How a digital twin helped AWS’ Alex Hooper navigate a heart health crisis and become a trusted partner

Last year, Amazon Web Services’ Alex Hooper was sitting on his couch watching the Australian Grand Prix when his smartwatch indicated that his pulse had climbed to 100 beats per minute.

At first, he brushed it off.

“I thought, ‘I’ve had a bourbon…I’ll be fine.’ Then, I started breaking out in cold sweats”, he says.

Earlier, while watching practice and qualifying sessions for the race, Hooper, an AI/ML Partner Development Specialist at AWS, had been building a ‘digital twin’ of himself: an AI-powered system capable of sending and receiving messages, and making and taking phone calls, all in his voice and style.

He switched on the service.

Initially, it was just an experimental exercise to explore the ethics of AI delegation. His wife was also connected and began receiving regular SMS alerts.

“What ended up happening is that my wife said, ‘I am really worried about your health’, which I had been playing down. But the AI took it quite seriously and encouraged her to proceed with booking me an appointment [with a doctor].”

Although the system fumbled the booking details, prompting his wife to make the appointment for him, it proved to be a small moment that may have saved his life.

Hooper was diagnosed with mitral valve failure, a condition that occurs when the value between the upper and lower left heart chambers fails to close tightly. As a result, blood leaks backwards instead of forward, which forces the heart to overwork.

He underwent surgery and spent months navigating a complex recovery: medications, physiotherapy, cardiology appointments and late-night anxiety.

An AI research partner and confidant

Throughout it all, AI became a constant companion, not a replacement for his medical team but something closer to a research partner and coach.

Hooper used AI to digest medical literature, feeding research papers into processing tools that converted them into podcast-style conversations between two AI voices. The format helped create emotional distance, making the information easier to absorb.

“It was very different as a journey”, he explains.

“There are many layers to it in terms the ethics, the understanding, the trust, the critical reasoning. You have to constantly assess whether AI is right or wrong.”

“I was never trusting the AI over my doctors, I was using it to do things like sequencing the order of taking medications when I was on multiple medications during or after surgery as well as understanding the physiotherapy side, [asking questions] like, ‘my arm is sore, is there something that I can do better?’

“But then also testing it with the doctors and saying, ‘what do you think of this particular report?’ A lot of them were very positive about it. Obviously, it wasn’t medically-endorsed; it wasn’t approved, it was purely just experimental.”

Doctors agreed with what the AI-generated reports had produced.

“One of the doctors turned around to me and said, ‘this is better than my thesis.’ I’m not endorsing that you trust AI over a doctor at first…but I liken it to when the TomTom [sat nav system] first came out, there were people that followed TomTom off cliffs because they followed it thinking it was a navigation system. The navigation for me is a bit like using TomTom when you know your way home. So, it’s not life or death”, he says.

Hooper found the AI companion was “therapeutic and reassuring’ when he was at his most vulnerable, sitting in a hospital bed with his brain running at a million miles an hour in the early hours of the morning, thinking about what he needed to do next.

“It was reassuring to [ask an] AI, ‘what do I need to make sure is in order if I don’t wake up from the surgery? Because with those kind of journeys, those sagas if you will, I take them as an opportunity and an experience of learning.”

How AI models behave under pressure

Over time, Hooper found that different AI models behaved differently under pressure.

He learned just how quickly the models can shift from being helpful to more conversative.

During one atrial fibrillation episode, his heart rate spiked to 180bpm before dropping to around 50bpm.

One OpenAI voice model gave a blunt directive: call an ambulance, I cannot help you. Whereas Grok was more encouraging but still refused to continue until he agreed to seek help.

“It’s just a slight tweak in it and I think when we’re architecting and designing these solutions as models change and become end of life, we need to be considering the architecture and how we…make the right selection of models because there is a massive change in terms of the user experience as we go through these changes.”

Hooper never took medical advice from the AI, he took it from doctors. But he used AI-generated advice to validate, clarify and explain medical information.

For example, he used AI to double check prescribed medications.

“I had five scripts and I wanted to check they were correct. [I asked the AI], ‘tell me exactly what this drug does?’ I was able to get an understanding, and it correlated with what the doctor had said earlier”, he says.

Helping dad out

More recently, Hooper built an AI assistant for his father using a repurposed voice-over-IP desk phone.

His father has dementia and is blind and the assistant allows him to connect with family and services without having to navigate a smartphone screen.

“So, he can pick up the phone, and the AI will, through an advanced Siri kind of function, [allow him to] call and text.

“The speed at which we can build these technologies, and solve problems is absolutely incredible. The bureaucracy, the time to actually get things implemented…is what takes time.”

He believes Australia’s population decline will drive further investment in this type of healthcare technology to make services more accessible and affordable.

“I think if we can do that in a way that is safe and within ethical guidelines…I think we’ve got an opportunity to completely redefine and reinvent what healthcare looks like for the entire world and I want to be part of that. I want to be driving that as part of my job because it’s so important to make sure that…we’re getting access to [good] healthcare.”