ARTIFICIAL INTELLIGENCE IN MEDICINE. PROSPECTS AND RISKS
Artificial intelligence among us
AI has entered our everyday life gradually, but thoroughly. Assistant programs correct our writing mistakes, study our tastes, preferences and needs in order to help us choose a restaurant at the right moment, offer relevant information for us, find old friends, get us out of traffic jams and so on. We can no longer imagine our lives without artificial intelligence. Moreover, we less and less often think about the fact that we are communicating with AI.
There is a phenomenon called the “odd paradox”. It consists in the fact that some time after a new technological discovery appears and is massively implemented, we start using it all the time and treating it as a part of our life, as something alive, but not as an AI. Thus, evolving, becoming more powerful and intelligent, AI penetrates all areas of our life, while becoming more and more invisible.
AI can be defined as autonomous computer systems capable of self-learning and independent decision-making. It is these qualities that distinguish an AI from an ordinary computer program.
There is narrow AI and super artificial intelligence (AGI). We are all familiar with narrow AI. It beats champions at chess with great efficiency, filters spam in mailboxes, and makes diagnoses based on MRI scans. In other words, it is excellent in a certain narrow field and nowhere else.
In contrast a super artificial intelligence (artificial general intelligence – AGI) will be able to do everything. It will absorb all the information that exists in the world today, and all the skills, while continuously learning and improving.
Its capabilities will surpass those of humans, as it will know no laziness and will be capable of absorbing gigantic amounts of information. It is likely that AGI will have a certain self-consciousness. We could already see a prototype of such technology in cinema, such as the V.I.K.I. superintelligence in the movie "I, Robot". It sounds pretty fantastic, but right now several world-renowned mega-companies are seriously working on the development of such AGI technology. The emergence of super artificial intelligence is only a matter of time. Some scientists assume that the world will become familiar with AGI by 2030, but it could happen much earlier. We will try to evaluate all the pros and cons of this event a little later, but for now, let's see how much AI helps humanity in the field of medicine.
AI in medicine.
AI-based programs have been created that can diagnose diseases using CT (computed tomography) and MRI (magnetic resonance imaging) and other scans. Such an AI learns from a large amount of data. That is, it needs to look at a huge number of images of different people. For this purpose, many hospitals and medical centers provide anonymized images of their patients. As a result, the AI demonstrates excellent results. It accumulates experience that exceeds that of the best specialists. It does not get tired and therefore is not capable of making mistakes due to human factor. After all, medical error is not that uncommon. Fatigue, family problems, emotional imbalance – all these negatively affect the ability to concentrate.
For example, according to statistics cited by Academician A.G. Chuchalin in his report at the Second National Congress of Physicians, every third diagnosis made by Russian doctors is incorrect. In the United States in 2015, 10% of patient deaths were due to misdiagnosis and medical errors.
Google's Deep Mind Health-based AI detects the early stages of diseases such as diabetic retinopathy and age-related macular degeneration by analyzing digital scans of the eye fundus. These scans are quite complex and take a long time for an ophthalmology professional to analyze them and make a diagnosis. Trained on millions of anonymous scans, AI does this faster and better than a doctor. What's more, early detection of abnormalities can save vision and prevent disease.
Google Health has created an app, Streams, a mobile medical assistant. This app checks patients' tests and scans. If alarming symptoms are detected, it instantly informs the attending physician and suggests the optimal treatment, saving up to several hours per day for each specialist on this painstaking, responsible, and routine task. Streams is able to predict with great accuracy the risk of acute kidney failure 48 hours before it occurs. This means that it saves lives and the health of people. It is also important to note that such prompt preventive actions significantly save treatment costs for both patients and hospitals. In Cambridge and Massachusetts, PathAI assists specialists in making cancer diagnoses.
