A Brief Evaluation of the Impairments Arising from the Use of Artificial Intelligence in Healthcare

Prepared by Büşra Nur Tokis Kuş and Kevser Çetin

INTRODUCTION

With the development of technology, the use of artificial intelligence algorithms and robots is also expanding. It cannot be expected that these tools, which we encounter in almost every field of life, will not be encountered in the field of medicine. These tools undoubtedly increase efficiency in the field of health. As a matter of fact, artificial intelligence algorithms and robots appear in many areas of the health sector. For example, if the heart rhythm is disturbed, the patient is intervened with a device called a defibrillator and returned to its normal rhythm. Recently, an instrument called an Automated external defibrillator (AED) has been developed. This device monitors the general course of the patient and intervenes autonomously in case of deterioration of heart rhythm. It is important to determine who will be responsible for death or injury caused by a malfunction during this intervention.

In this study, first of all, artificial intelligence tools used in the field of health will be classified and the purposes for which they are used will be discussed. Then, the criminal responsibilities of the artificial intelligence algorithm and possible persons in terms of the damages caused by the use of these tools will be examined in terms of Turkish Criminal Law.

I. AI TOOLS USED IN HEALTHCARE

With the invention of the microscope in the 17th century, the first steps of modern medicine were taken, and the understanding of traditional medicine left its place in cellular pathology applications over time. Thanks to imaging devices, x-rays, antibiotics, organ transplantation and genetic research, which are the important inventions of modern medicine in the 19th and 20th centuries, human lifespan has extended compared to the past. When we come to the 21st century, artificial intelligence tools have started to be used in the health sector with the development of computer technologies. Artificial intelligence tools used in areas such as radiology, dermatology, pathology, emergency medicine and intensive care medicine have become important assistants of health workers.[1]

Artificial intelligence tools are used for different purposes in the health sector. These usage areas will be briefly discussed below with examples.

A. Health Protection

One of the important benefits of artificial intelligence is that it helps people stay healthy and maintain their health. Thanks to some artificial intelligence applications and devices, people are encouraged to have a healthy lifestyle. They receive feedback and guidance support in this regard. Let us exemplify such applications:

  • Smartwatches produced by many companies remind their users of what they should pay attention to after surgery. It follows users’ menstrual cycle and blood pressure and gives feedback by measuring daily movements.[2]
  • The “Ada” application, which serves as a digital doctor, provides information about the causes of the disease by asking questions to its users and examining the findings. It also offers the opportunity to meet with a real doctor when necessary.[3]
  • Another artificial intelligence application which developed by IBM, Under Armor and Watson companies provides regular information about sleep quality, activities and nutrition to its users and provides fitness coaching.[4]
B. Robotic Surgery

In the field of medicine, artificial intelligence robots are used in fields such as surgery, hospital logistics, rehabilitation and patient care. Robotic surgery is very successful in minimizing human errors and supporting operations that are difficult to do with human hands. In addition, robotic surgery tools show satisfactory performance in reducing traumas, increasing healing speed, shortening operation times, and suturing incisions. On the other hand, with the success of robotic surgery tools, investments in the surgical robotics market are also increasing. Some examples might clarify them:

  • In 2017, artificial intelligence-assisted robotic devices were used to stitch extremely narrow vessels with a diameter of 0.03-0.08 mm at the Maastricht University Medical Center in the Netherlands. [5]
  • With the robotic surgical tool called “da Vinci”, which is also used in Turkey, the operation is performed without opening an incision in the patient. In this way, the patient recovers in a shorter time.[6]
  • The robot named “Mitra”, designed for nursing homes, helped healthcare workers by taking measurements during the Covid-19 pandemic. [7]
C. Clinical Judgement and Diagnosis

With the help of artificial intelligence, it is possible to evaluate, compare and analyze data. In this way, the diagnosis process to be made by a doctor becomes easier. On the other hand, artificial intelligence technologies are also used in determining the order of priority that will affect the course of treatment.[8]

Here are some relevant examples provided:

  • In a 2017 study at Stanford University, artificial intelligence was tested against 21 dermatologists. The artificial intelligence algorithm has been as successful as dermatologists in detecting skin cancer.[9]
  • The Google DeepMind application can accurately diagnose 50 types of eye diseases just by looking at the scans.[10]
  • With the cooperation of Optellum and Johnson & Johnson, artificial intelligence algorithms will be used for lung cancer prevention and early diagnosis.[11]
D. Treatment

Artificial intelligence algorithms can decide on the course of treatment, the dosage of drugs, and the patient’s care plan by comparing and evaluating the patient’s personal characteristics with the big data stack.[12] In this way, it increases the efficiency of the treatment.

