Artificial Intelligence in Medicine: Which Is More – Risks or Benefits?

So, the good news is that we have discovered the secret of immortality.

The bad news is that it's not for us.

Geoffrey Hinton, the "godfather" of artificial intelligence 

 

In our previous publications, we have focused on the strategies and policies by which states regulate the creation and use of artificial intelligence (AI) tools. Among the most important types of human activity in which AI is already used, medicine should be mentioned. Healthcare innovation is advancing rapidly, largely due to the incredible potential of AI, which is becoming more open, fair and affordable around the world.

According to experts from the influential think tank MacKinsey & Company, the use of AI can unlock trillions of dollars in value from banking services to medical and biological research.

Clarifying the economic component, experts claimed as early as the summer of 2023 that generative AI could have a significant impact on the pharmaceutical and medical industry, namely from 60 to 110 billion dollars per year.

In different countries of the world, the population (patients and the active public) have different knowledge about what AI is and have different views on this invention. In September 2023, in Japan, which is one of the most advanced countries in the development of AI, the indicative results of a study by the Commission on Patient and Public Involvement were published.

Patients and members of the public voiced 55 expectations (!) and 52 concerns (!) about AI.

The following were named in six groups of expectations:

• improving hospital administration (in particular, increasing the efficiency of hospital work);
• improved quality of care (including improvement of treatment technologies and easy diagnosis and treatment);
• positive changes in roles and relationships (in particular, friendly relations with AI and closer ones with experts);
• reducing costs (for example, reducing the financial burden on patients);
• better patient experience (facilitating access to diagnosis and treatment);
• reducing inequality (including differences in the level of hospitals and the quality of medical care).

Instead, the following were mentioned in the six groups regarding fears:

• concerns about changes in healthcare (for example, what will happen to patients and doctors in an era when AI becomes more competent and better than doctors?);
• limitation and loss of autonomy (in particular, loss of treatment options and opportunities);
• technical issues and liability (including misdiagnosis, accidents and liability issues);
• the emergence of inequality (in particular, between people who are familiar with and unfamiliar with AI);
• data management problems (risk of personal information leakage and problems with personal information processing);
• implementation costs (distrust in the system).

Although the number of respondents was relatively small, its results can be used in any country that plans to implement AI regulation.

 

Good examples are good news

AI and disease diagnosis

Several years of using AI in healthcare have yielded some useful results. One of the most important advantages of AI for practical medicine is the already improved speed and accuracy of diagnosis. At the same time, fast and accurate processing of large volumes of data makes it easier for medical professionals to diagnose and treat various diseases.

AI tools can already analyze medical images, such as X-rays and Magnetic Resonance Imaging (MRI), to detect patterns and abnormalities that a human doctor might miss. This can lead to earlier and more accurate diagnosis, which in turn leads to better outcomes for patients.

 

AI and clinical trials of new drugs

As reported by the British news agency Reuters, major drug manufacturers are already using AI to, for example, find patients for clinical trials more quickly or reduce the number of people needed to test drugs, speeding up their development and potentially saving millions of dollars.

 

AI and scientific research in the field of medicine

Hundreds of companies in different countries are already using AI for medical research. Striking results have been achieved in the use of AI systems for research on patient diagnosis methods, for example in the field of visually oriented specialties such as dermatology. Studies have shown that AI systems are capable of classifying skin cancer at a level of competence comparable to that of dermatologists. It is worth noting that AI systems need much less time to learn this medical specialty compared to doctors who spent years in medical school and relied on the experience gained during training and diagnosing patients for decades.

German manufacturer of medicines Bayer said it used AI to reduce the number of participants by several thousand for the latest trial of Asundexian, an experimental drug designed to reduce the long-term risk of stroke in adults. It used AI to link the results of mid-stage trials with real-world data from millions of patients in Finland and the United States of America and predict long-term risks in a population similar to the trial.

 

Is AI capable of helping patients in virtual wards?

Virtual wards (also known as hospital at home) allow patients to safely and conveniently receive the care they need at home, instead of in a hospital. The UK's National Health Service (NHS) at national level is successfully implementing virtual wards to support people in the place they call home, including in care homes.

 

Will AI replace doctor's consultations?

Most likely, it will not happen tomorrow. But in the American College of Radiology, you can already come across a radiologist vacancy announcement with such requirements as, for example: "must be certified by the American Board of Radiology and be a well-trained radiologist who is excited about the future where radiologists will be supported by AI and machine learning."

 

Is AI used in Ukrainian medicine?

