By merging a dry electrode brain wave wearable with a home sleep testing distribution channel, Beacon Biosignals is trying to build a platform that monetizes the same hardware through clinical diagnostics and pharmaceutical drug trial contracts.
A private neurotech company is running the same wearable brain-wave hardware in sleep labs and inside late-stage drug trials. The bet is that blurring those two markets is the business model, not just the product.
Boston-based Beacon Biosignals has spent the last several years building an artificial-intelligence scoring layer on top of electroencephalography, or EEG, the decades-old technique of recording electrical activity from the scalp. EEG is the same signal doctors read in a hospital neurology ward, but Beacon's version is captured by a soft, dry-electrode headband the company calls Waveband, designed to be put on by a patient at home rather than a technician in a clinic. The device has FDA 510(k) clearance, the regulatory pathway that lets a medical device reach the U.S. market once it has been shown to be substantially equivalent to an already-cleared product.
What changed this month is that Beacon folded CleveMed into that strategy. CleveMed, founded in Cleveland in 2000, built one of the larger home sleep testing, or HST, footprints in the United States, with its SleepView device and a national provider network used by clinicians to diagnose obstructive sleep apnea without sending patients to a sleep lab. Beacon completed a strategic transaction to acquire CleveMed, and as of June 2026 the CleveMed and SleepView brands are being retired into the Beacon name, with former CleveMed CEO Hani Kayyali staying on to run Beacon's sleep-testing operations. The deal gives Beacon a hardware distribution channel and a clinician customer base it did not have before, and it gives Beacon's existing AI-EEG platform a regulated clinical use case that already pays.
That last point is the structural move worth analysis. Until now Beacon's Waveband headband lived mostly inside pharmaceutical clinical trials, where sponsors pay to use it as an objective endpoint, a quantitative measurement that can stand in for a subjective symptom score, in studies of central nervous system, or CNS, drugs. A December 2025 partnership with Harmony Biosciences made that contract explicit: Beacon is supplying quantitative EEG endpoints for Harmony's Phase 3 work in hypersomnia conditions, the cluster of disorders, including narcolepsy and idiopathic hypersomnia, defined by excessive daytime sleepiness. Beacon is presenting AI-driven sleep research at SLEEP 2026, the sleep medicine field's annual scientific meeting, including work tied to that hypersomnia program.
The reuse logic is straightforward on paper. A sleep clinic gets a dry-electrode headband in the mail, the patient wears it overnight, the device streams EEG to Beacon's cloud, Beacon's models score the recording, and a clinician reads a report. The same headband, used under a research agreement with a drug sponsor, becomes an objective endpoint inside a Phase 2 or Phase 3 trial, where regulators and payers increasingly want quantitative measures of drug effect rather than patient-reported sleepiness scales. In both cases the underlying hardware, the scoring pipeline, and the regulatory clearance are the same. Beacon is essentially trying to monetize one device twice: once through clinical sleep-diagnostics reimbursement, the payment doctors and insurers make for a covered medical test, and once through CNS drug-trial endpoint contracts, where the customer is a pharmaceutical company.
That is a platform bet, not just a product bet. Platform companies in medical devices historically try to lock in distribution, then add adjacent use cases on top. Beacon is doing the same thing in reverse: it built the adjacent use case first, by selling EEG endpoints into CNS trials, and is now adding the distribution layer, by acquiring CleveMed's HST channel, to convert adjacent credibility into clinical volume. The $86 million funding round Beacon raised to accelerate AI-driven brain-health work, as reported by BioSpace, underwrites the conversion.
The dual-market bet has independent support, though it is not without skeptics. Dr. Bhavin R. Sheth, an associate professor of electrical and computer engineering at the University of Houston, has published research demonstrating that AI-based sleep staging can reach expert-level agreement with polysomnography, the in-lab gold standard, using simplified at-home hardware. Sheth's team showed in a 2024 study published in Computers in Biology and Medicine that a deep-learning model trained on 4,000 recordings "significantly outperforms current research and commercial devices" and achieves accuracy on par with clinician-scored polysomnography. His conclusion: AI-driven sleep analysis is on a credible path toward replacing the in-lab standard for many applications. That alignment matters for Beacon's thesis: if AI-EEG scoring can be validated against the gold standard in research settings, the same validation pathway that supports CNS trial endpoints could, in principle, support clinical HST adoption.
The practical obstacles, however, are real, according to sleep-technology experts. Todd M. Eiken, RPSGT, FAAST, vice president of product development at Dymedix Diagnostics, notes that EEG signals in real-world settings face fundamental challenges. "We're talking about trying to measure voltages that are really super small—in the microvolt range," Eiken says. "Everyday happenings like outside sounds and patient movement can make the signal unintelligible. The contact between EEG electrodes and skin must be pristine. Otherwise, it's just going to be noise." Eiken's point cuts to the core challenge for any dry-electrode home EEG system: clinical-lab conditions are controlled, but a patient's bedroom is not.
Richard Kaplan, PhD, president at General Sleep Corp, which makes Zmachine home sleep monitoring products, offers a related caution. "The challenge is to have an EEG system that is easy for the patient to apply without compromising robust clinical data about sleep stages—and all of this needs to happen without intimidating or confusing the patient," Kaplan says. The gap between a technical demonstration and a patient-friendly clinical product remains wide, he suggests, and ease of use is a separate problem from algorithmic accuracy.
There are honest limits to how far this story should be pushed. Beacon is a private company and does not disclose revenue, hardware shipment counts, or how many of its reported "40-plus trials" are commercially sponsored versus academic. Its marketing language describes Waveband scoring as reaching "expert-level accuracy" and as approaching equivalence with polysomnography, but the press materials reviewed here do not cite a peer-reviewed validation study establishing that claim. The CleveMed deal has closed and the brands are unified, but Sleep Review's coverage and CleveMed's own announcement frame the move in distribution terms rather than in clinical-outcome terms, which is the right level of caution for now.
What to watch next is whether the same scoring model that wins CNS endpoints also wins routine HST referrals. If it does, the two-market bet becomes a single flywheel: pharma contracts fund the model, clinical volume funds the hardware, and each side validates the other for the next customer. If it does not, Beacon is left running two thin businesses on one piece of hardware, with the brand unification serving mostly as a marketing convenience.