Does AI facilitate or endanger the clinician-patient interaction? Battle of the experts

Digital & Innovation: Artificial intelligence (AI) has significant potential to transform healthcare. Experts debated whether AI facilitates or endangers the clinician-patient interaction at the recent Australian Healthcare Week 2021 conference.

Professor Nilmini Wickramasinghe, Professor of Digital Health & Deputy Director Iverson Health Innovation Research Institute at Swinburne University, opened the debate by discussing why AI facilitates the patient experience. With advances in machine learning, the capabilities in AI have increased with sophisticated techniques extracting and making sense of large data sets.

“In healthcare today, we have huge amounts of disparate, multispectral data. It is cognitively not possible for any clinician or group of clinicians to quickly extract relevant knowledge and pertinent information to apply to a clinical setting or context without AI capabilities. In short, without AI, we would be making sub-optimal decisions in any clinical context.

“Further, a large part of my own research in healthcare is focused on designing and developing AI solutions. Through all the scientific studies we have done we have shown positive feedback from patients and clinicians who feel more empowered.

“We’re not talking about a silver bullet. We’re talking about using a tool in conjunction with clinicians to provide better patient care and better patient experience,” she added.

AI solutions can not only personalise and individualise treatments, but also ahead of time predict patients at risk. Therefore, appropriate preventative measures can take place by analysing large sets of data. Clinicians may gain better clarity around diagnosis and prognosis. By having a better understanding the number of unnecessary tests can be reduced.

Professor Wickramasinghe continued “It is important to understand the current challenges we’re facing in healthcare today – escalating costs, aging population, longer life expectancy, the rapid rise in chronic conditions and now we’re dealing with COVID-19.

“Today, we could not have developed the COVID vaccine so rapidly. We could not perform genomics and visual analytics in oncology.

“It’s just two simple letters, AI, but it is a very powerful tool to help us achieve better healthcare and better patient experiences,” she concluded.

Professor Ian Scott, Princess Alexandra Hospital Southside Clinical Unit, Faculty of Medicine, noted that dating back to the 1960s and 1970s there were great promises of expert systems, robotics and AI where the hype did not match the reality.

“More recently, we were told that the electronic medical records (EMR) is the great digital innovation to help patient interactions and improve clinical advice. In many cases EMR has dehumanised clinician and patient interactions. People now spend time in front of computers and monitors rather than actually at the bedside. More time is being taken away from clinical interactions with patients to fill out documentation. Clinicians also see problems with too many alerts, advice and recommendations not relevant to that particular patient,” he said.

According to Professor Scott, it is important to note that AI can produce harm. There have been cases whereby biased data influenced clinician decision making leading to wrong decisions and physical and psychological harm. IBM’s Watson for Oncology (WFO) is a classic example.

“We need more data and trials to understand where AI really works and where it doesn’t. I would argue that AI is promising, but is it time to replace the clinician-patient interaction with AI?” he said.

A/Professor Luis Prado, Executive Director of Academic and Medical Services and Chief Medical Officer at Epworth Healthcare, commented “This fear of dehumanising as a result of new technology advancements in healthcare has been with us for centuries.

“In the 1700s the stethoscope was invented. There was a fear that the presence of the stethoscope would physically and emotionally separate the doctor and dehumanised interaction. Of course this did not occur. In the 1900s, the Harvey Cushing sphygmomanometer was thought to be dehumanising for medical practice. Imagine trying to do clinical practice without one. Whilst the EMR is a problem, I’m sure my colleagues would never go back to paper prescribing over e-prescribing.

“Dehumanising is about fear. It’s the fear that I don’t know what I’m going to do in this new space, I’m going to have to learn something new, I’m going to be challenged or I might even be replaced,” he argued.

Healthcare professionals need to accept new technology, including AI, while determining where and how it should be best used to enhance patient care. AI has already shown to assist medical practice and it does not need to interrupt or disrupt the important patient-provider interaction.

“Reflect on the last time you were at general practice or saw a specialist. How much of that interaction was influenced by the use of technology on the desk, the stethoscope and other medical technology? Was it not about the doctor’s empathy, their care and their communication skills? AI can’t replace that because if it does, that’s on us,” he concluded.