The real life Dr. Shepherds – yes, this is a reference to the popular medical drama “Grey’s Anatomy” – who work in hospitals around the world may soon get a new assistant. No, not just another intern but an extra pair of virtual eyes to help them better diagnose stroke victims.
Tel Aviv-based MedyMatch Technology Ltd., which hopes to have its first commercially available product as soon as the first half of 2017, is developing an artificial intelligence (AI) platform for critical areas of patient care. The platform is meant to help study data more quickly and accurately than the human eye, and help physicians with their clinical decisions in a wide set of healthcare issues.
MedyMatch’s first area of focus will be for stroke patients.
“Speed is essential in treating stroke,” Gene Saragnese, MedyMatch chairman & CEO, said in an interview. “For every minute that ticks by, brain cells die.”
When dealing with stroke, the very first question doctors need to address is what kind of stroke they are seeing: is it a bleed in the brain or a blockage that prevents blood flowing to the brain? The two types of stroke are treated in a very different manner. A wrong diagnosis and treatment could mean the death of much-needed brain cells.
“Our goal is to make better decisions in that very first step in a stroke – so those patients can get moved into the appropriate treatment quickly,” Saragnese said.
The product is software that takes images from a standard CT machine and processes them in the cloud, using proprietary algorithms developed by MedyMatch. The software makes notes on the image, highlights areas for physicians so they can immediately see potential areas of bleeding – and sends the image back to the doctor’s workstation, together with the original.
The process is expected to enable doctors to get an expert opinion within three to five minutes, Saragnese said. With deep learning a series of examples are fed to the computer to set benchmarks for what is considered a baseline reading, said Saragnese, a former CEO of Philips Imaging Systems who joined MedyMatch as CEO in February. Then you upload a whole series of images to the computer – and the machine learns what a bleed looks like from those images. Basically, “you are training the computer with examples in a way that it can then start to read images by itself,” he said.
MedyMatch has secured billions of images from millions of cases via collaborations with hospitals in Israel and the US, including Hadassah Medical Center in Jerusalem and Massachusetts General Hospital in Boston. “Those places are where our experts are coming from. They help us train the software to read,” Saragnese said.
Stroke is the fourth-biggest killer in the US, and costs to treat stroke may increase from $71.6 billion in 2010 to about $183 billion by 2030, according to data published by the American Heart Association. Despite advances in medical imaging, the medical misdiagnosis error rate of around 30 percent rate has not changed for decades, said Dr. Gabriel Polliack, a director of strategy development at TEREM, a network of emergency medical centers in Israel.
“There is a need in the marketplace to provide radiologists and physicians a second set of eyes to help them overcome any limitation that is preventing them from providing a correct patient diagnosis,” said Polliack said. Polliack is a member of MedyMatch’s medical advisory board and has been actively advising the company since its founding over two years ago.
“The idea is great, and not only has significant clinical value, meaning improvement of patient outcomes, but it will have a direct impact on the cost of care,” Polliack said. “This means that MedyMatch is addressing the holy grail in the medical profession, that is, providing better patient outcomes while lowering the costs.”
MedyMatch is in the process of raising an additional $8 million in funding, after it completed an initial financing round of $2 million earlier this year, Saragnese said. Its product will need to get FDA and other approvals.
“Stroke is clearly a global problem and a global opportunity for the company,” he said. “There are people doing machine learning in images, but it is for cancers. There is really no one working in this particular space of stroke right now, so there are not many competitors.”
Deals involving artificial intelligence-focused startups have increased almost sevenfold over the last five years, from just four deals in the first quarter of 2011 to 27 in the first quarter of this year, according to CB Insights, a New York based data firm. Around 15 percent of the deals in the first quarter went to companies with a focus on AI applications in healthcare, the data firm said.
AI startups have raised an aggregate $967 million in funding since 2010, with investments going to companies in 13 countries and 10 industry categories, including business intelligence, e-commerce, and healthcare, Deepashri Varadharajan, a tech industry analyst at CB Insights said by email.
“Specifically, in healthcare, AI is being used to leverage huge amounts of medical data to predict risk and make diagnosis more accurate,” Varadharajan said.
There are, however, challenges ahead for MedyMatch, a central one being making sure there is a valid cloud infrastructure in place to move images from the hospital to the cloud and back, Saragnese said. This infrastructure is the focus of giants such as IBM, General Electric and Philips, he said. “These are companies that are all developing cloud infrastructure that we expect we will be able to leverage.”
Another challenge could be overcoming the threat physicians may feel from the technology. “The technology looks interesting as long as it is used as an additional tool to extend your level of confidence,” Dr. Guy Raphaeli, a stroke neurologist and interventional neuroradiologist at the Rabin Medical Center in Petah Tikva, Israel said by phone. Raphaeli did not know about MedyMatch before he was approached by The Times of Israel for his opinion.
“It may also be useful for rural and remote hospitals in which there is a shortage of expertise. But I don’t believe a computer can ever replace the clinical skills of a physician, who can touch and understand the patient and integrate the whole medical picture,” Raphaeli said. “I don’t believe we need this technology here in Israel because we work at a high level of confidence and many of the hospitals are interconnected, so doctors can ask each other for assistance when needed.”
Indeed, customers of the product will not necessarily be the big general hospitals, where there is no lack of experts, but smaller rural or community hospitals, where the workers may be less experienced, Saragnese said. “Imagine you are in a rural hospital in China, that could be a tool to help the less experienced reader. Those are the target hospitals,” he said.
There are a number of revenue models the company is looking at, one of them being a subscription-based model, Saragnese said. “It would be less than $10 every time it is used, so very small.”