of processing these data. Algorithms of this type exceed the capabilities of human brain, so they will start very quickly to support psychiatrists in diagnosing patients, and then…
…you’re not trying to say that they will replace the psychiatrist, are you?
I am convinced that the physician will only be supervising the whole process, making sure the machine is working correctly.
People are afraid of this. Most of us do not want to be treated by machines.
I think there is no way back now. In matters of diagnosis this is bound to happen. In therapy? I’m sure that artificial intelligence may provide more effective treatment than any psychiatrist in the world. Combining diagnostic artificial intelligence and therapeutic artificial intelligence will create a super-psychiatrist.
But psychiatry is a very specific area. Human brain is more complex than the heart or the liver. It seems to us that we are so mentally complex and it is so difficult to unravel our emotions and actions that there is no chance for “some machine” to help us.
To me psychiatry is just as simple as cardiology is to a cardiologist. Here, too, we are dealing with symptoms of specific intensity. This information may be represented in the digital reality, and the system may decide to prescribe one form of therapy or another. I see no special differences here.
Is there a chance for telemedicine to take root in Poland right now? What if you create solutions that will end up in limbo because there is no demand for them or because they don’t fit the reality?
It won’t happen. One just needs to think holistically: have the implementations in mind from the start. Yes, in the West telemedical solutions are better known and it is easier to sell them there. They are the most popular in countries in which people live far from one another: in the USA, in Australia, or in Scandinavia.
There’s no point in waiting until the world notices us; we need to act. I’ve learnt of it the hard way and I don’t intend to waste my efforts again. For example, recently I have taken one of my applications to Helsinki, where I met a man who administers a chain of private addiction treatment centers. We live in interesting times – we are going back to the renaissance. Scientists can no longer just sit behind the desk and create; they need to be versatile once more.
But the mere fact that we function in Polish is a pretty big limitation…
Why? One of our applications was released with 15 language versions. Tools for processing of natural languages are improving every day. “Impossible” doesn’t appear in my vocabulary; if a certain solution does not exist, then it simply needs to be created. Artificial intelligence will change the world. We have two paths: wait and watch it happen or be the designers of those changes. I would like to have actual influence on what our reality looks like and that is why I have been implementing all those projects.
Have you observed different reactions to such therapeutic methods between younger and older patients?
For “digital natives” these methods are the most natural thing in the world. For children who spend several hours a day in virtual reality this is their natural environment. I have patients who don’t know how to look me in the eye; they have no need for it because they communicate almost solely through internet messengers. US Army is having difficulty recruiting volunteers because young boys’ brains are adapted from childhood to flat screens and they cannot perceive depth. Such people can’t shoot because they are unable to estimate distances.
This is obviously an alarming sign, but generally speaking, for the young generation virtual therapies are going to be something completely normal. A virtual physician for a virtual society.
It still seems to me that the role of direct contact with another human being cannot be overestimated. Another person will always act as an inspiration…
If we look at the robots that look after the elderly, it turns out that most of them befriend these machines after a few days. It is easy to establish such a relation. Who knows, perhaps at some point we will be able to give machines personality or even create a digital simulation of a specific person’s brain and gain digital immortality? For now, it is still science fiction.
But, yes, indeed, in the modern world psychotherapists act as spiritual guides. People need such roles models. It would be good, however, if such guides first became masters in their field and were able to prove the validity of their claims. They aren’t always able to do it.
Right now, we are working on a bot that will provide therapy for agoraphobia by using virtual reality
In theory we could create such a virtual master, something like a virtual Marek Krzystanek who would talk to people. Machine learning algorithms would have to be incorporated here. I am able to see personally one hundred people a month at most. A virtual Marek Krzystanek could see any number of patients. After a while it could turn out that he is capable of answering more questions than me. It would mean that progress has overtaken us.
Reportedly, the number of people suffering from depression is growing constantly. Can applications help them?
Yes, the number of reactive depression cases is growing. We are unable to cope with civilizational stress and the pace of our lives; our nervous system cannot deal with this. The things that happened over the last two hundred years have exposed our nervous system to extreme stress, which resulted in adjustment disorders: depression and anxiety. Evolution and progress collide. We need to set certain limits, because if we only keep adapting to others, we start to get ill. We need to wake up and understand that it is up to us to decide how much we live for ourselves and how much for the rest of the world. But this is primarily the problem of one’s living philosophy. It would be bad if an application was telling people how to live.
But therapeutical applications, similar to therapy in general, can help a person to restore the barrier between them and the world so that they could hear their own thoughts once more, get in touch with their emotions, and feel their own body. Right now I am working on a VR application for multi-sensory relaxation. Another application of ours teaches breathing techniques. We also have two smartphone applications for self-diagnosis of depression and anxiety disorders. Telepsychiatry has finally become fashionable. Today, no one is surprised anymore by the word that I coined.
*Professor Marek Krzystanek, MD, Ph.D., psychiatrist, sex therapist, and certified cognitive behavioral psychotherapist, director of the Psychiatric Rehabilitation Clinic of the Medical University of Silesia in Katowice. Chairman of the Scientific Department of Telepsychiatry of the Polish Psychiatric Association. Innovator, inventor, precursor of telemedicine in Poland. He creates new mobile applications for the e-health/mobile health sector that use virtual reality and artificial intelligence algorithms to improve human health. Author of more than 90 scientific publications, 100 congress lectures, and two books. President of the Science for Life foundation and director of EduFactory s.c..