Connect with us


Tackling chronic conditions with conversational AI



Health Tech World hears from the co-founder of remote monitoring start-up Aide Health about the unmet needs of people with long-term conditions and the potential of conversational AI to help bridge the gap.

Long -term conditions like asthma, diabetes, hypertension and heart disease account for around 50 per cent of GP time, half of all hospital beds and 70 per cent of all primary and acute care spending in the NHS.

This economic impact has extended to the wider economy, with long-term conditions deemed to be the second most common reason for workplace absenteeism.

One company looking to ease the burden of chronic illness on patients, clinicians and the wider health service is Aide Health, a London-based start-up that last year secured $1.2m (£1m) in pre-seed funding.

It was founded by Ian Wharton and Brian Snyder in 2021.

Wharton, the company CEO, comes from a design background, developing digital products for the likes of Formula 1 and H&M before launching Aide Health.

The inspiration to enter the medical space came from personal experience having lived with long-term conditions at various points in his life.

“I know firsthand the challenges of long-term conditions,” Wharton told Health Tech World.

“I know the impact they have on health care systems and individuals.

“We just wanted to turn our skillset and our experience to try and help people navigate those a little bit better.”

Bridging the gap

Aide seeks to fill a gap in the market for a digital health service specifically catering to long-term conditions.

And with 26 million people living with one or more long-term conditions in the UK right now, the gap is significant.

“Conditions have a significant human and economic impact in pretty much every developed country in the world,” Wharton said.

“Our view is that most of the time, it’s just because people don’t have the tools or the knowledge or the insight to self-manage those conditions as well as possible.”

Aide Health is especially interested in comorbidity – patients with two or more conditions – which Wharton believes is “completely missing from healthcare delivery today”.

“If you think about the challenge today, if you have type 2 diabetes, you might have an annual review of your diabetes; you might see a clinician once, maybe twice a year.

“And then the rest of the time you’re kind of fending for yourself”, Wharton said.

“There are all sorts of conversations that clinicians and patients would love to have in that time about the importance of retinal screening, about what HbA1c means and why that is important for someone’s health, the importance of maintaining regular exercise.

“But, the resources just don’t allow those conversations to happen.”

In one of these annual or biannual appointments, a clinician will ask the patient about their health since their last visit, but this relies on the individual’s ability to remember the ups and downs of their condition over the course of six months or a year.

Aide Health is working address this with with three elements to its platform.

Structured monitoring, which captures the likes of blood pressure, peak flow and symptom severity, provides insights and trends in their condition.

The platform also contains a conversational AI function which has short daily conversations with patients to help manage their day-to-day symptoms and improve their health literacy.

Finally, medicines optimisation helps users understand their medicines, take them more efficiently and share the challenges they may have faced with a particular therapy.

Medication non-adherence: “One of the most pressing problems of healthcare worldwide”

 Medical non-adherence – when a patient does not take their medications as prescribed – is a widespread issue but one that Wharton believes is not spoken about enough.

There are numerous reasons for medical non-adherence. These can be nuanced and complex, especially for people with long-term conditions and comorbidities.

A patient may have difficulties understanding the instructions, physical issues with using the medication, an inability to afford their treatment or simply forgetting about it.

“Long-term conditions, in general, are really complex, very nuanced, very multifactorial,” Wharton said.

“There are all sorts of scenarios and complexities of life that make them hard to manage, and medicine adherence is part of that.”

Regardless of the reason, the WHO estimates that approximately half of patients fail to take their medications as prescribed.

As a result, around 50 per cent of all treatment failures can be attributed to medication non-adherence. And the implications don’t stop at the patient level.

It also has a huge financial impact, with a cost of £930 million per year to the NHS.

Part of this is the immediate costs of medicine wastage, but the knock-on effect on the exacerbation of patients’ conditions put an additional financial burden on health services.

“It’s an enormous magnitude of a problem,” Wharton said.

