How AI is getting more women to cervical cancer screenings

Missed medical appointments cost billions, and can mean the difference between early detection and a devastating diagnosis. A pioneering AI platform is tackling medical ‘no-shows’ and getting more women into lifesaving cervical screenings.

The US loses US$150bn to missed medical appointments each year, with the cost of no-shows thought to be around £1.2bn in the UK.

Not only do these missed appointments have a huge financial burden, but they oftendelay diagnosis and treatment that could potentially save lives.

“One of the biggest problems in healthcare, especially in the US, is no-shows,” says Neil Dunwoody, co-founder and COO of medtech company, SPRYT.

“On top of that, over $1 trillion – more than a quarter of the entire US healthcare budget – is spent on administration and scheduling alone. And this isn’t just a US issue. No-shows and the costs of getting patients to their appointments are global challenges. There isn’t a single health system in the world that’s unaffected.”

In a bid to reduce the number of missed medical appointments, including oncology screenings, SPRYT has developed an agentic AI platform, designed to simplify the process of booking appointments.

ASA’s predictive model is capable of forecasting appointment no-shows with up to 92 per cent accuracy, reducing communication costs by 30 per cent.

From sport to healthcare

Initially a sports platform, linking people with others who wanted to take part in activities, SPRYT looked at algorithms that could predict whether someone would show up for activities such as a five-a-side match or a tennis game.

“We realised that, if that kind of behavioral data exists, why not apply it to healthcare? After all, there’s even more data available, and in many ways, the motivations are more explainable,” says Dunwoody.

“People might skip five-a-side because they’re anxious about who they’re playing against. But in healthcare, someone might miss an oncology appointment because they’re terrified of what they might hear. The emotional barriers are very real, and very different.”

For Dunwoody and co-founder Daragh Donohoe, the project became personal when they lost a close friend to cancer.

“He had back pain and was referred for an MRI, but couldn’t make the appointment,” says Dunwoody.

“He tried to reschedule, but the system made it so difficult that he just gave up. A year later, he was diagnosed with stage four cancer and passed away shortly after.”

Since the Covid-19 pandemic, these problems have been exacerbated, and with many GPs only offering a short time slot in the mornings to book appointments, being seen by a healthcare professional can be difficult.

“The truth is that health systems are built for administrators, not for patients. We decided to flip that,” Dunwoody adds.

Meeting patients where they are

In order to reach patients more easily, the platform uses common messaging platforms such as Whatsapp.

“We realised early on that there was no point building another app. There are already too many healthcare apps with low engagement. People just don’t use them,” Dunwoody explains.

“Instead, we looked at how people communicate. Your SMS inbox is filled with banks, utilities, spam, and scams. But your WhatsApp or Messenger threads? That’s where your friends, family, and loved ones are. People you trust. So why not manage your healthcare there?

“It’s free for patients, far cheaper for health systems than SMS, phone calls, or letters, and critically, it’s where people are most responsive. That’s why we built the system the way we did – not around apps, but around the channels people already live in.

“We were the first company in the world to integrate WhatsApp and generative AI into a live health system, specifically with the NHS’ EMIS system.”

How ASA works

At the core of the platform are two key models – a no-show prediction model, predicting with up to 92 per cent accuracy whether someone is likely to miss their appointment, and a no-show reduction model which helps to change no-show outcomes using behavioural science, linguistics, and psychology.

The no-show prediction model combines historical data pulled from electronic medical records, synthetic data to simulate millions of potential reasons someone might not show up, and live conversational data.

“The live conversational data is where it gets really powerful,” says Dunwoody.

“ASA is a fully conversational agent that can speak in 161 different languages. Through these conversations, we can detect all sorts of signals: fear, hesitation, low health literacy, whether English is someone’s first language, and so on.

“ASA analyses how people respond, whether they hesitate, how fast they reply, if they seem confused, and adapts in real time. If someone isn’t understanding the message, ASA can rephrase it, simplify it, or switch to another language to make sure they get it.

He continues: “With the no-show reduction model, the goal is to get the patient to do one of three things: book the appointment, reschedule it, or cancel ahead of time, so that slot can be filled by someone else.

“Most systems stop at prediction. They flag the risk and leave it to the health service to act. What makes ASA different is that it engages directly with the patient, in a natural, empathetic conversation.”

SPRYT has also developed a Retrieval-Augmented Generation (RAG) model which ensures the AI is always referencing pre-approved, accurate information such as FAQs from clinic websites, official guidelines from organisations, and operational information like clinic locations, parking, or childcare availability.

Additionally, Dunwood explains, in areas where literacy or language barriers are common or where people are visually impaired, SPRYT has developed a feature called “voice note tennis.”

ASA sends a voice message in the patient’s preferred language, and they reply the same way.

“It’s more natural, more human, and genuinely accessible,” he adds.

Boosting cervical screenings in the NHS

Following a successful pilot focused on improving take-up of the HPV vaccine, ASA is currently operating within the NHS in north central London, managing cervical cancer screening appointments, and has already seen promising results.

“Booking rates for cervical screenings jumped from 10 per cent to 160 per cent,” says Dunwoody.

“Administrative workloads dropped by over eight hours per week per clinical administrator. We’ve achieved a 33 per cent reduction in SMS costs and over 60 per cent when you factor in saved costs from letters and phone calls. Now, 25 per cent of patients book outside normal office hours, often at night or early morning. That’s the beauty of ASA, it doesn’t sleep.”

Alongside this, SPRYT is also working on other initiatives to support the NHS’s objective of eliminating cervical cancer by 2040, such as providing women with at-home screening tests.

“We contact a patient twice for smear tests and, if they don’t respond or book, we’ll send them an at-home cancer screening test,” says Dunwoody.

“There is a big push in the redirection of NHS England, to keep patients out of hospitals or clinics, and the best way to do that is at home testing for things like this.”

Meanwhile, in the US, the SPRYT ASA platform is now part of the Mayo Clinic innovation exchange, and the company is looking at rolling out a programme at Cleveland Clinic, as well as looking to expand into other areas of healthcare including diabetes and lung screening.

SPRYT’s ASA platform won Femtech World’s Cancer Innovation of the Year Award 2025.