HypnOS: nightmare-detection research, a Class II strategy and a remote-monitoring program
Chief Medical Technology Officer and first-named inventor at betweenMD: led the invention, intervention design and algorithm research for HypnOS, a planned prescription digital therapeutic designed to detect nightmare onset from wearable physiological signals and interrupt it with haptic feedback; prepared its Class II regulatory and clinical-study strategy; and built the reimbursement models for a remote-monitoring program.
Problem
HypnOS was planned as a single-purpose prescription device: an Apple Watch and iPhone kit that does one thing, detect a nightmare during sleep and interrupt it with a gentle haptic signal. The hard part is timing. A rise in heart rate and movement may appear only once the sleeper is already waking, too late to help. Sleep physiology also shifts through the night and from night to night, and people move in their sleep for many harmless reasons, so a single fixed threshold misfires.
The remote-monitoring program faced a different, more ordinary problem: in a clinic, remote monitoring fails if it does not pay for itself under the reimbursement rules, or if the devices do not fit the clinical workflow.
Architecture
I led the invention, the intervention design and the algorithm research and development for a detection approach that adapts throughout each night to the sleeper's own physiology, uses PPG-derived heart-rate variability as a stress signal, accommodates the low motion of REM sleep and periodic limb movements, and excludes the gross movements of trauma-associated sleep disorder and other REM parasomnias. In offline analysis of wearable recordings it flags stress events during still sleep in people with nightmare disorder as candidate nightmare onsets; detection performance was to be established in the planned study. On the intervention side, I designed the haptic feedback to adapt to the individual sleeper, intended to keep it effective as the sleeper habituates. The detection work was built first for Apple Watch and iPhone, then extended to the Biostrap wearable. The commercial plan was to bill HypnOS as durable medical equipment, following the reimbursement path of the first FDA-authorized device in its category, through a durable-medical-equipment partner. I also coordinated the wearable-platform evaluation, the clinical-study design and the Class II regulatory strategy needed to test it.
Planned next stages: add oxygen saturation and respiration to exclude apneic and non-REM events, so that a flagged event is more likely a nightmare; then use frequency-domain heart-rate variability to catch the autonomic shift toward sympathetic activation earlier.
Separately from HypnOS, I built financial models for remote patient and therapeutic monitoring that tie enrolment, monitoring time and billing codes to reimbursement, so a practice can see whether a program pays for itself before it starts.
What was built
- Nightmare-detection algorithm research from wearable signals: PPG waveform and inter-beat intervals, heart-rate variability and stress indices, with movement used to qualify an intervention rather than trigger it
- Wearable-platform evaluation (including Biostrap) for PPG and inter-beat-interval quality, with an ECG reference planned for validation
- Safety reasoning before any haptic intervention: normal REM movement, periodic limb movements and other parasomnias treated as confounds
- A draft pre-submission for a Class II 510(k) pathway with special controls and a randomized, sham-controlled study design; the draft was not submitted
- Named lead Principal Investigator on the SBIR proposal for the planned study
- Sleep-medicine and clinical-workflow requirements translated into technical specifications for data capture, signal processing and machine-learning development
- Clinic financial models for RPM and RTM: workbooks for a prospective behavioral-health (CBT) practice and a prospective respiratory/musculoskeletal practice. Inputs are the CMS remote-monitoring billing codes, minutes per monitored patient, staff cost by role (clinical psychologist, nurse practitioner, respiratory therapist, medical assistant), platform fee, patient compliance, payer acceptance and overhead; outputs are patients per staff member, billable revenue and margin at 25, 50 and 75 percent of staff time.
- Remote-monitoring program design around wearables, including device integration planning for Biostrap
- Backend APIs integrating Postgres and MongoDB for data extraction, transformation and loading
Verification
The detection approach ran offline on recorded wearable data; establishing its detection performance was the job of the planned study. The financial models were checked against the reimbursement rules they depend on, because a remote-monitoring program's first test is whether a practice can run it at all.
Evidence
- VERIFIED — PRIVATEHypnOS draft pre-submission (Q-Sub), sponsor betweenMD, not submittedheld
- VERIFIED — PRIVATEAlgorithm notebooks, 2022-11 to 2023-06, and signal-validation memoheld
- VERIFIED — PRIVATEWearable-platform evaluation deck and device matrixheld
- VERIFIED — PRIVATEU.S. provisional patent application 63/502,459, System and Method to Assess Sleep Disturbances (filed 2023-05-16; first-named inventor)USPTO Filing Receipt held
- VERIFIED — PRIVATERPM and RTM financial models — two workbooks (behavioral health; respiratory/musculoskeletal), 2022held
Status
Historical. First-named inventor, U.S. provisional patent application 63/502,459, System and Method to Assess Sleep Disturbances (2023).
Historical case. The draft pre-submission was not submitted; the notebooks, models and planning documents are held. · updated 2026-11-02