LUX AI company has developed QTrobot to work with children suffering from autism syndrome. It is known that such children have difficulties in social communication, learning, and it is extremely difficult for them to focus their attention on anything. Working with autistic children is laborious and time-consuming. QTrobot showed good results. Children held their attention on it for twice as long as on a human therapist. A normal person can show frequent changes in emotion, voice intonation, and gestures that make an autistic child uncomfortable. The QTrobot doesn't often change its facial expression and is more stable, which allows children to avoid distractions and feel at ease. Accordingly, learning material is better and faster. The QTtobot is currently being tested in various therapeutic centers in Europe.
Development of new drugs
The path from the idea for a new drug to its release on the pharmaceutical market is a long and laborious one. It takes years or even decades of meticulous work to find the right formula. A group of collaborators works on synthesizing dozens of molecules capable of theoretically beating the disease. But during testing, most of them are rejected because of unforeseen side effects or low efficacy.
Thanks to its ability to process gigantic amounts of information in the shortest possible time and to anticipate risks and side effects, artificial intelligence reduces the time and cost of developing new drugs many times over.
In 2007, a robot named Adam was tasked with studying the functions of yeast. Adam went through billions of data points and came up with a very precise hypothesis about the functions of 19 yeast genes.
Adam's robot friend Eve discovered that the toothpaste component "triclosan" can be used to fight parasites that cause malaria.
For example, in 2020, a new drug was developed for the treatment of OCD (obsessive-compulsive disorder). The British startup Exscientia and the Japanese pharmaceutical firm Sumitomo Dainippon Pharma worked on it.
Thanks to the computational power of AI, instead of the minimum 5 years, it took only 12 months to create the new formula.
The Babylon Health 360 platform has been introduced in the United States. It takes care of a person 24 hours a day, 7 days a week. Babylon Health, using AI capabilities, analyzes symptoms, suggests what might have caused it, if necessary, quickly connects the patient with a doctor or a nurse. In addition, it offers electronic tools for continuous health monitoring, develops a personal diet and physical activity plan, etc.
The One Drop system monitors and treats chronic conditions such as diabetes, heart disease, hypertension and even overweight.
A psychologist in a pocket
It is no secret that in recent decades the number of people with mental disorders and psychological problems has increased dramatically. But not everyone goes to a specialist. Some are afraid of getting a diagnosis, some get labeled as "mentally ill", and some simply for financial reasons. In order to make the help of a psychologist available to everyone, special programs based on AI are being developed.
Elomiya is ready to talk to you day and night. She is a chat-bot with artificial intelligence, a start-up of a Ukrainian student. This project originated from the idea of making psychological help more accessible. She will always listen, sympathize, and helps you understand yourself. Her training is conducted and supervised by qualified psychiatrists. Elomia is now a chat-bot used by over 4,000 people in different countries. So far, her help is practically free of charge. Soon the creators of Elomia plan to enter the big U.S. market and make a paid subscription.
But in any case, Elomia's services will remain much cheaper than those of a human psychiatrist.
Wysa is a life coach with artificial intelligence to help support emotional and mental health. Wysa has studied over 100,000 professional dialogues. At the beginning of a conversation, she determines the person's condition and, accordingly, selects the right kind of therapy. By observing the interlocutor and interpreting the smallest details, such as the timbre of the voice, coherence of speech, impulsiveness in pressing the keyboard, etc., the AI determines how the person feels. The user may not even be aware that therapy is being applied to him. From the outside it looks like an ordinary conversation. But most notice a decrease in anxiety, an increase in self-confidence, and an improvement in sleep.
The American Psychiatric Association estimates that digital mental health has become a multi-billion dollar industry and includes more than 10,000 apps. Apps range from guided meditation (Headspace) and mood tracking (MoodKit) to text therapy by licensed counselors (Talkspace, BetterHelp).
Today Consis and PharmaBot robotic pharmacists are in the greatest demand. They are a system for storing and dispensing medications. All you have to do is enter the medicine into the robot, and then it finds the best place to store it. The pharmacist makes a request on the monitor screen, and in seconds the robot dispenses the medicine for sale! It is able to make an inventory, taking into account the quantity and expiration date. It performs the most labor-intensive, time-consuming and responsible work in pharmacies, freeing up time for human pharmacists and pharmacy technicians to communicate with customers.
