Epilepsy care boosted by wearables and machine learning

0
39
Dr Page
Dr PageEpilepsy is one of the most common, most serious and most unpredictable neurological conditions. According to the Epilepsy Society, one in 20 people will have a one-off epileptic seizure at some point in their life – but this doesn’t necessarily mean that they will go on to have epilepsy. One in 50, meanwhile, will have epilepsy for some period of their life – but not its entirety. People who do have epilepsy, meanwhile, may get little warning of an imminent seizure – of which there are around 40 different types. Afterwards, they may forget it ever happened.

It is in part this unpredictability that not only makes it very hard to live with epilepsy – social anxiety and isolation are common problems – but also tricky for medics to support people with the condition, especially against a backdrop of NHS cuts. Dr Rupert Page, a consultant neurologist and clinical lead at the Dorset Epilepsy Service, sums up the demands on his time like this:

We’re a team of three – myself and two specialist adult epilepsy nurses – and we support around three-quarters of a million people in Dorset. So we have to spend our time where it’s most needed. For example, we look to see patients who have had their first seizure as soon as possible, to identify whether they’re at risk of further seizures. But if they never have another seizure, and they’re otherwise well and there aren’t any other risk factors, follow-up appointments at three months, six months, one year and so on might be unnecessary. What we really need to do is direct that kind of support, and those appointment times, to the people who really need them most.

Mulling over this problem of prioritization, Page began to consider how technology might be used to address it. Information technology is an interest of his: he’s chief clinical information officer (CCIO) for Poole Hospital and also its clinical safety officer, which means he takes an active role in assessing new systems and applications that the hospital is considering implementing.

A more data-driven approach, he decided, would enable the Dorset Epilepsy Service to allocate appointments in a different way. More specifically, he saw a role for wearable technologies and mobile apps for collecting data from individual patients and then conveying it to the medical team supporting them:

This would open up the possibility of only offering appointments when people needed them. Because you’re not offering appointments – perhaps unnecessary ones – on a formulaic basis, you can potentially keep them open until much closer to the time of the appointment and then offer them to those who need them most, based on real-time messaging about their level of need. For example, if you’re getting notifications that a patient who you are expecting to get seizure-free through medication is experiencing more seizures, then this is the person who really needs your time. It just gives us an opportunity to be far more responsive.

New solution

Last month, the results of Page’s thinking – and his work on engineering a solution as part of a private/public consortium of medics and healthcare IT partners – were unveiled. Called myCareCentric Epilepsy, the solution combines a Microsoft Band wristband, a mobile app, shared care records, machine learning and data analysis tools.

MyCareCentric Epilepsy works by bringing a wide variety of patient information into a single record, held in a CareCentric system from clinical records specialist Graphnet Health. Some of that information comes from the patient themselves, either by them entering it into the mobile app or web portal, or alternatively via a Microsoft Band wristband that they wear, with sensors measuring movement, sleep patterns, exercise, heart rate, temperature and galvanic skin response. This information can be extremely useful in detecting a seizure and identifying its type, says Page. Other information is contributed by the patient’s GP, the healthcare Trust and other parts of the care community.

The combined data is presented to medical staff in a single dashboard view and alerts notify them when a patient has experienced a seizure. They’re also notified if a patient has been admitted to hospital, so that epilepsy specialists can be on hand to provide advice and guidance to the other medics taking care of them. This has already proved extremely valuable in one particular case:

We had a situation where a patient was admitted to hospital with a fracture, unrelated to her epilepsy. The orthopaedic team were taking good care of her, but were planning to put her on a fast, prior to surgery to fix the problem – but not giving her her medication might have led to her having frequent seizures and probably being admitted to the intensive care unit. But because we knew she was in hospital, the epilepsy team was able to catch up with her on the ward and advise the team looking after her on the medication she requires, in order to prevent that from happening.

Right now, myCareCentric is being piloted by a number of patients. And as more come on board, they will collectively be contributing significant volumes of data. This information, according to Page, will be invaluable in helping medics analyse, for example, to what extent medication regimes and lifestyle factors (such as sleep or exercise) impact on the frequency and severity of seizures. This analysis work will be performed in Microsoft Azure, using machine learning technologies, and over time, could help medics anticipate seizures before they occur and proactively offer advice to patients, so they might take steps to preempt them.

According to Graphnet Health, patient Sean Hamilton was keen to sign up for the service as soon as he heard about it:

The one key thing for me is about being able to have a safety net and a support system so I can actually regain my independence. What tends to happen now, particularly if I am out on my own, is that end up back in an ambulance or in A&E. I hope to get better control of my seizures and ideally to get the life back that I used to have.

Facebook Comments

LEAVE A REPLY

Please enter your comment!
Please enter your name here