NEW YORK – Israel-based medical informatics firm Medial EarlySign is looking to expand its presence in the US market, targeting both healthcare providers and diagnostic firms with its machine learning-based tools for analyzing lab and patient data.
The company in January expanded its US leadership team, adding three new executives: Christopher Brown as chief growth officer, Darrell Atkin as vice president of marketing, and Adam Dooley as vice president of business development. It is also planning a funding round this year to support its expansion that will target at least $20 million, said CEO Jeremy Orr.
Medial currently has ongoing research partnerships with Geisinger Health System and Kaiser Permanente through which it is developing and testing its informatics tools. It also sells its services to the University of Alabama, Birmingham and St. Louis University and expects to close deals with several other healthcare providers this year.
Founded in 2013, Medial uses machine learning to analyze lab results, clinical data, and electronic health records to identify risk factors that might inform patient care. To date, much of its work has focused on lab data where its effort to add value above and beyond the raw test result fits with broader trends within the lab industry, particularly around the movement that has become known as Lab 2.0.
Orr noted that clinical labs have in the last few decades become seen as providing something of a commodity. In recent years, downward pressure on test reimbursement driven by both government and private payors have led some in the industry to explore how they might generate greater value from lab data, using it to, for instance, guide population health efforts.
"They desperately want to get back to providing insights that are much closer to clinical care, and they are all trying to develop, one way or another, better clinical decision support and augmented interpretation of lab results that actually helps clinicians make better decisions for individual patients," Orr said.
Several leading hospital systems and lab companies, such as Long Island's Northwell Health, New Mexico-based TriCore Reference Laboratories, and Geisinger are working to develop such solutions in house, using lab results to, for instance, identify gaps in patient care. Medial believes it can offer additional information and risk prediction by using machine learning to analyze lab data.
The company's lead product is its LGI-Flag tool, which uses results from a standard complete blood count to detect patients likely to have a gastrointestinal condition causing occult bleeding, such as an ulcer or colorectal cancer. Because it is based entirely on CBC results, the tool allows providers to, for instance, identify at-risk patients to prioritize for colonoscopy without requiring any additional testing, Orr noted.
Medial has also developed a product for identifying unvaccinated patients at high risk of developing serious complications from the flu that it has implemented with Israeli health system Maccabi Healthcare Services and with Kaiser. Additionally, it is working with Geisinger to implement a tool for identifying pre-diabetic patients who are likely to become diabetic in the next year.
"The thing [these conditions] have in common is that they are all relatively common, expensive conditions, where being early [with treatment] matters," Orr said.
Thus far the company has primarily worked with healthcare providers, which Orr noted have access to the patient lab data most of its tools are based on. It is also currently working with some large in vitro diagnostic instrument manufacturers with the aim of embedding some of its treatment algorithms into their machines.
"We hope that you will see some announcements later this year about some multinational [IVD] manufacturers that want to put our algorithms in their machines and distribute them that way," Orr said.
In the case of the company's LGI-Flag tool, for instance, the algorithm underlying that tool could be embedded in a hematology analyzer so that it would be automatically run as part of every CBC test for a particular group of patients and flag any results that indicate high risk of a GI issue.
"The provider would get that blood count report and then there would be an additional line of interpretation based on the additional machine leaning analysis saying that this patient is considered particularly high risk and this is the recommended action," Orr said.
He noted that such an approach could prove an effective way to drive uptake of the company's technology.
"Some of the diagnostics companies are multinational and have massive distribution with these machines that they are selling and installing all over the world," he said. "If we can put it in a machine and take a small fee per calculation times thousands and thousands and thousands of calculations everywhere the machines are implemented, that is a nice business model."
This will likely involve more regulatory scrutiny of the company's tools than they have received thus far. Orr said. He said that as currently sold, Medial's products are considered low-risk clinical decision support and are lightly regulated, but he added that it expects "that heavier regulation is likely coming to our space."
"You're starting to see more scrutiny of machine learning algorithms in every human endeavor, and especially clinical medicine, where the [US Food and Drug Administration] has suggested they are going to require explainability, studies showing efficacy," he said. "There will probably be a regulatory pathway that is more like a typical medical device approval process, and we are ready for that."
In addition to working with providers and IVD manufacturers, Medial is pursuing relationships with payors, though Orr noted that the company has less experience working with medical claims data, which is what payors primarily have access to, rather than lab data, which is typically housed with providers.
It is working to improve its use of claims data, Orr said, citing its participation with Geisinger in the Centers for Medicare and Medicaid Services (CMS) Artificial Intelligence Health Outcomes Challenge, which calls for participants to apply AI to Medicare claims data to predict unplanned hospital admissions and adverse events. In November, the pair was selected as one of 25 participants to advance to the first stage of the challenge.
Generally speaking, though, the claims analysis space is a crowded one with a number of well-established players, Orr said.
Medial currently has 35 employees, 28 in Israel and seven in the US. The company has raised $50 million since its launch.