INSIDER MAGAZINE: Could artificial intelligence really one day replace human doctors and therapists?

In as little as five years, the experience of visiting the GP could change radically. Technological advances in the health sector are expected to surge in the near future with the possibility of visiting a medical practice without actually seeing a doctor a potential reality. Instead of checking in with a medical receptionist before waiting your turn to see a doctor, patients could be greeted by ChatGPT-style software, which could input your symptoms and summarise your electronic health records before suggesting a diagnosis. A doctor would then confirm the results before prescribing medication to treat your illness or condition. Experts say this futuristic scenario is just one example of the role artificial intelligence can play in the medical field — and significant strides are already being made in Australia. In February 2022, the Therapeutic Goods Administration approved an AI-enabled device from US firm iHealthScreen that can screen for three eye diseases. The iPredict system was designed for the early diagnosis of diabetic retinopathy, age-related macular degeneration and glaucoma, including retinal imaging. Alex Jenkins, director of the WA Data Science Innovation Hub, says the iPredict technology is just “the tip of the iceberg”, adding healthcare and medical research are the two industries set to benefit the most from AI. The hub, a WA Government initiative supported by Curtin University, aims to ensure the State remains at the forefront of data science and AI-led technologies. “Already, there are significant advances in areas such as medical imaging and AI is really good at analysing data and extracting insights from that data,” Jenkins says. According to Jenkins, AI has the equivalent of thousands of years worth of doctors’ experiences. “What we do when we train these AI is that we combine data from many, many different sources,” he says. “That is a very powerful technique that will enable us to make better treatment decisions and more personalised recommendations.” Jenkins clarifies he is not advocating for AI to take all the power. “Instead of AI replacing a doctor’s job, think about how a doctor’s service might be improved if they had a whole team of assistants working with them for every patient. You can delegate some of your tasks (to AI),” he says. “The end goal here is not to replace people’s jobs, the end goal is to increase the level of medical care that we can provide to the general public.” Simon Kos, chief medical officer at Microsoft for Australia and New Zealand, says patients should not be scared of an intangible technology like AI. “At this current point in time, patients’ lives are in the hands of clinicians using technology of all sorts of shapes and types, so (AI) is just another technology and more technologies will come,” he says. “The difference is this is a new technology . . . and that discovery has come before the regulation that gives people added assurance and confidence.” Kos says the science of medicine, which involves large volumes of data, will become increasingly computerised. However, the human element of medicine will also remain important. “I think we need humans to step up and be empathetic . . . break bad news and develop trust,” he says. “I think there’s accountability, so when these decisions get made, however they get made, at the end of the day, someone needs to make the final decision.” Within Microsoft’s research unit is a major division dedicated to health called Health Futures, which conducts primary research into AI and cloud systems for health use-cases. Kos says Health Futures focuses on research with near-term applications, creating intellectual property that can be used in products, open-sourced or spun-off. He works with Australian health providers to see how these innovations and other more mature technologies can digitally transform health care. Adrian Anthony, Adelaide-based general surgeon and Royal Australasian College of Surgeons censor-in-chief, believes AI has a “definite” role to play in the medical field given the amount of data captured. Anthony says while AI already has a role in reading X-rays and pathological specimens under the microscope to diagnose health conditions, it is also emerging in robotics surgery. “(This is where) AI will be tracking what the surgeon is doing and then potentially will say, ‘no, don’t cut that, that is a dangerous manoeuvre’,” he says. More importantly, Anthony adds, surgical training will be heavily supplemented by AI in the next 10 years. “At the moment, the supervisor has to find the time to sit down and have that discussion with the trainee. If you have the AI capability, that might happen more in real time or more frequently,” he says. “Potentially, (training) happens in a very non-judgmental unbiased way. If we get (along well) with a trainee, we might give them glowing reports . . . AI is much more likely to be objective.” Kos is also excited about the potential to train the next generation workforce with AI. “When I went through, I read textbooks . . . you’d develop clinical acumen and skills in a simulated environment,” he says. “I think we could use this technology to provide a highly realistic simulation where practising doctors, nurses, allied health professionals can interact with (AI) and get human-like responses and hone their skills.” Anthony says AI will have to meet broad parameters in surgery. “One of the challenges that we have with AI being incorporated into surgery is ensuring that it is reliable, it’s valid and that it actually supports patients’ safety and efficacy of treatment and outcomes, because they’re the sort of things that we now look for in human performance,” he says. He admits the medical field, particularly surgical practices, is still trying to “find our feet” in understanding the concept of AI and the breadth of the technology’s application. He admits there are some fundamental challenges regarding ethics and the legality of capturing patients’ data. “In healthcare and surgery, a lot of that information is very sensitive and highly confidential information,” he says. “One of the