Our vision

In our vision, we imagine an Aotearoa New Zealand in which the benefits of AI have been captured, at least in primary care.

Jade is a GP in Newmarket and is part of a team practice in the heart of a bustling local community. At lunchtime one Wednesday, Jade calls her grandad to check how he is doing. He asks why she’s not at work and laughs when she says she’s on her lunch break. They didn’t have lunch breaks in his day. He was a GP who retired early in the 2020s, completely burnt out. Her mum remembers the brutally long days he worked during the COVID-19 pandemic and the time it took to clear the backlog of non-urgent medical tasks and follow-ups afterwards. She spent a long time trying to persuade Jade to think of alternate career options. But thankfully, the workload is manageable for those in the health sector now. The smooth rollout of artificial intelligence support across the New Zealand public health system in the late 2020s completely changed the game. At the end of her medical training – which heavily utilised AI, preparing her for AI-supported practice – Jade was excited to be accepted into the GP training programme. This is now one of the most sought-after careers for graduating doctors who enjoy building relationships with their patients in local communities.

In this practice, Jade can access a full range of AI support modules. The basic ones are available throughout Aotearoa New Zealand, with all GPs trained to understand their role in human-centred medicine. The advanced modules are only available in hospitals or large practices in the major centres, with specialist training needed. So Jade also supports rural GPs and their patients remotely. In her practice, most patients arrive having already done a preliminary consultation with the personalised AI healthcare module on their phones. Biometric data is collected on a smartwatch, issued by the practice if the one they normally wear isn’t compatible with the software. If patients have a particular condition or set of risks, specialist monitoring is set up in their home.

So when Hēmi arrives for his appointment, Jade already knows that he has been having issues with his heart rate and blood pressure for some weeks now. The AI has suggested he call in because he has been working on his fitness and sometimes feels very faint after exercise. Jade logs into his file and sees what Hēmi has been told. He is a patient who has opted to receive quite a lot of technical detail as he is very health literate, but the system still has deeper information accessible to Jade. He definitely needs his medication adjusted, and the AI offers a range of possible treatments for Jade to discuss with Hēmi. This is a very efficient conversation, as he had already done some reading and made some preliminary decisions, and so the consultation is there to discuss these and provide some reassurance.

Jade adjusts the medication in the systems and alerts Hēmi’s pharmacist to assess the dosage and any potential interactions by the time that Hēmi gets there. There is also time for Jade to ask some more general questions about his wellbeing, and how things are going in his life. Jade knows there are often additional important personal issues that people do not enter into their health record and prefer to discuss face to face. It turns out there are some stress factors that he can talk through with Jade, including his wife’s health. Hēmi’s wife Ngahuia has been struggling with a wound on her big toe that won’t heal. This is likely exacerbated by her diabetes, which Hēmi worries she is not managing well. While Hēmi is in the consultation with Jade, Ngahuia talks to Colin. Colin is one of the nurse practitioners at the practice and provides patient support, teaching, and monitoring. Ngahuia and Hēmi had their appointments booked for the same time by the AI timetabling system, which was able to access both their schedules and those of the GP and nurse. This system seamlessly books their appointments to enable them to attend simultaneously. While assessing the integrity of the wound, Colin teaches Ngahuia to take care of the toe at home. Ngahuia indicated that she needed some reminders about what she learned, so Colin asks the clinic’s AI assistant to send Ngahuia a virtual simulation about wound management around her big toe. He also chats with Ngahuia about her blood sugar management plan and her concerns about specialist monitoring of her wound at home. Unlike Hēmi, Ngahuia has been reluctant to adopt biometric data monitoring and sharing, so the only data available is from six monthly blood tests. Colin assures Ngahuia that she can receive care as she feels most comfortable. On the way home, she talks to Hēmi again about how his data-sharing supports his health and wonders whether she might trial using the data-sharing system in the future.

Hēmi is also pondering how widely to share his data. His heart issues are likely to include a genetic predisposition, and understanding the risk might be useful for his family. Jade took Hēmi through the options for sharing his data in the whānau-sharing system. Patients can opt in or out of the system, acknowledging that not everyone wants their siblings and cousins to know their medical history. Still, the data-sharing mechanism means that family members’ GPs can be given general risk factors without any specifics, allowing treatment to be optimised for particular genetic risks without personal data being compromised. Hēmi decides it’s a good option for the health of his wider family. With all the basic data, scheduling, and diagnostics handled by the AI, there was more time to discuss the benefits and concerns of data-sharing. Hēmi doesn’t go to the doctor very often, but when he does, he enjoys a trusting relationship.

Jade’s next patient, Sheila, is concerned about her upcoming mammogram, especially after her mum tells her horror stories about the extent of breast compression during the procedure that she asserts are essential to get a good image. Jade explains that in the early days, it was indeed quite an uncomfortable experience, but the image analysis is now much more sophisticated and in three dimensions, which means that optimal compression is much less painful. The AI systems first introduced in the early 2020s are now much more sophisticated, and each mammogram is compared in detail to the patient’s previous image, carefully separating natural changes in breast density from unexpected findings. Abnormalities can now be highlighted and assessed very quickly by an experienced radiologist, supported by AI. Jade takes time to talk Sheila through the process and explains how early detection means that very few women now suffer from advanced breast cancer. These days invasive biopsies are much less likely to be required, thanks to the sophistication that AI has brought to image analysis. She also talks Sheila through the protections in place for her children, whose data won’t be shared beyond the immediate family until they are old enough to consent to this themselves.