Some of the current examples on the subject are:

  • An artificial intelligence algorithm called MediKanren helps patients in the treatment of ADNP (Activity-dependent neuroprotective protein) syndrome, a rare neurological disease.[13]
  • In collaboration with the University of Leeds and Roche Diagnostics, the artificial intelligence algorithm was tested and yielded successful results in determining the best treatment option for patients with bowel cancer.[14]
E. Research

Existing clinical trials require long-term research and experience, and are costly. According to the report published by the California Biomedical Research Association, only five of the five thousand drugs that have started preclinical tests can enter human testing, and only one of them is approved for human use.  On the other hand, it costs an average of 359 million dollars to develop a new drug in research laboratories.[15]  Thanks to artificial intelligence algorithms, the production and development of new drugs and the preparation of content have been facilitated and accelerated.

Some of the current developments on this topic are as follows:

  • Pfizer has contracted with IBM Watson for Drug Discovery for oncology research. Drug combinations and appropriate treatment strategies are determined using an artificial intelligence algorithm.[16]
  • Pharmaceutical company GlaxoSmithKline has also signed an agreement with Insilico Medicine, which develops an artificial intelligence algorithm for drug research and development.[17]
F. Elderly Care

With the developments in technology and medicine, the human lifespan has extended compared to previous generations. However, towards the end of life, diseases such as dementia, heart failure, and osteoporosis, which limit human movements and lead to death slowly, are frequently encountered. In this period, it becomes difficult for people to do their own self-care.[18] However, elderly care services are also facilitated by artificial intelligence algorithms.

Let us exemplify such applications:

  • An artificial intelligence algorithm named Lilli records the movements of the elderly, allowing them to receive fast and early care or treatment.[19]
  • The robot named Dinsow chats with the elderly and ensures their follow-up, reminds them of medication times and helps them communicate with their relatives.[20]
G. Education

Artificial intelligence algorithms are used not only for patient-centered usage but also for the improvement and development of medical education.[21] Artificial intelligence algorithms draw natural simulations that normal computers cannot do, making it easier to understand and also making it possible to learn outside the clinic.

II. CRIMINAL LIABILITY OF ARTIFICIAL INTELLIGENCE VEHICLES USED IN THE FIELD OF HEALTH IN TERMS OF TURKISH CRIMINAL LAW

Artificial intelligence algorithms and robots used in the field of health are increasing day by day. It is important to determine who will be responsible for the damages arising from their use. There is no specific point to discuss in terms of the damages that occur during the control of these vehicles by human hands or with the help of a command. As a matter of fact, in this case, these vehicles will only be in the position of a vehicle. The main point to be discussed is who will be held criminally responsible for the damages they cause during autonomous and semi-autonomous use of these vehicles. First of all, it will be examined whether artificial intelligence algorithms can be perpetrators, and then the possible responsibilities of the people behind these tools will be discussed.

A. Artificial Intelligence as the Perpetrator

Criminal law, by its nature, is a legal discipline that restricts individual rights and freedoms. Therefore, it is tightly bound to some basic principles. The first of these is the principle of fault. The principle of fault is based on the principle of “no-fault, no crime and no punishment”. In order for the perpetrator to be punished, it is not enough to have committed the criminal act, but also to have acted with fault. In other words, it is not enough that there is only causation between the act of the perpetrator and the result, it is necessary for the perpetrator to act with fault. Shortly, the fault is the result of free will.[22]

On the other hand, Turkish Criminal Law does not accept the criminal responsibility of persons other than real persons. In other words, legal entities have no criminal liability. However, security measures may be ordered for legal persons. The basic premise of this thought is that people other than real persons cannot act based on free will.[23] In order to punish artificial intelligence and robots used in the field of health, it is necessary to examine whether they act freely in terms of criminal law and whether they are condemnable in the eyes of society. In other words, artificial intelligence and robots must have preferred the criminal act. Artificial intelligence algorithms and robots that can learn to do not have the consciousness that humans have and cannot distinguish between good and bad. Since artificial intelligence algorithms are completely human products, they act within the boundaries drawn for them. In this respect, it should be underlined that they did not cause the act constituting a crime by fault.[24]

