The case in summary: Artificial intelligence (AI) is becoming increasingly widespread in the healthcare system, where it is used, among other things, as a support tool for clinical decisions. Some believe that the use of AI can lead to discrimination, as the technology is based on data from the majority of the population and can thus overlook minorities. AI can also increase the risk of the spread of sensitive information and computer attacks against hospitals. The use of AI can potentially help solve some of the healthcare system’s future challenges, such as an aging population and a lack of staff. There is concern that the current legislation is not sufficient to protect against misuse of AI in the healthcare system. The summary is made by an AI service from OpenAi. The content is quality assured by news’s journalists before publication. In recent years, the use of artificial intelligence (AI) has become increasingly widespread in the world, and within a number of areas of society. In the healthcare system, the technology is already used as a support tool for clinical decisions. – It can be diagnoses, treatment choices, prioritization of who should receive health care first and it can be risk prediction, i.e. assessing which patients are at risk of developing a condition that something can be done about. This is what PhD candidate at UiT Norway’s Arctic University in Tromsø, Mathias Karlsen Hauglid, tells news. BIG STEP: In 2019, Mathias Karlsen Hauglid started work towards the PhD thesis, which he has now submitted. Photo: TRUDE HAUGSETH MOE/UNIVERSITY OF TROMSØ Hauglid recently submitted his doctoral thesis “Bias and Discrimination in Clinical Decision Support Systems Based on Artificial Intelligence” at the Faculty of Law at UiT. There he addressed both the positive and negative aspects of using artificial intelligence in the healthcare system. The use of artificial intelligence causes the head teacher at the Norwegian Armed Forces Staff School, Tom Røseth, to warn that it may increase the risk of computer attacks against Norwegian hospitals. Sensitive information There are not only positive aspects to using artificial intelligence in the healthcare system. Several players news has spoken to also point to another challenge: the spread of sensitive information. In order for artificial intelligence to function optimally, a larger amount of data is needed, for example images of hands, if it is to be used to evaluate X-ray images. AI: At Bærum Hospital, artificial intelligence is used, among other things, to analyze X-ray images. Archive image. Photo: Rolv Christian Topdahl / news Even if the images’ metadata does not contain sensitive information, this can be disclosed, the PhD candidate believes. – There is research that suggests that artificial intelligence can recognize and place patients in groups, without the people who have developed the systems or use them being aware of it, says Hauglid, before he continues: – Then we have lost track of how sensitive information is used in a decision-making process. Discrimination One of the country’s foremost researchers on artificial intelligence, Ishita Barua, is impressed by Hauglid’s work, and believes it is unique that we are now getting research into how to take into account both technology, law and medicine in the face of new technology. – It is completely unique in the world context. Barua says that there are physical differences between the majority and the minority of the population, which will be important to take into account when assessing artificial intelligence assessments. UNIQUE: Mathias Hauglid’s PhD work impresses AI researcher Ishita Barua. Photo: Ola Hana / news This is because the technology is based on what it has of data, which to a greater extent will contain data from the majority of the population. At the same time, the technology is dependent on the fact that the minorities have been recruited, the data base for artificial intelligence must be representative. Barua believes this can lead to a particular problem in the health service, all the time collected data does not check whether a person has, for example, an ethnic background. – This is what leads to algorithmic discrimination, and it is something I think we must be very aware of, because now we are in the initial phase, she says. She warns that the use of artificial intelligence can affect hospitals’ ability to offer equal health services regardless of the background of the individual patient or relatives. – I think we are in danger of weakening that offer, of discriminating against certain people if we do not have an offer for those who get an AI tool on them that it was not trained on. Data attacks Barua says that the use of artificial intelligence within the healthcare system occurs when those who want such a tool either develop the data set themselves, or buy “off the shelf”. Which means that they buy a ready-made data set. The fact that such potentially sensitive information is collected in data sets could make hospitals more vulnerable to computer attacks, says head teacher at the Norwegian Defense Staff School, Tom Røseth. – It is possible. Artificial intelligence will increase the amount of data, it will probably also increase the analyzes of actors, i.e. individuals, individually or in groups, he says. RISK: Head teacher at the Defense Staff School, Tom Røseth, believes that the use of artificial intelligence in Norwegian hospitals will make the hospitals more attractive to attackers. Photo: Trygve Heide / news Røseth says that there is a general interest among actors in access to health data for various reasons, and that there may be evil actors who seek to use these for intelligence purposes or financial purposes. He points out that the health data of an actor in crisis or war, such as Russian President Vladimir Putin, is very important. Furthermore, he points out that it is often during peacetime that data is collected. INTEREST: Health data about important players in war and crisis, such as Vladimir Putin, is very important, Røseth points out. Photo: MIKHAIL METZEL / AFP Spies? From the fact that intelligence activity previously revolved around classic handovers of state secrets, for example military capabilities or government security, espionage now takes place digitally. In the Norwegian Security Service’s (PST) latest national threat assessment (external link), it emerged that network operations constitute “a large part of Russian and Chinese intelligence activity in Norway”. THREAT ASSESSMENT: In 2023, Russian intelligence services will pose the biggest threat in Norway according to PST’s assessment. Photo: Helge Carlsen / news Røseth says it is difficult to say whether espionage takes place in Norwegian hospitals, physical or digital, or not. – But it is likely that there is ongoing activity against interesting government officials and health data. – Will there be an increased risk of espionage against Norwegian hospitals with AI? – With AI, there is an increased risk that more complex data can fall into the hands of unwanted actors, but this of course depends on security. Effective For a long time, the healthcare system has looked at how to solve the challenges of the future. An increasing number of elderly people in Norway, combined with an increased shortage of employees, is eating away at the health sector. PhD candidate Mathias Hauglid believes that artificial intelligence can help solve some of the challenges that are now being faced. – That is one of the reasons why it has already been put into use at Bærum Hospital, where a system is used to diagnose small fractures, he says, before continuing: – Artificial intelligence has been introduced there to shorten the time it takes to get a diagnosis, and the resources that go into making a diagnosis. news has previously discussed how artificial intelligence is being used at Bærum Hospital to analyze X-ray images: Strong increase PhD candidate Hauglid believes one of the challenges when using advanced technology such as artificial intelligence is the EU’s outdated legislation. He himself believes that today we use artificial intelligence on an old set of rules. – I think the protection mechanisms are not good enough. – It is better in the upcoming regulations, but we are in a phase now where we risk artificial intelligence being used on a large scale, without the recommended safety mechanisms in place. He believes there is no doubt that a race is currently underway between the Western world and China, and presumably also Russia, related to artificial intelligence. In the coming years, he believes we will see an explosive increase in the use of artificial intelligence here at home. – I believe that in 2024 there will be a doubling of the number of artificial intelligence systems used in patient care in Norway, he says, and believes that it will increase further in the following years.
ttn-69
Increased risk of computer attacks – news Troms and Finnmark