Aide Health is working to address the problem with AI and natural language processing, specifically by identifying potential reasons for medical non-adherence and providing ongoing support to help patients build positive habits surrounding their medication.

“This is where the benefit, we believe, of natural language can come in,” Wharton said. “One [reason] is people just plain forget to take that medicine.

“They don’t form a positive habit with their medicine around their day-to-day routine. We have a way in natural language of supporting that.

“If you have any concerns about your medicines, chances are you’ll stop taking them. So if you experience any side effects or presumed side effects you’ll become non-adherent.

“You might only engage in that conversation about that with your clinician six months later.

He added: “These are really complex reasons and very specific to the individual. This is part of the reason we chose natural language.

“We can design something highly individualised, something that’s generative, but something that can be deployed at scale to support as many people as possible.”

“AI is not something to be messed around with”

As awareness grows around the possibilities and potential dangers of AI, Aide wants to use AI responsibly, keeping patients at the centre.

The natural language model that powers Aide Health is designed to never go rogue.

On the back end, it may be able to answer any question that is thrown at it, but the model has been designed only to engage in clinically-validated conversations to avoid the dissemination of unreliable and unsafe information.

“It comes down to the question, does a patient know what to ask, what to research and what to uncover about their health? The answer is almost definitely no, they do not.

Ian Wharton

“We can’t rely on them to prompt these systems. We have to design systems that initiate these conversations over time.”

Wharton acknowledges the huge potential of using AI models, particularly in diagnostics, but he stressed that they should never replace or circumvent the clinician.

“There’s definitely a gap there,” he said. “Our view is, it has to be additive to a clinician rather than replace them.

“Looking at it from the lens of chronic disease, what comes with that is, how do you do this safely? How do you do this ethically? And the answer is, it’s very hard.

“When it comes to health, this is not to be messed around with, this is people’s livelihoods.

“Do people need help? Absolutely. But they need help in a way that’s meaningful to them and safe.

“We should definitely explore, experiment and assess the value of these things, but we shouldn’t just use them blindly,” Wharton added.

Integrating into the NHS

Aide Health’s potential was captured in a recent NHS pilot which targeted people with asthma and type 2 diabetes.

The six-month feasibility study involved a small sample size but provided promising early signals for what the platform could bring to the NHS.

Results from the pilot showed that Aide users achieved 75 per cent adherence, 35 per cent more than patients without the app. It also produced positive results around the speed of use.

Users took an average of 10 seconds to report a symptom and around 30 seconds to go through one of its check-ins, which record patient confidence levels and mental wellbeing levels.

Finally, the pilot showed promise for its ability to engage older patients, who are often excluded from health tech innovations.

“Most health tech and the associated research that goes with it tends to have a cut-off at age 65, which we think is strange,” Wharton said.

“It’s needlessly limiting, restrictive and exclusionary. So we didn’t have an upper [age] limit. In fact, our most improved user over the course of the pilot was age 71.

“It’s the first signal that regardless of socioeconomic status, regardless of age, smartphone adoption rates are very, very high. The ability to use conversational AI is very, very high.

“We shouldn’t restrict use from the group of people who have a higher propensity for long-term conditions.”

Aide is now selling actively within the NHS, predominantly primary care networks. The platform currently caters to two conditions – asthma and type-2 diabetes – with hypertension soon to follow thanks to grant funding from the NIHR.

The goal over the next 24 months, Wharton said, is to support the top ten long-term conditions.

“First and foremost, we absolutely want to make an impact in the NHS. That’s how we help people most in this country and that’s the reason we started,” he added.

“We’re speaking to a handful of primary care networks and integrated care systems already.

“The NHS in general is very open and welcoming to innovation to help support their patient populations. It’s quite the opposite of what you hear from certain sides of the industry.

“We’re ambitious. Our goal is 100 million lives globally supported using Aide Health.”

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending stories