At the same time:
- storage space is saved. For example, Consis stores up to 3,500 packages per square meter;
- the time required to find a drug is reduced;
- the turnover process of goods is optimized;
- pharmacy owners save on staff.
Undoubtedly, the day is not far away when AI will be trained in everything a pharmacist knows and will be entrusted with the job of a "first assistant" – communicating with customers and directly selling drugs. Such a prospect raises the issue of technogenic unemployment.
The Conѕis system is being actively implemented in pharmacies in capitals and major cities around the world. So far its cost is quite high. But over time, the cost will come down, and many pharmacy business owners will begin to think about the benefits of such an AI-based employee. After all, it doesn't get sick, doesn't get tired, doesn't make mistakes, doesn't demand a paycheck, and is always in a good mood.
A nurse robot named Moxi was developed to relieve nurses and to solve the problem of an acute shortage of nurses, for example, in the United States. Moxi connects to the hospital's electronic system, determines the assignment of patients to wards, as well as the schedule of procedures. It can distribute medications, deliver tests to the laboratory, and do a whole range of other important things. Interestingly, Moxi is able to maintain a conversation with a patient and provide temporary psychological assistance.
An innovation by Japanese scientists Robear is a robot for caring for immobile patients. It easily lifts the patient, carries him or her, seats him or her in a chair, etc. Robear will be a great help in caring for the elderly.
Hanson Robotics introduced to the world a nurse robot Grace, who can speak 3 languages and successfully copes with a wide range of duties.
Epson is a robot that takes blood from a vein. Using infrared sensors, it determines the location of veins, the thickness of the skin, and the resistance of tissues. The injection of such a robot is much more accurate and less painful than that of a regular nurse.
Prospects and risks
Artificial intelligence is a wonderful tool that, if used correctly, can take medicine to a very high level. Let us imagine that in the near future AI will help us cope with many chronic diseases, as well as diseases that are currently considered incurable. These include diabetes, cardiovascular disease, osteoarthritis, AIDS, and many others. For starters, AI will help find drugs that completely cure these diseases. But that is only the first step. Then, with the help of AI technology, it is very easy to establish constant monitoring of the condition of each person's body. And this will move medicine into the preventive field. In other words, we won't let people get sick because AI will identify risk factors in advance and take preventive measures. Even more opportunities in medicine will open up with the advent of Super Artificial Intelligence. Just imagine that everyone will have a personal, the best doctor in the world, to whom you don't have to go to an appointment, because he is always with you, always by your side. It sounds wonderful, and these technologies are just around the corner, but they will be widely available only in the format of the Creative Society.
Right now we are facing the ethical question of how to use AI. Because the number of benefits in this area is directly proportional to the number of risks.
At the request of the WHO, an international group of experts has prepared a report entitled "Ethics and Management of Artificial Intelligence for Health", which shows some of the risks and provides recommendations for avoiding them. To summarize, we are already facing or on the cusp of man-made unemployment, AI bias, liability issues, ethical breaches in data collection, the prevalence of private commercial interests over ethics and patient health, the risk of monopolization and use of AI for surveillance and social control.
Developers of AI-enabled robots tirelessly claim that their technology is by no means intended to replace humans in the workplace, but rather to relieve them by making their work easier, more efficient, and more comfortable. That’s how it works at first. But the harsh format of consumer society forces us to chase profits, reducing production costs. In this race for survival, humans cannot compete with AI. As a result, robots are already replacing humans. Employers are finding this solution economically advantageous. Technogenic unemployment already exists, and it will increase every year.
This problem is easily solved in a Creative society by the introduction of UBI (unconditional basic income). No human being will be left out of life and starve to death. Moreover, he will be provided with housing, food, and everything he needs by birthright. The harmonious development of society for the benefit of man implies free education, retraining and employment for all.
