The Gap — Why SmartShot Should Exist

The Structural Observation

Someone dies because the naloxone is across the room.

That sentence contains the entire argument for SmartShot. The medication exists. The diagnosis exists. The patient is wearing a sensor that knows exactly what is happening. And nothing happens, because the system that watches cannot act.

Remote Patient Monitoring is a $16 billion industry in the United States as of 2025 (MarketsandMarkets, 2025). It measures heart rate. It tracks SpO2. It logs respiratory rate, skin temperature, motion, sleep cycles. It transmits that data to a dashboard, where a clinician or an algorithm reviews it and decides whether to intervene. The intervention, when it comes, is a phone call. Or a text. Or an alert that routes to a nurse who is covering forty other patients.

RPM monitors. It does not intervene. Monitoring without intervention is observation, not care.

The gap is structural. RPM devices are built to detect and report, not to respond. The sensor fires. The alert routes. The human must be conscious, present, and fast enough. In the three scenarios that follow, the human is none of those things.

Scenario 1: Opioid Overdose, Alone in a Room

A 34-year-old man uses fentanyl alone in his apartment. His respiratory rate drops below 6 breaths per minute. His SpO2 falls below 85%. A wearable sensor, if he were wearing one, could detect both signals within 90 seconds. A naloxone auto-injector sits in his kitchen drawer. He is unconscious on his bathroom floor.

He dies.

This is not a rare event. In 2024, 54,045 Americans died from opioid overdoses (CDC NCHS, May 2025). That number represents a 27% decline from 2023 — a decline driven by fentanyl supply shifts, not by any improvement in the ability to reach people who use alone.

The critical variable is solitude. Studies consistently find that 39% to 83% of fatal opioid overdoses occur when the person is alone at the time of use (PMC, 2022). In one cohort, 75% of decedents were using alone. Even when bystanders are present, naloxone administration by a layperson occurs in only 4% of deaths involving illicit opioids (CDC MMWR, 2018).

Naloxone works. It reverses respiratory depression in minutes. It is available over the counter. It costs under $50. None of that matters if no conscious person is in the room.

A wearable device that detects respiratory depression and delivers subcutaneous naloxone without requiring a conscious operator would address the single largest failure mode in opioid overdose mortality: the absence of a bystander.

Scenario 2: Anaphylaxis in a Child, EpiPen in the Nurse's Office

A 15-year-old eats a granola bar at lunch. It contains tree nuts. She did not know. Her throat begins to close. Her heart rate spikes. Her blood pressure drops. She reaches for an auto-injector that is not in her backpack — it is locked in the school nurse's office, per district policy.

The average time from first anaphylaxis symptom to cardiac arrest is 14 minutes (University of Bristol / Clinical & Experimental Allergy, 2026). Fourteen minutes. That is less time than most school periods.

A University of Bristol study of 19 UK pediatric anaphylaxis deaths between 2019 and 2023 found that in nearly three-quarters of cases, the child received either no epinephrine or only a single dose before cardiac arrest. In 37% of deaths — 7 of 19 — the child did not have an auto-injector with them when the reaction began (NCMD, 2024).

Nearly 90% of fatal pediatric anaphylaxis cases occurred in children aged 10 to 17. Almost half were 15 to 17. These are teenagers who are old enough to carry a phone but whose school may prohibit them from carrying their own medication.

The United States records approximately 225 anaphylaxis deaths per year across all causes and age groups (Turner et al., Journal of Allergy and Clinical Immunology, 2014). Approximately 60% are drug-triggered; the remainder split between food and venom. The case fatality rate for anaphylaxis that reaches a hospital is 0.3%. The deaths cluster where the epinephrine does not reach the patient in time.

An EpiPen requires a conscious user who can identify the reaction, retrieve the device, remove the cap, and press it against their thigh. A wearable that detects the histamine cascade — heart rate spike, blood pressure drop, skin conductance change — and delivers intramuscular epinephrine without requiring the child to be conscious or to have the device in hand goes straight at the access delay that kills.

Scenario 3: Cardiac Event in a Rural Home, 40 Minutes from the ER

A 62-year-old woman in eastern Kentucky feels chest pressure at 2 a.m. She lives alone. The nearest emergency department is 38 miles away. She calls 911.

Median EMS response time in rural areas is 11.5 minutes, compared to 7.3 minutes in urban settings (PMC, 2025). A 2025 study from England found rural response times exceed urban response times by more than 3.5 minutes, with every additional kilometer from an ambulance station adding 37 seconds to arrival (NCBI, 2025).

For out-of-hospital cardiac arrest, survival decreases 5% to 12% for every minute of delay in treatment (Norwegian study, Clinical Cardiology, 2025). The overall survival-to-discharge rate for out-of-hospital cardiac arrest in the United States has been approximately 10% for two decades (Sudden Cardiac Arrest Foundation; AHA, 2024). Approximately 350,000 OHCA events occur annually. Roughly 315,000 of those people die.

