The most important applications of artificial intelligence in medicine

Artificial intelligence is a slogan that evokes various reactions – from delight to fear. And although we meet him very often, it is usually difficult for us to cite specific examples of the use of artificial intelligence. And it is used more and more commonly, among others in medicine.

When we hear about artificial intelligence in medicine, it’s easy to imagine tons of intricate-looking items that fully replace doctors. Such a picture is often presented in popular culture, where instead of a traditional medical visit, we see a virtual assistant and self-diagnosis tools that immediately implement appropriate treatment. On the one hand, reality may seem more prosaic, and on the other – much more fascinating.

What is Artificial Intelligence?

At the beginning, it is worth answering the question of what artificial intelligence is at all. It turns out that talking about it today is not entirely justified. We distinguish between its three levels:

1. narrow artificial intelligence Artificial Narrow Intelligence — ANI), 
2. general artificial intelligence Artificial General Intelligence— AGI),
3. artificial superintelligence Artificial Super Intelligence — ASI). 

Currently, we can only talk about the first stage. The next ones – respectively – equalizing and surpassing the human mind – may seem scary to many people.

So, when we speak of artificial intelligence, we mean systems or machines that mimic human intelligence in performing tasks and can further improve functioning based on the collected information. So they are not tools to replace people, but rather to help them do their daily work.

Artificial intelligence in medicine

One of the areas where the huge potential of using artificial intelligence is seen is medicine. It is worth noting how much hope is attached to it, for example, by the European Commission, which already a few years ago developed a plan for financing AI research in various markets. The amount of funding increases from $ 4 billion in 2018 to as much as $ 20 billion in 2020.

One of the problems that AI is supposed to answer is staffing problems in medicine. As estimated by the American Medical Association, doctors in the United States can spend only 27 percent of their time on contact with a patient. For comparison, as much as 49 percent of this time is spent on keeping medical records. There is no need to explain how this can negatively affect the effectiveness of treatment and the relationship between the doctor and the patient.

In Poland, we also very often receive information about problems with the medical staff. In February 2020, we had just over 140 doctors. If dentists are added to them, the number is close to 180. Taking into account that we have almost 38 million inhabitants in the country, it is easy to calculate that there is one specialist for over 200 people. At the same time, it is worth noting that over 50 doctors, i.e. over one third (!), Are over 60 years old.

“The changes taking place in the field of digital medicine are largely based on the use of artificial intelligence, analysis of large medical data sets or the use of mixed reality in imaging medical data. This is a very real and not as distant, as many people think, dimension of modern health care. It is medical care – and I consider these solutions to be the most interesting at present – which effectively responds to the challenges of an aging society, the growing deficit of medical staff and the need for much better prevention in relation to such civilization diseases as cardiovascular diseases, diabetes or cancer “- says Krzysztof Mędrala, CEO of MedApp SA

1. Also read: Development of digital health strategies in Poland and in the world

It is not surprising that there are voices describing the introduction of AI to the medical market as a technological revolution. It is worth looking at it in practice. Let’s look at how artificial intelligence is used in different fields of medicine.

Cardiology

Cardiovascular disease is the most common cause of death in the world. No wonder that exactly cardiologist is a field in which new technologies are used very widely. Even today, devices for remote diagnostics are more and more popular, which measure, among other things, the work of the heart around the clock. However, there remains the question of using this data and finding any possible inaccuracies.

This is where artificial intelligence comes in handy. Algorithms are able to support doctors by taking a full ECG reading. While human analysis may take a long time, which will delay diagnostics and possible medical intervention, artificial intelligence will be able to analyze the collected data in no time.

How effective can such algorithms be? Just give the example of Rima Arnaut, a cardiologist working at the University of California in San Francisco. He and his colleagues used deep learning mechanisms to teach artificial intelligence to properly classify echocardiograms according to a specific pattern. The algorithm achieved an accuracy of 92 percent, while the team of cardiologists performing the same classification – 79 percent.

It’s not everything. Another team of scientists, this time from the University of Nottingham in the UK, has created a system that scans the routine medical data of nearly 300 patients. Risk factors such as high blood pressure, cholesterol, smoking, diabetes and age were analyzed. On their basis, the algorithm determined the risk of a heart attack or stroke over the next 10 years. The results showed that the algorithm predicted a 7,6 percent risk of a heart attack or stroke. more often than doctors, and at the same time he had 1,6 percent. fewer false positives. According to scientists, using artificial intelligence mechanisms, it was possible to inform 355 patients in advance that they were at risk of a heart attack or stroke.

Oncology

In Poland, the most common cancer among women is breast cancer. The basic examination for the detection of this disease is mammography. Already today, scientists are looking for an alternative that would allow for a faster and cheaper method that would be non-invasive at the same time. One of the proposals is Thermalytix, a body mapping device that uses the mechanisms of artificial intelligence. It is based on thermal sensors, and the first studies using it show promising results.

Women who have the so-called dense breasts are more likely to develop cancer. Moreover, their mammograms may be less accurate and may not show any disturbing changes. Scientists from the University of California, San Francisco have successfully developed software that automatically classifies breast density. It can be a support for radiologists when the density of the breasts do not allow for an accurate diagnosis.

Skin cancer is another type of cancer that uses AI for diagnosis. From smartphone applications to remote diagnostic devices or artificial intelligence mechanisms, experts are trying to find ways to detect it quickly. Stanford University researchers have developed a system trained to recognize nearly 130 skin lesions, including moles, rashes and other lesions. It allows for proper detection of neoplastic changes on the skin of patients without the need for medical intervention. The system was tested by a team of dermatologists, according to which the accuracy of recognizing individual types of birthmarks exceeded 90 percent. Scientists suggest that in the future it could even be used to detect skin cancer at home, even with the help of a mobile application.

