Epitel - From Research to $26 Million Series B

Epitel just closed a $26 million Series B to expand remote wireless EEG monitoring, and in this episode Mark Lehmkuhle, the company's CTO and founder, and J Carroll, Director of Software Engineering and Architecture, walk DeAndre Harakas and Grant Chapman through how they got here, which started in a rodent lab at the University of Utah and ended up as a wearable sensor that people are supposed to want to wear for days at a time. Mark describes Epitel as a machine learning company for the detection of neurological disorders that just so happens to make its own wireless wearable EEG sensor, a little chicklet with two gold electrodes that attaches to the scalp with a one piece sticker that is both adhesive and conductive hydrogel, so no goopy gels. The hardware, as he puts it, is the means to the end, because the real product is turning an enormous stream of data into something clinically actionable. From there the conversation gets into the electrode work, including recreating a study from the early 1970s to find out how few electrodes you actually need, which came down to four locations, two at the corners of the forehead and two behind the ears, and then a second round of work on how close together those electrodes can sit before the signal stops being useful. Mark makes the point that self-reporting, which is still the primary outcome measure in a lot of clinical trials, is maybe fifty-fifty at best, because people genuinely do not know when they have had a seizure, and a lot of them happen at night. The back half is about the part Mark says is actually hard, which is not the tech stack, it is human factors. Their population runs from neonates to people in their 90s, seizure forecasting raises the question of what a 40 percent chance even means to a given person, and none of the signal quality matters if somebody will not wear the thing. J talks about garbage in and garbage out, and why knowing your data is bad is a far better situation than having a thousand patients and no idea which recordings to trust. Grant and DeAndre get into human factors studies you have to fail once to take seriously, the lily pad analogy for why small companies cannot jump the whole pond in one go, why the first FDA clearance was a milestone rather than a sellable product, and why a connected product is never really done as long as it is supported. Chapters 0:00 Welcome and where we are in season three 1:34 J Carroll on an eclectic path into medical devices 2:47 Mark Lehmkuhle, CTO and founder 3:41 What Epitel actually builds 4:24 Hardware as the necessary evil 5:51 The $26 million Series B, and rewinding to the start 6:30 The NIH screening program at the University of Utah 7:52 Borrowing the idea from a tire pressure monitor 8:43 J's years in a rat research lab 10:52 What academia can risk that product development cannot 11:34 How many iterations it took 12:39 Four days, nine hundred dollars, and a drive to Boston 18:30 How few electrodes do you actually need 20:39 How close together can they sit 20:52 Why worse data gets the job done 22:12 Pulling the models back to the edge 23:14 J turns a question back on Mark 24:51 Why self-reporting is fifty-fifty at best 27:23 The tech stack is tractable, human factors are not 28:14 Does a forecast change the outcome it predicts 28:39 Garbage in, garbage out 30:10 Designing for messy, real-world humans 31:38 Watching session health from the cloud 32:54 You have to get burned by a human factors study once 34:24 The way you do anything is the way you do everything 36:28 Mark solved a human factors problem, not a technical one 37:39 Every problem you solve reveals the next one 39:31 The Delta touch faucet and the Thanksgiving turkey 41:32 The next weak link, from patient to practice 42:27 Cybersecurity as an ever moving target 44:55 A connected product is never done 47:47 Ship or die, and the first iPhone had no App Store 49:07 Four engineers, an FDA clearance, and a product they couldn't sell 51:20 The lily pad analogy 52:57 Would you do it again knowing what you know now 54:17 Ignorance as an underrated asset 55:27 DeAndre on walking up a mountain three steps at a time Learn more about Epitel: https://epitel.com Learn more about Glassboard: https://glassboard.com #hardtech #hardware #medicaldevice #EEG #epilepsy #machinelearning #startups #productdevelopment #IoT #humanfactors