Rural communities show higher rates of bystander CPR (49.6%) than urban areas (40.6%), yet survival rates remain lower (PMC, 2025). Bystander willingness is not the constraint. Distance is.

The Zoll LifeVest proves that autonomous wearable cardiac intervention is viable — it detects arrhythmia and delivers defibrillation without human input. But defibrillation is electrical therapy. For the patient whose cardiac event requires nitroglycerin, or whose arrhythmia presents as a medication-responsive episode rather than a shockable rhythm, no wearable can help. The device category ends at electricity.

A wearable that extends autonomous intervention from electrical therapy to pharmaceutical therapy — sublingual or subcutaneous nitroglycerin delivered upon detection of ST-segment changes or hemodynamic collapse — would fill the gap between the LifeVest and the ambulance that is 40 minutes away.

Why No One Has Shipped This

The gap is obvious. The three scenarios above are not obscure edge cases. They are common causes of death in the United States. So why has no device shipped that closes the loop from detection to medication delivery?

Five reasons. Each is real, and none is trivial.

1. Regulatory complexity. An autonomous drug-delivery wearable is a Class III medical device under FDA classification. It requires Premarket Approval (PMA), not 510(k) clearance. PMA demands well-controlled clinical trials demonstrating safety and effectiveness. The FDA review process alone takes 6 to 18 months after submission, and the clinical development preceding it can take years. Because the device combines a drug component and a device component, it is classified as a combination product, subject to review by both CDER and CDRH, with jurisdiction assigned by the Office of Combination Products based on primary mode of action (FDA.gov). No predicate device exists for most of SmartShot's target indications, which means no shortcut through substantial equivalence.

2. Liability vacuum. If an autonomous device administers naloxone to a patient who is not overdosing — who is, say, deeply asleep — who is liable? The device manufacturer? The prescribing physician? The algorithm developer? Good Samaritan statutes cover human bystanders administering naloxone. They do not clearly cover autonomous devices. Product liability for a device that makes a pharmacological decision without human confirmation is largely uncharted. No insurer has priced this risk, because no device has created it.

3. Miniaturization constraints. A wearable must be small enough to wear continuously, which means the drug reservoir, injection mechanism, sensor array, processor, battery, and wireless radio must fit in a form factor roughly the size of a continuous glucose monitor or insulin pump. Closed-loop insulin systems took decades of miniaturization to reach their current size, and they deliver a single drug at micro-doses. SmartShot's target drugs — naloxone, epinephrine, nitroglycerin — require larger single-dose volumes and faster delivery mechanisms.

4. Drug stability. Epinephrine degrades through oxidation, producing adrenochrome and losing potency. Current storage recommendations require controlled room temperature (20-25 degrees Celsius), with permitted excursions only between 15 and 30 degrees Celsius (USP standards). A body-worn device exposes medication to sustained skin-contact temperatures, humidity, and motion. A 2023 study in the Journal of Allergy and Clinical Immunology found that EpiPen and Symjepi showed rapid and extensive degradation under extreme temperature conditions, becoming subpotent within weeks. Naloxone is more thermally stable than epinephrine but still requires controlled storage. Maintaining drug potency in a wearable worn 24 hours a day, through showers and summer heat, is an unsolved engineering problem.

5. Sensor accuracy requirements. A false negative — failing to detect an overdose — is fatal. A false positive — injecting naloxone into a sleeping patient — causes precipitated withdrawal: vomiting, agitation, tachycardia. Neither error is acceptable. The detection algorithm must achieve sensitivity and specificity levels that exceed current wearable-grade sensors. Clinical-grade pulse oximetry is accurate to plus or minus 2%. Consumer wearables are accurate to plus or minus 5-8%. The margin between "deeply asleep" and "respiratory depression from opioids" is narrow enough that consumer-grade accuracy is insufficient.

What Would Need to Be True

For SmartShot to work — not as a concept, but as a device a physician would prescribe and a patient would wear — the following conditions must be met:

• Wearable-grade SpO2 and respiratory rate sensors must achieve clinical-grade accuracy (plus or minus 2%) in continuous, multi-day wear without skin irritation or signal degradation.
• A subcutaneous or intramuscular injection mechanism must be miniaturized to fit a body-worn form factor while delivering a full therapeutic dose (0.4mg naloxone, 0.3mg epinephrine) in under 10 seconds.
• Drug cartridges must maintain potency at skin-contact temperatures (32-37 degrees Celsius) for at least 14 days between replacements, validated through accelerated stability testing.
• The detection algorithm must demonstrate sensitivity above 95% and specificity above 98% for each target condition in controlled clinical trials — not bench testing, not simulated data, not retrospective chart review.
• The FDA must establish a regulatory pathway for autonomous combination products that deliver emergency medications without real-time physician confirmation. This may require a new guidance document or a novel use of the Breakthrough Device designation.
• Product liability insurance must be available at a price that does not make the device economically unviable. This likely requires either legislative safe harbor (extending Good Samaritan protections to autonomous devices) or sufficient clinical trial data to demonstrate a favorable risk-benefit ratio.
• Reimbursement pathways must exist. CMS does not currently have CPT codes for autonomous emergency medication delivery. New codes would need to be established, or existing RPM codes (99453, 99454, 99457, 99458) would need to be extended.

