HTN recently caught up with Tom Ratcliffe, GP and digital lead at Modality Partnership, and Hannah Allen, GP and chief medical officer at Heidi, to learn more about their experience implementing Ambient Voice Technology (AVT) in primary care.
“I’m a GP in the North of England, and I work in a large group of practices based just outside of Leeds and Bradford that is part of the Modality Group,” Tom shared. “I also lead on digital transformation for the Modality Group across England – we have 54 sites serving around half a million patients.”
As part of his work in digital transformation, Tom has overseen the rollout of AVT to Modality’s clinicians. “In primary care, we’re increasingly managing higher levels of complexity, with cases that would have originally been managed in hospitals, being handled in the community,” he explained.
This increased demand has made it difficult to retain people in general practice, according to Tom, who also thinks AVT is “one of a number of tools which are now available to help us with the increasingly time-consuming administrative side of our role”.
Hannah also shared her background: “I’ve been a GP for 14 years, and I made the leap into digital health because I wanted to be able to make an impact at scale,” she said. “I was seeing patients coming in with different problems, and I felt frustrated that I couldn’t help them more in the timeframe I had.”
It was also about empowering clinicians with the latest tools and technologies to be able to deliver care at the top of their game for Hannah, who spoke about founding a women’s health startup after working on a telemedicine project in Rwanda where there were 12 million patients and only 800 doctors.
“I had this lightbulb moment when I tried using Heidi for the first time,” Hannah told us. “I remember thinking how it was just going to change everything – the entire landscape and how we practice and deliver healthcare. I wanted to be at the forefront of that technology, really improving the lives of clinicians and the care we give to patients. I’m still a GP, but now I’m full time at Heidi supporting growth in the UK and further afield.”
Moving beyond AI scribes
Heidi has now moved from being an AI scribe to an AI care partner, Hannah explained, “so now we’re not only using large language models and generative AI to essentially listen to and structure clinician-patient conversations; we’re supporting with downstream tasks like ordering blood tests and X-rays”. At Modality Partnership, that has helped achieve a reduction of 61 percent in out-of-consultation documentation, and an 85 percent time saving in certain departments.
Expanding on what it means to be an AI care partner, Hannah discussed Heidi’s focus on communicating with patients and surfacing the latest evidence to inform and direct care. “We’ve just launched our evidence product, which is super exciting – it provides the most up-to-date protocols, guidance, and frameworks for all clinicians, according to national guidelines like NICE and best practice, but also local and regional guidance specific to the trust or ICB.”
Benefits and impact
Tom mentioned a number of benefits around AVT use in primary care: “It’s not just the time saving, which can be enormously beneficial, especially for colleagues who tend to write longer notes; in my experience it also enables you to spend more time focusing on the patient and being present during the consultation. Making eye contact and picking up on non-verbal cues are key to making the consultation work better.”
AVT can also lessen the day-to-day cognitive load that is driving a lot of people away from general practice, Tom told us. “It means when we come to the end of our surgeries we feel less exhausted – the notes still need careful review, but it’s less draining using AVT.”
The quality of documentation can also be improved through the use of AVT, with Tom pointing to enhanced thoroughness of notes and referral letters. “As organisations using the technology build on additional capabilities, like Heidi’s evidence search product, or voice agent integrations, that will open the door to more potential time savings.”
Benefits extend even beyond these examples, Tom explained, “as AVT can be useful in things like helping draft responses to complaints, or summarising clinical meetings, as you can’t use open access AI platforms where patient identifiable information might be used. Using a tool like Heidi, where security is already set up to allow you to do that, is really helpful.”
Hannah noted the benefits of having Heidi as an AI care partner, working as an “orchestration layer” to pull together tasks and administrative duties that clinicians are normally faced with and optimise the workflow. “That means an end-to-end experience of a clinician-patient interaction, with Heidi AI acting as a digital front door to avoid the 8am rush, pulling together patient information in a way to reduce cognitive load, and then handling those downstream actions like test ordering and patient follow-up.”
Describing her own experience of using Heidi, Hannah said: “It takes away a lot of that middle-of-the-night panic, waking up and wondering if I remembered to order tests for this patient, or to follow-up with that patient; it’s like having a junior doctor in the room alongside you to do the admin side of clinical practice, leaving you to focus on the bit you love, the human part, supporting patients.”