The problem of AI bias. Psychologists have identified more than 180 cognitive distortions of the human mind, and this is not the limit. These distortions can be unintentionally passed on by the AI from its creators. Another cause of distortions may be poor Data, that is, insufficient quantity and quality of training information. So AI can make the wrong decisions or decisions that infringe on human rights.
Here's a prime example: In the U.S A., an AI-based program determined which patients might need additional medical care. This algorithm worked with a huge number of citizens – more than 200 million people. In the end, it turned out that the results produced by this algorithm were wrong. The AI favored people with white skin color! All because the algorithm was trained on the date (information) of the population's previous health care expenditures. Unfortunately, whites prevailed among them because their incomes are higher than with a different skin color. This is how the distortion of our society affected the results of AI.
Hence the problem of responsibility. We are entrusting AI with more and more complex and responsible areas concerning human life and health. But who will be held responsible if it makes the wrong decisions? Experts advise that full responsibility for any decisions made by the AI should always be assigned to a specific person or group of people. And the AI itself should never have the status of a person or a legal entity. That is, we must create conditions such that the AI is always under the control of people. Again, this is only possible in a Creative society in which the concept of "power" does not exist. Instead, there is the responsibility of each member of society for developing and living conditions.
Preventing monopolization. The modern consumerist format assumes that everything is bought and sold, the only question is how much. As we can see, all the world's capital is concentrated in the hands of a very narrow circle of individuals. It is not surprising that advanced technologies, which can bring fabulous profits, will also be monopolized if nothing is done about it now. Preventing the monopolization of AI by commercial, private individuals and commercial companies is an important prerequisite for the fair and beneficial use of technology for the benefit of every individual and all of society. Otherwise, it will become just another "progressive" enrichment tool for the few. It is even more frustrating to realize that AI technology can be used for total control, and not for good purposes. In the Creative Society, this problem ceases to exist as a result of the absence of monopolization. Any mineral resources, any energy sources, any advanced technologies belong to all mankind and every person by birthright on the planet and are used exclusively for the benefit of society.
Some experts talk about another problem not mentioned in the report: how will the introduction of AI affect general education? If there is no need for specialists, we will stop training them, then who will monitor the correctness of AI?
Today it is extremely important to raise awareness and involve the entire population in solving these problems. This is a prerequisite for avoiding ugly distortions.
Today, the only project that offers a solution to all of the above problems at once is the Creative Society. This project offers a peaceful, evolutionary way to change the format of society from a consumer society to a Сreative Society in which the value of human life will return to first place. Power will be replaced by responsibility, with the obligatory participation of each person in the development of society. The Creative Society is already supported by people from over 180 countries. They are building bridges towards each other, bridges to a peaceful, safe and secure world. To learn more about the project and volunteer activities, and to support the initiative, visit creativesociety.com. It is very important for everyone to take a conscious stance: support the modern consumerist format or the Creative society. Whatever the choice, it should be made consciously. And to do this, it is good to understand for ourselves:
- What is the consumerist format?
- How does it affect our lives?
- What is the Creative Society?
- What future awaits us if we choose one path or the other?
The main thing to understand is that a position of non-interference is the active support of the consumerist format.
Right now our world is at that very critical crossroads point, when you can’t leave things as they are.
- The paper "Ethics and Management of Artificial Intelligence for Health" is the result of two years of consultation by a group of international experts appointed by WHO.
- Article "Artificial intelligence in medicine: applications and perspectives". – Azoft
- Article "Artificial Intelligence in Medicine. – XX2 century
- Artificial Intelligence and Machine Learning in Medicine" article. Hubr
- Evidence of Human-Level Bonds Established With a Digital Conversational Agent: Cross-sectional, Retrospective Observational Study» – JMIR
- «The odd paradox in defining artificial intelligence». Chad Woodford
- «Bias in AI: What it is, types, examples and 6 ways to fix it in 2022».
- «Sexism and Chauvinism in Artificial Intelligence. Why is it so hard to overcome?» – Cem Dilmegani