A pilot project on the use of AI in medicine was launched in Odesa. It is about the BrainScan telemedicine project. This system analyzes computer tomography images in automatic mode and detects pathological changes, significantly speeds up the process of diagnosing brain diseases or injuries, when time is a critical factor.

Ukrainian scientists consider the use of AI to be promising in such areas of medicine as diagnostics, ophthalmology, cardiology, orthopedics, and stomatology.

 

Risks. Or thinking about bad consequences

In the National Health System of the United Kingdom, not only do they say that the country can become a world leader in the field of AI in health care, but at the same time they suggest that it carries risks. First, it will use AI that is (possibly) dangerous. Second, and more likely, the opportunity to use AI to help patients will be lost or delayed as both healthcare workers and innovators hold back while waiting for a regulatory framework to allow them to fully develop.

The Organization for Economic Co-operation and Development (OECD) 2023 report “Artificial Intelligence in Healthcare: Huge Potential, Huge Risks” identifies several other key risks:

• unclear responsibility for managing AI today and its evolution in healthcare and other sectors;
• reducing the capacity of the workforce in the field of health care;
• human and technical resources invested in individual health solutions that benefit only a certain part of the population;
• algorithms that are biased or opaque;
• leakage of personal data due to privacy and security breaches / cyber threats.

 

How states prevent problems in the application of AI in health care

It seems that the United States of America is currently the leader in the development of AI in terms of the number of laws passed that regulate relations in the field of health care. According to the Electronic Privacy Information Center (EPIC), already in 2020-2022, six state laws were passed, the content of which, in particular, related to the regulation of AI as part of comprehensive bills on the privacy of consumers, in particular patients.

Seven states, expressing distrust and skepticism that algorithms can make better decisions than medical professionals when it comes to health care, have proposed bills to regulate the use of AI in physical and mental health services. For example, a bill in Illinois requires AI algorithms used to diagnose patients to be certified and demonstrate accurate results. It would also give patients the right to know that an algorithm has diagnosed them and to opt out of using it to determine their diagnosis.

As we wrote in previous publications, in some countries, in particular the United Kingdom, the state is in no hurry to pass laws specifically to regulate AI.

However, a reliable regulatory system is key to guaranteeing the safety and efficiency of technologies that contribute to the reliability of AI systems. Given the axiomatic need for such regulation and to reduce risks, a completely reliable, efficient and transparent regulatory system has been introduced and is already operating in the country, which includes: (1) regulation of AI products; (2) regulation of health care institutions and (3) regulation of activities of medical workers.

To avoid complexity and confusion in the sprawling regulatory landscape for AI developers and users, regulators of the United Kingdom work with international bodies to produce training materials on the regulation of AI technology and provide access to the information AI developers and suppliers need to ensure products meet regulatory requirements.

The National Agency on Corruption Prevention (NACP) will involve artificial intelligence in the verification of declarations

Regulatory requirements for AI products vary depending on whether the AI product is classified as a medical device. Medical devices must be registered with the Medicines and Healthcare products Regulatory Agency of the United Kingdom (MHRA) and are subject to the Medical Device Regulations, UK Medical Device Recognition (MDR) 2002.

While medical devices are regulated by the UK Medicines and Healthcare products Regulatory Agency (MHRA), healthcare facilities are regulated by the Healthcare Quality Commission (CQC). The CQC deals with the control and inspection of services and assesses whether they are safe, effective and well administered. It publishes standards of care and ensures that services meet basic standards and do not fall below them.

Health professional regulatory bodies such as the General Medical Council (GMC) and the Nursing and Midwifery Council (NMC) are responsible for setting standards of competence and behaviour. These bodies are also responsible for assessing the quality of education and training courses to ensure that healthcare professionals have the skills and knowledge to ensure safe and competent practice.

In Britain, since October 24, 2023, within the framework of the National Health Care System, a special instruction has been in effect that sets the minimum requirements for care in virtual wards for people with heart failure.

International organizations are keeping up with national governments and are also introducing various legal innovations to deal with AI. As early as January 2024, the World Health Organization (WHO) issued new guidelines for the ethics and management of large multimodal models (LMMs), a type of rapidly growing generative artificial intelligence (AI) technology with applications throughout the health care system. The guidelines contain more than 40 recommendations for governments, technology companies and healthcare providers to consider to ensure that large multimodal models (LMMs) are used appropriately to promote and protect public health. The Ministry of Health of Ukraine has already informed Ukrainian doctors about this important document.

New information about the development of AI in the field of healthcare appears every day. The speed and impact of this information has now reached the level of a tsunami. Immortality is still far away. Improving the quality of health care is already in hospitals.

Read this article in Ukrainian and russian.