Safety and compliance
Tom talked about how introducing AVT 18 months ago had felt “daunting” at the time, as it had not yet been widely used in the UK. “We spent a lot of time going through compliance and governance arrangements that different providers had put in place, and we ended up choosing Heidi.”
Reflecting on the procurement process, Tom considered the benefits of having a large group of practices with a central governance and compliance team, which helped when putting together safety cases to comply with DCB 0129 and 0160 cases. Information governance was another consideration, he continued, and standardised operating processes were put in place to be used across the whole group.
“For us, it was much more straightforward than it would be for a smaller organisation,” Tom considered, “because that whole process was a significant amount of work.” Measures since developed by NHSE such as criteria for AVT and a list of approved suppliers, is helpful, “but I think for smaller providers it will be difficult to work toward the governance to safely use AVT”.
Existing regulation and governance for Ambient AI can be complex, Hannah shared, “but done well, regulation is really good for innovation – it offers you those guardrails to ensure you’re deploying in a safe way”. She agreed that things like the AVT approved supplier registry have been “brilliant” in providing clarity to clinicians and organisations, and helping demonstrate the impact the technology can have both now and in the future.
“We’re learning a lot,” Hannah continued, “and as a company we will continue to build on our products, as well as helping to share the latest evidence based on updated ICB guidelines, and prompting clinicians to change the way they think about managing their patients. In future, I hope we’ll give clinicians a lot of that work-life balance back, and support them in getting back to the human side of healthcare and improving outcomes for patients.”
Navigating change management
Ensuring clinician and patient voices shape product development has been central to Heidi’s approach too, Hannah told us – with the team holding regular learning groups with both clinicians and patients. “We also release tasks to all of the clinicians using Heidi, which cover all of the actions coming out of consultations – blood tests, follow-ups, X-ray ordering; we’ve generated over 50 million tasks to date, and that has shown us all the ways we can improve and what clinicians want from us in future.”
AVT offers a way of managing things better on a population health level, Hannah said, “which in practice means the end-to-end of patient interaction with any kind of health service is improved, so the majority of patients we see right now don’t need to see us, because we’re practicing more preventative healthcare”. Technology like AVT will help support patients with self-care, she went on, and with pulling information together to enrich conversations or consultations patients do have with clinicians.
“For me, that means instead of asking patients what is wrong with them, I’m asking what is important to them,” Hannah considered, “and we’re not able to do that right now, because we’re in the kind of environment where we’re data rich but information poor”. Ambient AI is helping to level that playing field by surfacing the right information at the right time, to the right person, she added.
When talking about the people and the wider change element of introducing AVT, Tom noted how existing demand for the technology meant it was “almost a ground-up” approach. “Quite often, when you’re doing change management in healthcare, it can be that you’re trying to persuade a lot of people in busy environments to change. With AVT, it was refreshing, because people were crying out to be able to use it – it was more a case of trying to roll out in a quick but safe way so people could start using it,” he said.
The solution has now been successfully rolled out to around 320 clinical staff across the country, according to Tom. “No one technology on its own is going to make general practice sustainable going forward,” he said, “but primary care is a volume business. You’re dealing with a huge amount of results, patients, documents, and so on, so you just chip away at that, you get a bit of a time saving across multiple areas, and suddenly that adds up to a big difference.”
From a patient perspective, Tom told us how he found AVT to be “very acceptable to patients”, due in large part to the widely accepted use of transcription tools in day-to-day life outside of healthcare. And for administrative staff, he outlined how feedback has suggested that AVT has helped with processing things like referral letters, “which can sometimes be difficult to decipher when written by busy clinicians”.
Key takeaways
Tom suggested that when looking to implement AVT, selecting a product and a supplier is key. “There are lots of checks and due diligence activities that need to be done. It’s also important to train staff on the safe use of AI, and to look at what added value the different products available are going to be able to add as they develop. Over the next 18 months we’re going to see AVT platforms becoming much more useful in many different ways.”
“My advice would be to really think about the problem you are trying to solve before deciding to implement,” Hannah shared. “There will be solutions out there that can help you, but talk to your clinicians, ask what problems they are having, and then map those to the right kinds of solutions. The other thing I would highlight is adoption – adoption is important, and it’s our North Star metric – if your clinicians are not using a product, you’re not getting an ROI from it. If people like it, they will use it, and then you can get the return you need for your organisation, your workforce, and your patients.”
We’d like to thank Hannah and Tom for sharing these insights with us.
