Neuro Event Labs reduces guesswork in neurology

The work of neurologists can entail a startling degree of guesswork – of the highly educated kind – when it comes to treating epilepsy.

Because patient self-reporting remains the principal source of information for neurologists on epileptic outpatients, determining the efficacy of medication and the nature and frequency of seizures becomes decisively more difficult as soon as the patient has left the appointment with a prescription in hand.

“There are studies showing that the neurological events people report don’t correlate to real events at all,” tells Kaapo Annala, the chief executive of Neuro Event Labs.

Night-time seizures can be particularly problematic. A German study, for example, has found that epilepsy patients fail to report almost 86 per cent of seizures occurring during sleep.

“Was there a seizure? And does the patient remember it? These are the most important pieces of information for physicians making decisions about the care pathway and administration. If they don’t understand the history and response of an epileptic patient, it’s a bit of a guessing game, like flipping a coin,” describes Annala.

One of the surely many medical professionals vexed by the problem was Jukka Peltola, a professor of neurology at the University of Tampere and chief physician at Tampere University Hospital (TAYS).

His vexation came to a head in 2015; Neuro Event Labs was founded.

Focus on biomarkers

The Tampere-headquartered startup began its journey by evaluating the advantages and disadvantages of different technologies for monitoring night-time seizures. Sensors, for example, were deemed to provide an adequate degree of accuracy but also present a number of practical challenges, according to Annala.

“We quickly realised that attaching sensors to patients suffering from severe neurological conditions wasn’t an option because the sensors can come off during seizures, they may not detect subtle body movements, the patient can forget to attach them and they can make it harder to sleep,” he lists.

“Then we realised that video provides all the necessary information – and more,” he adds. “The degree of accuracy is higher than with any other solution.”

Nelli, the solution born out of the realisation, consists of four elements: a camera and microphone that record the data; a small locally integrated computer that sifts through the data to detect biomarkers of abnormal neurological events; a cloud-based analytics system that utilises more complex algorithms and machine learning to determine the type, duration and location of the events; and an interface that presents the analyses to neurologists almost in real time.

“The integrated computer uses algorithms to detect all kinds of abnormal events – be it a stiffening of the arm, smacking of the lips or change in respiratory rhythm,” tells Annala.

“The second part is knowing what these biomarkers mean. We’ve taught our system with a large amount of video data – some of which we’ve collected ourselves and some of which is from hospital databases – that this kind of mannerism, movement or signal is indicative of, let’s say, a clonic seizure.”

The analyses enable neurologists to identify changes in the nature and frequency of seizures from one monitoring period to another and compare the findings with changes in the patient’s medication.

“Even though our algorithms have been validated, physicians can review video of each seizure to confirm the analysis,” adds Annala.

One month of monitoring for price of one day

The benefits are fairly obvious as far as neurologists are concerned. More accurate data enables them to make more accurate diagnoses and adjust the medication and treatment in a safe and individualised way.

“Another benefit is understanding what’s going on in the patient’s neurological world. We’ve dealt with lots of cases where it was known that something unusual was happening at night, but it was uncertain whether it was because of epilepsy, a sleep disorder, a movement disorder or a psychological condition,” says Annala.

“By screening the patient with Nelli, we know what the right treatment path is.”

Hospitals, meanwhile, can reduce their costs notably. The current global standard for monitoring seizures is video electroencephalography (VEEG), which costs 3 000–5 000 euros a day and can have a wait time of several months. Neuro Event Labs can start monitoring a patient roughly a week after the request and continue the monitoring for a month on the same budget.

“It means that we’re more likely to catch a seizure,” highlights Annala.

He points out that the solution should come as a relief also to the loved ones of many epileptics: “We’ve encountered lots of difficult-to-treat epileptics whose mums and dads have stayed up for 20 years because they’ve been told that the better they can count seizures at night, the better their child can be treated. They haven’t really been sleeping because of this.”

One of a kind

Neuro Event Labs over quadrupled its turnover from the previous year to 400 000 euros in 2018 and expects the growth to continue at almost the same clip in 2019. Its headcount has grown even more rapidly, jumping from three in early 2018 to almost 30 in mid-2019.

“We have 11 different nationalities and a good balance between men and women, making us a company with quite a lot of diversity,” tells Annala. “Diversity is something we pursue actively. We believe it allows us to get better results than if we were all cut from the same cloth. We can tackle problems from many different angles.”

Some of the future growth will be sought from other sub-branches of neurology. Nelli has already been used in research projects to study sleep disorders, identify cerebral palsy in infants and detect motion disturbances in patients with Parkinson’s disease.

“These are research studies, but daily clinical use isn’t a reality yet,” he specifies.

Despite the well-known limitations of current seizure-monitoring systems, the marketplace is not especially cramped – yet.

“There’s no other system like ours that’s based on automatic video analysis. It really is one of a kind. But we’ll surely face more competition also from video-based systems in the coming years,” Annala tells, sounding unperturbed by the prospect.

“Our advantage is that we’re much further in product development. We’ve created and commercialised a huge annotated library for artificial intelligence and machine learning. It gets bigger the more neurological events are fed into the system – it’s constantly learning and increasing our competitive edge.”