challenges we have is (figuring out) who owns that information. (There are questions about) my responsibility to tell the patient that their sensitive information is being captured and how we protect anonymity. “There needs to be real time component to how information from AI is being incorporated into surgical practice.” Asked if AI will replace surgeons, Anthony’s response is: “Certainly not in the foreseeable future.” “There will always be a (human) surgeon interacting with the patient, interacting with other staff, there will always be a surgeon making the key decisions,” he says. “Potentially, it could mean that surgeons make a decision but then AI actually confirms that decision, as opposed to (surgeons) going to AI to help us make a decision.” There are also questions about how the use of AI in surgery will impact equity of access, says Anthony. “Every time we see new technology appear in surgery, there is inherently inequity in who accesses that technology,” he says. “In Australia it relates to geography and to some extent, socio-economic class.” Perth-based Dave Newman, founder and chief executive of PlanCare, has realised AI can be used in something as “unsexy as aged care or older adults”. The platform he founded in 2021 works as a provider of home care packages for older Australians. It allows users to choose from a list of local service providers, such as health professionals and cleaners. PlanCare also provides a monitoring feature allowing a doctor or health professional to write comments on the system after a visit. This is aimed at detecting early signs of deterioration in the client’s health. PlanCare, which services more than 1000 clients across Australia, is now conducting tests to determine how AI can provide value to its clients. As part of an internal project, PlanCare took doctors’ summary notes on patients and put them into what Newman calls a version of ChatGPT. “When a human looks at shift notes, it’s really hard to get context of trends of a client’s well-being,” he says. “When we put the same shift notes into ChatGPT, we just asked it a simple question, ‘Do these shift notes show any cause for concern?’ and sure enough, it came back and said, ‘Yes, this particular client shows early signs of cognitive decline.’” Newman says it was an “aha” moment of discovery. “(We found out) that in this instance, we can use AI technology. It can really provide a benefit because what it can do is look at trends overtime, which is a lot harder for a human,” he says. “We are really at the early days of it. When tools like ChatGPT came online last November, it’s just opened up this new wave in the health industry for companies like us to really start to leverage these things and drive value for clients.” Melbourne-based Meg Price, co-founder of AI-powered health coaching platform Noa Coach, says more people are willing to talk to AI about mental health and stress instead of a human being. Price pointed to separate research from US computer software company Oracle and mental health app Wysa, which showed nearly three in four people would rather speak to AI than approach human resources staff in the workplace to talk about such issues. More than two in three people would also rather talk to AI than their GP about stress and mental health. “A lot of it seems to come down to ‘the AI doesn’t judge me, I don’t have to even think about what I say or how I say it’ . . . and the AI is available 24/7,” Price says. “What does that say about our human skills to be able to really truly listen and empathise with someone if an AI is doing a better job?” Price says Noa Coach — available on Google Play, the Apple app store and on the web — allows users to engage in confidential and personalised conversations with an AI coach. It ensures individuals can access support without the potential discomfort or unease associated with face-to-face conversations, she says. Jenkins has also noticed more people using ChatGPT as a form of therapy. “Certainly, ChatGPT was not designed with this in mind. But people are often reporting that they’re able to tell things to an AI and be more frank and honest than they would be as opposed to sitting in front of a doctor because they know that they’re not going to be judged,” he says. More than 1000 users across Australia, China, the United Kingdom, North America and the Netherlands, have signed up to Noa Coach since it was launched in July. Price expects this number to grow at an exponential rate based on its marketing efforts. She encourages more people to see the potential of AI before completely disregarding it, saying there is nothing to be scared of. “I think the main thing that people are concerned about is what’s happening to their data and I think that’s an absolutely valid concern,” she says. “We are very strict with what we do . . . we depersonalise all data and what we use for machine learning is only if the user says we can. I think it’s really important that people assess (where their data is going) with any AI they’re talking to and we probably don’t do that enough. How many times do we talk to Siri or Alexa and we’re not checking where that’s going.” Jenkins says we should not be afraid of AI but advises everyone to insist their personal medical data is protected at all times. “We’ve all seen what’s happened with the obvious data leaks throughout the year,” he says. “If we are going to use these software systems to handle people’s data, then we have to have the absolute highest level of protection to ensure people’s private information is safe. For me, that is a higher risk than, for instance, an AI making an incorrect assessment of my health. Given the rigorous processes that we have to follow in this country in order for any kind of medical technology to be implemented, people should not be scared.” Price hopes to see more women and people from different countries and cultures building AI to operate in harmony with humans. “Those that are thriving in an AI world will actually have a different level of digital literacy,” she says. “Those people that are successful (in the future) are going to understand how AI works and they’re going to be able to evaluate online information critically.”