In the last decade, the development of artificial intelligence (AI)-based digital technologies for healthcare purposes has increased rapidly. A key to their success, however, is how patients and the public perceive these technologies. A research team has now studied the attitudes of the general public towards AI in healthcare, with a special focus on people with diabetes, a group they expected to have a positive perception, because many of them are dependent on technological solutions. Despite generally positive views towards AI-based solutions in healthcare, people with diabetes expressed the importance of human interaction.
AI-based solutions are already revolutionising medicine. For example, radiologists use AI to analyse X-rays and cardiologists use AI to detect abnormalities in electrocardiograms. AI has thus demonstrated its potential in healthcare systems and can potentially improve patient outcomes and reduce the treatment required and costs. Although AI research is well funded, there is very little knowledge about what makes people more willing to adopt AI-based solutions.
“Patients’ acceptance of AI-based healthcare technologies strongly depends on how they perceive and trust AI. We asked 12,755 people about their current technology use, attitudes towards data sharing and perception of AI in healthcare, including six scenarios for AI technologies. The most surprising thing about the results is probably that, even though the majority saw more benefit than risk, many of the 8,420 respondents were still very uncertain whether the benefits of AI outweigh the risks. Most crucial for people’s openness to AI seems to be that it must not replace contact with a healthcare professional, especially among people with diabetes,” explains a researcher behind the study, Jonas Frey Rosborg Schaarup, PhD student, Steno Diabetes Center Aarhus.
People with diabetes give priority to human interaction
The new study is based on an online survey of 8420 respondents who took a position on six possible future scenarios relating to the use of AI in healthcare and answered 10 questions about technology use, data sharing, knowledge about AI and the benefits and risks of AI. The purpose was to learn how the general public perceive AI-based technologies, the use of which is heavily researched within disciplines such as medical imaging and cardiology. Nevertheless, far more technologies could potentially be rolled out in hospitals and elsewhere in society, where they may affect patients’ daily lives even more directly.
“Despite the enormous potential to transform healthcare, there are still many obstacles to implementing these technologies. What determines whether people have positive or negative attitudes towards these AI technologies has not previously been thoroughly investigated in Denmark’s population, especially among people with diabetes, and better insight into this can certainly affect how digital solutions are designed and how they are implemented in clinical care settings,” says Jonas Frey Rosborg Schaarup.
The six possible scenarios for using AI-based solutions in healthcare were:
- automated image analysis for screening for skin cancer;
- wearable devices that monitor patients at home;
- a chatbot at acute care to assess the severity of the symptoms; and
- monitoring movement using smart sensors on clothing for improving rehabilitation.
In addition, the two scenarios specifically for people with diabetes were:
- automated image analysis for screening for diabetic eye disease – retinopathy – that can cause blindness; and
- automatic analysis of blood glucose measurements for tailored glucose-lowering treatment.
“Many people with diabetes already often use healthcare technologies in their everyday life to manage their diabetes. We therefore thought that their attitudes towards the new technologies would generally be more positive than those of people without diabetes, but we found no evidence for this. In fact, the study suggests that having diabetes is associated with a less positive attitude to replacing healthcare professionals with AI-based technologies, but people with diabetes are still comfortable with the technology as long it is controlled by humans,” explains the senior researcher behind the study, Adam Hulman, Senior Data Scientist, Steno Diabetes Center Aarhus.
No replacement for human contact
Of the 8,420 respondents, 88% had previously heard of AI, 46% agreed that the benefits of AI outweigh the risks, whereas only 2% agreed that the risks outweigh the benefits.
“30% were unsure what to think, and that is a surprising number, especially among people with diabetes, who are used to using technological solutions to manage their blood glucose. We certainly interpret the uncertainty as fear of the unknown and think that education and being transparent about methods is really important,” says Adam Hulman.
According to the researchers, the results also emphasise an urgent need for a more patient-centred process for developing AI-based solutions in the future. This applies to structuring otherwise unstructured clinical notes about patients, collecting patient data and improving and personalising treatment.
“All of this provides incredible insight, the opportunity to improve methods and better results for patients. However, our study very clearly shows that people are still only open to AI if a person controls the intelligent systems, so despite a generally positive perception of AI and its benefits in healthcare, human interaction seemed to play a decisive role in defining positive attitudes – and especially among people with chronic diseases such as diabetes,” concludes Adam Hulman.