None of these conditions is impossible. Several are already partially met. But none is trivially met, and the list explains why this device does not yet exist despite the obvious need.

Positioning Against Existing Solutions

Four products occupy adjacent territory. None occupies SmartShot's.

EpiPen (and equivalents: Auvi-Q, generic epinephrine auto-injectors, neffy nasal spray). These deliver the right drug. They require a conscious user. The patient must recognize the reaction, retrieve the device, and activate it. In 37% of pediatric anaphylaxis deaths, the auto-injector was not with the patient. In three-quarters of deaths, epinephrine was either not administered or administered too late. The failure mode is not the drug. It is the human in the loop.

Remote Patient Monitoring (RPM devices: Biobeat, Masimo, Philips BioIntelliSense). These detect the right signals. They cannot deliver medication. RPM is observe-only by design. When a patient's vitals cross a critical threshold, the system generates an alert. The alert goes to a clinician. The clinician calls the patient. If the patient is unconscious, the clinician calls 911. If the patient is in a rural area, 911 response time is 11.5 minutes median. RPM watches people die in high resolution.

Closed-loop insulin delivery (Omnipod 5, Medtronic 780G, Tandem Control-IQ, Sequel twiist, iLet Bionic Pancreas). These are the closest analogue to SmartShot. They sense a biomarker (glucose), make an algorithmic decision, and deliver a drug (insulin) without human confirmation. They prove the concept. But they manage a single chronic condition with a single drug delivered in micro-doses over hours. The first closed-loop insulin system (Medtronic 670G) received FDA approval in 2016, after decades of development in glucose sensing, insulin formulation, and pump miniaturization. SmartShot's target conditions require bolus delivery of emergency medications — a fundamentally different dosing profile, a different risk calculus, and a different regulatory posture.

Zoll LifeVest (wearable cardioverter defibrillator). This device autonomously detects life-threatening arrhythmias and delivers defibrillation without human input. It has been worn by more than one million patients. It proves that autonomous wearable intervention in cardiac emergencies is clinically viable and regulatorily approvable. But the LifeVest delivers electrical therapy, not medication. For the subset of cardiac events that require pharmaceutical intervention rather than defibrillation — and for every non-cardiac emergency — the LifeVest model does not apply. SmartShot extends the LifeVest's structural logic from electrical therapy to pharmaceutical therapy.

The RPM-to-RMD Progression

The argument for SmartShot is not that it is a novel invention. It is that it is a structural inevitability.

Healthcare technology follows a consistent pattern: observe, then intervene. Diagnostic imaging preceded surgical robotics. Continuous glucose monitoring preceded automated insulin delivery. Remote cardiac telemetry preceded the implantable cardioverter-defibrillator. In every case, the monitoring technology matured first, created a data layer that clinicians trusted, and then the intervention technology followed — sometimes decades later.

RPM is the observation layer. It is mature. It is reimbursed. It is a $16 billion market growing at 12.6% annually (MarketsandMarkets, 2025). It has proven that wearable sensors can reliably detect the physiological signatures of cardiac events, respiratory depression, anaphylaxis, and hypoglycemia in real time.

The intervention layer has no commercial product.

Remote Medication Delivery — RMD — is the term SmartShot uses for that layer. The term is original to this project. RMD is not a product. It is a category. It describes any system that combines continuous biosensing, algorithmic detection, and autonomous medication delivery in a wearable form factor.

The progression from RPM to RMD mirrors the progression from CGM to closed-loop insulin delivery. That progression took roughly 15 years from the first reliable continuous glucose monitor (Medtronic CGMS, 1999) to the first hybrid closed-loop system (Medtronic 670G, 2016). The sensor matured. The algorithm matured. The delivery mechanism matured. Then they converged.

RPM sensors are mature now. The algorithms are advancing. The delivery mechanism is the missing piece. SmartShot is a design thesis for that missing piece.

I registered SmartShot because I watched this pattern from inside the healthcare system — years of building utilization management programs, designing care pathways, writing clinical criteria — and saw the same structural gap every time. We could identify the patients who were going to have emergencies. We could flag them. We could monitor them. And then we waited for the emergency to happen and hoped someone was nearby.

That is not a technology problem. It is a category problem. The category that solves it does not exist yet. SmartShot is an attempt to name it, scope it, and describe what it would take to build it.

The gap is four minutes. The medication exists. The sensor exists. The patient is dying. No shipping product connects them.

That is why SmartShot should exist.