Psychology and psychiatry

The use of artificial intelligence in psychology or psychiatry can be difficult to imagine. However, such attempts are also made. An algorithm has been developed at the Vanderbilt University Medical Center in Nashville to predict the likelihood of a patient leaving hospital attempting suicide. It takes into account factors such as hospitalization data, age, gender, medications, and diagnostic history.

How effective can such a mechanism be? Data from over 5 patients admitted to hospital because of self-harm or suicide attempts show that 84 percent of the time the algorithm correctly predicted a suicide attempt within a week of leaving hospital. Moreover, 80 percent predicted that a given patient would attempt suicide within two years.

Sometimes, simpler mechanisms are also used. An example of this is applications that, based on face scans, are able to detect, among others, mood changes, early emotional disorders, and even suicidal thoughts. This is especially important in the case of people who resist working with a specialist.

Support for doctors

The use of artificial intelligence in areas such as cardiology and psychology sounds impressive. On the other hand, there are also activities aimed at supporting the daily work of doctors. They can be a good example tools for scanning drug databases and the effectiveness of their use in specific disease entities. For doctors, this can be a huge help and save time.

It is also worth mentioning the increasing use of artificial intelligence in initial diagnostics. The situation in which In a special application, the patient can perform an e-medical interview, in which he receives further questions based on the answers he gives. Then it is the AI ​​mechanisms that make the patient’s initial diagnosis, which is only later verified by the doctor.

Currently, such solutions are already on the Polish market, but they need to be widely adopted in health care. Without this, the potential cannot be fully realized. “In Poland, we have several very good projects in the field of artificial intelligence in health that deserve attention, among others, StethoMe, Cardiomatics, AIDA Diagnostics and AioCare. I believe that all solutions enabling remote examination of the patient’s condition at home are particularly important for the development of telemedicine and will become more and more popular in the coming years, ”says Piotr Orzechowski, CEO of Infermedica.

1. Also read: The most important trends in e-health 2021

Artificial intelligence and medical devices

The development of artificial intelligence is closely related to the emergence of new medical devices. They occupy a special place here devices for remote diagnostics. Mentioned here include devices for remote heart rate measurement and tools for recognizing skin birthmarks. There are many more devices of this type, including a telespirometer, a remote stethoscope or a thermometer.

Remote monitoring of the patient’s health, combined with AI algorithms, can lead to much faster and more effective diagnostics. “Personally, I am fascinated by the possibilities offered by home or personal medical devices, which allow for remote monitoring of the patient’s health and early identification of disturbing symptoms (eg ECG bands) and for carrying out the initial examination by myself. It would not be possible without artificial intelligence and algorithms, trained on huge data sets. Through such applications, artificial intelligence already allows the patient to significantly greater control of their own health, without the need to contact a doctor directly, ”says Tomasz Domino, COO Infermedica.

Such solutions are offered, among others, by MedApp SA, one of the well-known Polish medtechs. “In order to meet the expectations of modern medicine, in this area, as part of remote monitoring and diagnostics of patients, we offer the CarnaLife System solution, which is an advanced digital medicine platform that allows a doctor to assess and monitor patients’ health and conduct consultations at any time and anywhere. It is an application that aggregates data from patient measurements using remote devices connected to the system. The research results obtained in this way are analyzed 24 hours a day, 7 days a week, using artificial intelligence and big data analysis tools. In the event of any deviations from the norm, the system alerts the medical staff who takes care of the patient “- informs Krzysztof Mędrala, CEO of MedApp SA

1. Also read: 7 trends in telemedicine for 2021

AI in search of a cure for COVID-19

The COVID-19 pandemic has caused a real shock to the medical world. From the very beginning, intensive work was carried out on the invention of a vaccine, but also on the use of existing drugs to help patients with severe disease. As the development of a new drug usually takes many years, the researchers, with the support of artificial intelligence mechanisms, wanted to use existing ones.

According to Nature Communications, one of the studies in this area was carried out by Caroline Uhler’s team from the Massachusetts Institute of Technology. He developed a method of drugs available on the market that could potentially be helpful in combating COVID-19, especially in the elderly. The results turned out to be promising.

The developed system took into account, inter alia, the change in gene expression in lung cells caused by disease and aging. The RIPK19 protein could be a target for COVID-1 drugs. The algorithm identified three approved drugs that affect its expression and may indirectly contribute to inhibition of COVID-19 in the elderly.

Will AI replace doctors?

Artificial intelligence is just beginning to appear in medicine, but it is already a topic that raises many emotions. Some are excited about it, others – on the contrary – apprehensive. There is no shortage of a vision of medical facilities without doctors, where people are treated by error-free, but also emotionless robots. And while this vision may seem interesting to some, it is also – at least for the moment – extremely unlikely.

The main purpose of using artificial intelligence in medicine is primarily support of medical staff. The algorithms being developed are to support doctors in their daily activities – from completing documentation to initial diagnosis, then continuous monitoring and, if necessary, implementing appropriate treatment as soon as possible. Artificial intelligence is not to defeat doctors, but to help them so that they actually have as much time as possible to care for the patient. And it is the COOPERATION between digital technologies and people – both doctors and patients – that is an extremely important step in the development of medicine.

1. Check also other articles available in Digital Health Magazine on medTvoiLokony