“One minute delay in stroke treatment causes a reduction of twenty-two healthy life days for a patient,” says Razmara Nizak, Head of Innovation at Nicolab.
Speaking at the sidelines of AWS Summit in Amsterdam last month, Nizak was not only showcasing Nicolab’s Stroke Viewer, a stroke diagnostic technology software but also the impact of AI in emergency care.
While AI is seemingly everywhere these days, it has been assisting doctors for quite some time. With its Stroke Viewer, Nicolab is particularly assisting doctors who treat stroke patients.
Founded in 2015 by Merel Boers and Renan Sales Barros as a spin-off from the Amsterdam University Medical Centre, the Dutch startup is a great example of how human and artificial intelligence can save lives.
Solving a time sensitive disease
Nicolab was born from MR CLEAN, a randomised trial of intra-arterial (IA) treatment for acute ischemic stroke that aimed to answer the question of whether an IA intervention with usual care can lead to improved neurological outcome.
In the case of stroke patients, every minute counts and faster a patient is treated and better are the chances of survival.
Nizak explains that the MR CLEAN trial led to international approval of a treatment for stroke patients, “where instead of giving the patient a medication, you really go into the patient’s arteries and vessels, removing the blood clot.”
This type of intra-arterial intervention restored the blood circulation in the brain and offered stroke patients a higher chance of survival and an improved quality of life.
Boers and Sales Barros, who were part of the MR CLEAN trial, saw the potential of this life changing treatment, and began working on an application that further helps the doctors.
Called StrokeViewer, the application uses AI to analyse imaging data and helps doctors to diagnose stroke faster and better than previously possible.
The application not only helps doctors make decisions faster but also makes patient data and images accessible to them anytime and anywhere.
A patient’s journey
Nizak believes the best way to illustrate how Nicolab’s StrokeViewer works is by looking at a stroke patient’s journey.
He says when a patient gets a stroke, the first step is usually to call an ambulance and the patient is transferred from their home to a hospital.
Once in the hospital, the stroke patient is first given a very basic CT scan and the doctors use the imaging data to decide whether the stroke is caused by bleeding or by stoppage.
“And that’s very important, because the treatment for the two diagnoses is different,” he says.
If there is a stoppage, the Nicolab’s algorithm kicks in to identify where the stoppage is and also how bad the brain tissue is damaged.
This is possible because Nicolab has trained its AI algorithm on a variety of images acquired through national and international collaborations.
Nizak explains, “This is done in order to make sure we have a heterogeneous dataset accounting for variation in population and imaging devices.”
Through the data, Nicolab has built an AI algorithm that it calls generalisable and has reduced risk for bias.
“The Al models and improvement of the models are bound to strict requirements defined by regulatory bodies and laws (FDA USA and MDR EU),” Nizak says.
“We keep improving by listening to and learning from feedback provided by end-users (clinicians) and can retrain models by adding new data or tweaking existing models.”
In case of the aforementioned stroke patient, the longer delay in treatment could lead to worse damage in brain tissue. With its algorithmic data tool, Nicolab helps doctors make the decision on whether to treat the patient faster.
While it serves a purpose when the doctor is available in the hospital, its purpose alleviates when the experts and doctors are not in the hospital.
Nizak says decision making can get delayed when the doctors are not immediately available and multiple people need to be consulted since they don’t have access to the information.
With Nicolab’s StrokeViewer, the data is available for the doctors anytime and anywhere, which allows for them to reach the decision faster and save lives in the process.
AWS to power platform-agnostic tool
StrokeViewer, Nizak says, is a platform-agnostic, cloud-based web application and a mobile application that can be used on iPhone or an Android smartphone as well as a tablet or a computer.
This flexibility to access data means doctors can access the patient data and scan images through any device at their disposal.
Using the data on StrokeViewer, doctors can decide whether the patient needs treatment even before getting to the hospital.
Nicolab is a cloud-native solution that runs on AWS. Nizak says they chose AWS because of its ease of use and scalability.
Another reason being that many hospitals use AWS as their preferred cloud platform, which “enables Nicolab to run its StrokeViewer platform and improve patients’ lives.”
Nizak explains that they have both small hospitals as well as big ones using StrokeViewer and the number of images coming in can vary depending on the hospital.
With AWS, Nicolab is able to scale its platform use for each and every hospital while its AI algorithms, which rely on graphics performance, continue to improve in the background.
Nicolab uses Amazon EKS, a managed Kubernetes service, as well as its authentication service to correctly process the data and give access to the right users.
While Nicolab securely processes the data at its end, it also ensures that only authorised users are able to access that data by using AWS-powered authentication system or hospital’s own authentication.
As a leading cloud service provider and designated provider for some organisations, Nizak says AWS has also helped them speed up customers’ contracting and tender processes.
Human and artificial intelligence
When it comes to AI, there is a duopoly of thoughts. The AI proponents argue that the tech will supplement human intelligence while its opponents often paint a doomsday scenario.
Nizak says human intelligence and artificial intelligence together make a better doctor. “Humans and AI together can achieve a higher performance,” he adds.
He wants us to see the use of AI by doctors in the same way as doctors consulting with their human colleagues.
In case of hospitals which don’t have multiple doctors to consult, Nizak sees AI filling the gap with its expertise and allowing doctors to make more confident decisions.
Nizak has no doubts that the introduction of AI “improves the accuracy and also benefits the patient and patient outcome.”
Can AI replace human doctors? Nizak is quick to admit that AI is not 100 per cent accurate.
He draws a parallel between medical AI and generative AI, where a tool like ChatGPT seems incredibly intelligent at times but also fails to answer simple questions.
With StrokeViewer and Nicolab AI, Nizak says AI is only analysing the images but only the doctor has complete patient history, allowing them to contextualise better.
It is remarkable how Nicolab has found use of AI in not only training patient data but also helping doctors draw better inferences for treatment.
However, the more remarkable fact is that Nicolab has achieved all this with A$13M in funding and around €2.2M in grants from the European Innovation Council.
The Amsterdam-based startup with strong roots in Australia has grown from 12 to 50 people since 2019 and has expanded to the US.
Technology to treat other acute diseases
While Nicolab is mainly focused on helping doctors treat stroke patients faster and in an efficient way, it is also looking to expand to other acute diseases.
Nizak says the infrastructure designed by Nicolab can be used by any medical device or medical imaging setup.
He says they are looking at areas where their technology can be used to speed up the workflow across hospitals and their networks.
“For acute diseases where time is critical, every minute saved dramatically improves the patient’s outcomes,” he reiterates.
He further adds that Nicolab is looking into leveraging its software and platform in other disease areas including treatment of heart attack, abdominal traumas, abdominal aneurysm or ruptures.
The startup is currently piloting expanded usage of its software platform with a hospital in Belgium.
Nicolab’s software and AI technology is designed to benefit any healthcare system where the treatment is dependent on image exchange and there are bottlenecks in collaboration or communication.
The startup is primarily operating in developed countries because of access to cloud technologies and robust mechanism for licensing medical software but it doesn’t rule out offering its service to developing countries.
“Our aim is really to remove bottlenecks in collaboration within and across hospitals,” says Nizak, adding, “that can be translated also outside developing countries.”
Read the orginal article: https://siliconcanals.com/news/startups/nicolab-strokeviewer-ai-acute-diseases-treatment/