Jade also offers Sheila the option for some genomic screening. The full set of genes associated with breast cancer is increasingly well understood. Sheila’s family opted not to enable whānau-sharing with their personal data, but she can still opt to have her genetic information factored into her breast cancer health programme. It gives her a good handle on her personal risk factors and the optimal frequency of mammograms for her. Some women have mammograms every six months and others every five years, enabling the service to target those at highest risk.

Jade doesn’t share the latest research findings because they are a long way from being implemented into clinical practice but is excited by the latest developments in precision medicine that were flagged in the news section of the AI diagnostic module. An early clinical trial on women with a particular mutation has just been carried out, showing that hormone replacement therapy which includes a specific inhibitor for one of the proteins that result from the mutation lowers the risk of
specific breast and ovarian cancers. Jade is not expected to be able to keep up with the rapidly growing body of research; instead, the system provides her with a literature synthesis as well as recommendations and alerts.

Jade’s next patient is new to the practice. Akshita has recently emigrated from the UK where the NHS offers much less advanced options than here in New Zealand. In an extended 30 min appointment, Jade explains how things work here and offers Akshita the option of having blood tests and a full medical exam to populate her baseline data in the system. She talks Akshita through how AI at the clinic can feed her personalised biometric data into the health system from a wristwatch in real-time. Jade emphasised that all data-sharing is strictly opt-in where personal health information can be used in tools that are approved for use in her clinical care, while de-identified aggregated data are used to directly inform the improvement of health services for all. Akshita is somewhat reassured that the data is tightly held for medical purposes only. Having generally low trust in the government, she goes away to think about which option she will take in the degree to which her data and AI will support the relationship with her GP. Jade showed Akshita the health system’s AI chat tool in case she thought of any questions about her data and consent later. She is impressed that the AI can translate into any language.

Next up is Fred, who recently had a hip replacement and is here to discuss his rehabilitation. Although the operation was only a month ago, he seems very mobile as he enters Jade’s office. Ahead of the operation, Fred had a series of scans, which gave the surgeon a precise understanding of the shape of his hip joint. The scans generated a blueprint for a bespoke 3D-printed implant that was seamlessly inserted during the operation using laser-guided robotic placement. Taking the guesswork out of the surgery significantly reduced the duration of the operation and the detrimental impact of the anaesthetic, making the surgery much better tolerated. Both Fred and Jade are excited about Fred‘s improved mobility, which has enabled him to start thinking about going for longer walks again and increasing his general fitness. Jade supports his idea of joining a community walking group, which will also improve his mental health. Fred was left alone since he lost his partner, and one of the worst impacts of his hip problems was reducing his social activity, triggering depression. Fred seems positive as he plans to reconnect with his mates.

Jade suggests that Fred check in with Colin about his health management. On chatting with Fred, Colin notices that he is getting a little forgetful in taking medication and installs a memory-jogging app on his watch. In fact, Colin uses a similar memory jogging app himself. His system prompts him to ask about the gap in Fred’s biometric data due to him forgetting to put on his watch. Colin suggests an alternate memory-jogging system, which might make it easier to remember things. Fred already has a device providing an instant connection device to medical support on a screen near his bed, which he can access by pressing a button. Run by AI, the avatar on the screen knows Fred well, and they have a good relationship. It reminds him of basic daily tasks and automatically alerts the practice if there are any new concerns. Colin suggests they get the AI to remind Fred to put on his watch each morning.

The last patient of the day for Jade, before a couple of home visits, is a Telehealth appointment with Karen who lives on Great Barrier Island. Karen normally sees a local GP when she can afford to but has been referred to Jade before the GP drops by her place to assess whether she may have COVID-19. Karen is on home dialysis and has developed a nasty cough and a temperature over the last couple of days. Jade is trialling a new AI module that listens to a patient’s voice and cough and gives a probability that the cough is indeed caused by COVID-19. There’s a nasty new variant this year, so the health system is on alert. Karen chats to Jade, and the AI listens in and thinks there’s an 85% chance it might be COVID-19—helpful information to relay to the local GP. Fortunately, predicting severe infection from COVID-19 (or other infectious diseases) and particularly the need for hospitalisation has become quite accurate over the last few years. Since performance measures are regularly and accessibly communicated to the public, their conversation reassures Karen. Jade knows that the local GP has undertaken the AI module on early detection of deterioration in patients with underlying health conditions like Karen. While there, Jade asks Karen if she’d like a retinal scan to confirm that her high blood pressure is being managed as well as checking for new problems such as diabetes. She agrees and puts her eye close to the camera on her phone, which is especially adapted
for high-resolution retinal images. The AI runs a quick diagnostic, and everything looks in order, which is reassuring. Jade lets Karen know this, and her GP will be in touch later that day to deliver a COVID-19 test and give advice on the cough.

Before heading home, Jade asks the AI to run through the day, check all the notes, and alert her to any anomalies or omissions. She reviews her schedule the next day, and asks whether she should come in before 9 am to prepare for any appointments. One patient’s file suggests a high uncertainty in the AI diagnostics, so Jade asks the AI to schedule 20 minutes for her to look at this ahead of time, and heads home to her family dinner with Grandad, confident that nothing is forgotten.​