Another principle is the principle of legality. The principle of legality is regulated in Article 2 of the Turkish Penal Code as follows: “No one can be punished, and no security measures can be applied for an act that is not clearly considered a crime by law. No penalty and security measures other than the penalties and security measures written in the law can be imposed.” The principle of legality prevents the state from the arbitrary treatment of individuals.[25] Currently, there is no regulation in the Turkish Penal Code to penalize artificial intelligence algorithms. However, even if a regulation is considered, this situation should not be contrary to the principle of being a last resort (ultima ratio) of the criminal law.[26]

B. Who May Incur Potential Criminal Liability

It has been revealed above that artificial intelligence algorithms and robots used in the field of health cannot be the perpetrators of a crime. In this section, artificial intelligence algorithms and the criminal liability of the manufacturers, users and programmers of robots will be examined.  

The first thing we will consider is the use of artificial intelligence as a means. In this case, artificial intelligence or a robot is no different from a tool. In this case, the manufacturer, programmer and/or user will be responsible. Of course, in this case, it is necessary for these people to be held responsible for their faulty behavior.[27]

On the other hand, who will be responsible for the damages that occur despite the use of these tools in accordance with the law? Since the damages caused by artificial intelligence algorithms and robots used in the health field can occur in more than one form, it becomes difficult to detect this problem. In the first possibility, although the healthcare worker, often the physician, has shown the necessary attention and care and there is no systemic issue with the artificial intelligence algorithm or the robot, damage may have occurred. Technically, negligence cannot be mentioned because the necessary attention and care has been taken. However, if the health worker is able to prevent the result, negligence liability arises.[28]

In another possibility, the resulting damage is not caused by the physician, but by the artificial intelligence algorithm or the robot. An example of this possibility is a software error, a mechanical error, or an error in new skills after the update. To be concrete, the patient may die or be seriously injured as a result of the robot used in operating rooms cutting the patient’s artery. In this case, it should be said that the physician will not be liable to criminal liability, but after the wrong intervention of the robot, the physician must take over the intervention immediately. On the other hand, there is more than one potential responsible in this possibility. First of all, the responsibility of the manufacturer may arise. The manufacturer may have produced a robot of lower quality than it should have been, by not fulfilling its duty of care during the production of the robot. Similarly, damage may have occurred due to the software developer’s violation of the duty of care in the first version or subsequent updates. On the other hand, the responsibility of the hospital, which does not take care of the robots used, does not provide adequate and necessary training to the health workers, and does not provide the hygiene conditions, is possible according to the conditions of the concrete case.[29]

Finally, the emerging harm may not be caused by artificial intelligence algorithms or robots, but by the ignorance or inexperience of the healthcare worker. The use of artificial intelligence algorithms and robots requires technical knowledge and experience. The knowledge and experience of the health worker are very important, especially in semi-autonomous vehicles. As a matter of fact, a semi-autonomous vehicle needs the guidance of the health worker. In order to avoid this situation, it is essential to provide training to health workers. Otherwise, the health worker will be penalized. [30]

CONCLUSION

Artificial intelligence algorithms and robots used in various branches of the health field are increasing day by day. Although it provides great benefits in the field of health, sometimes some damages occur due to the use of artificial intelligence algorithms or robots. It is a contentious issue, who will be responsible for the damages caused by autonomous or semi-autonomous artificial intelligence algorithms and robots. As a matter of fact, the management and administration of these vehicles are not done by the direction of a single person, but by means of the developed software.

First of all, it should be noted that artificial intelligence algorithms and robots cannot be perpetrators in terms of Turkish Criminal Law. As a matter of fact, Turkish Criminal Law seeks only criminal responsibility and fault liability of real persons. Although artificial intelligence algorithms and robots act independently, it is obvious that they do not have the consciousness to distinguish between good and bad. Also, there is no specific point to be discussed in terms of the deliberate use of these tools in crimes. In this case, the artificial intelligence algorithm or robot will be in the position of goods and the provisions of the Turkish Criminal Law regarding intent will find application. The point that draws attention is who will be responsible for negligent crimes caused by the use of these tools. The first issue to be examined in negligent crimes is the duty of care and attention. Since this situation will differ according to the conditions of each concrete case, it is difficult to determine. However, in terms of artificial intelligence algorithms and robots used in the field of health, it should be underlined that first of all, healthcare professionals, the programmer and manufacturer who wrote the artificial intelligence algorithm, and the hospitals responsible for making these tools ready for use and providing the necessary training to healthcare professionals, have a duty of care and attention.

REFERENCES

*All web contents in the bibliography were accessed on 13.12.2021.

[1]  İlhan Kılıç, ‘Tıpta Yapay Zeka Kullanımında Etik Sorunlar’ in Yener Ünver (ed) Ceza Hukukunda Robot, Yapay Zeka ve Yeni Teknolojiler, (Seçkin 2021) 121.

[2] https://healthcareglobal.com/digital-healthcare/next-generation-wearable-technology  

[3] https://webrazzi.com/2017/04/19/yapay-zeka-destekli-saglik-uygulamasi-ada/   

[4] https://newsroom.ibm.com/2016-01-06-Under-Armour-And-IBM-To-Transform-Personal-Health-And-Fitness-Powered-By-IBM-Watson

[5] https://www.roboticsbusinessreview.com/health-medical/ai-assisted-surgery-improves-patient-outcomes/ ; https://www.engineering.com/story/autonomous-robots-in-the-operating-room

[6]  https://www.haberler.com/ameliyathanede-robot-var-13658260-haberi/

[7]  https://www.theguardian.com/global-development/2020/dec/02/robodoc-how-india-robots-are-taking-on-covid-patient-care-mitra

[8] Pınar Bacaksız and Seda Yağmur Sümer, Robotlar, Yapay Zeka ve Ceza Hukuku, (1st, Adalet 2021) 211.

[9] https://news.stanford.edu/2017/01/25/artificial-intelligence-used-identify-skin-cancer/

[10]  https://www.businessinsider.com/google-deepmind-ai-detects-eye-disease-2018-8

[11]  https://hitconsultant.net/2021/08/23/optellum-j-j-lung-health-initiative/#.YSTebo4zZPZ

[12]  https://www.ibm.com/topics/artificial-intelligence-medicine

[13] https://www.techtimes.com/articles/264042/20210811/ai-known-medikanren-helps-find-treatment-rare-neuro-condition-other-illness.htm

[14] https://www.med-technews.com/news/Medtech-Diagnostics-news/ai-test-could-hold-key-to-bowel-cancer-treatment/

[15] https://www.pwc.com/gx/en/industries/healthcare/publications/ai-robotics-new-health/transforming-healthcare.html

[16] https://www.pfizer.com/news/press-release/press-release-detail/ibm_and_pfizer_to_accelerate_immuno_oncology_research_with_watson_for_drug_discovery  

[17] https://www.fiercebiotech.com/medtech/gsk-taps-baltimore-s-insilico-for-ai-based-drug-discovery

[18] https://www.pwc.com/gx/en/industries/healthcare/publications/ai-robotics-new-health/transforming-healthcare.html

[19] https://www.bbc.com/news/technology-58317106

[20] https://www.webtekno.com/yaslilara-yardim-eden-robot-dinsow-seri-uretime-girdi-h30532.html

[21] Bacaksız and Sümer 214.

[22] İzzet Özgenç and İlhan Üzülmez, Ceza Genel Hukuku (3th edn, Seçkin 2020) 26.

[23] Tuba Kelep Pekmez, ‘Otonom Araçların Kullanımından Doğan Cezaî Sorumluluk: Türk Hukuku Bakımından Genel Bir Değerlendirme’ (2018) 6(2) Ceza Hukuku ve Kriminoloji Dergisi-Journal of Penal Law and Criminology 173, 183.

[24] Koray Doğan, ‘Sürücüsüz Araçlar, Robotik Cerrahi, Endüstriyel Robotlar ve Cezai Sorumluluk’ (2019) 21(1) D.E.Ü. Hukuk Fakültesi Dergisi (Prof. Dr. Durmuş TEZCAN’a Armağan) 3219, 3236.

[25] Özgenç and Üzülmez 28.

[26] Doğan 3236.

[27] Sinan Altunç, ‘Yapay Zekâ ve Ceza Hukuku Sorumluluğu’ in Eylem Aksoy Retornaz and Osman Gazi Güçlütürk (eds), Gelişen Teknolojiler ve Hukuk II: Yapay Zekâ (On İki Levha 2021) 354.

[28] Bacaksız and Sümer 218.

[29] ibid 219.

[30] ibid 220.