Autonomous Closed-Loop Bioelectronic Wound Dressings for Chronic Wound Therapy

1. Problem Statement

Chronic wounds — defined as wounds that fail to progress through the normal healing cascade within 30 days — affect 8.2 million Medicare beneficiaries annually in the United States and impose a direct treatment cost of $28 billion per year on the Medicare system alone (Nussbaum SR et al., Value in Health, 2018;21(1):27-32). The three primary chronic wound types — diabetic foot ulcers (DFUs), venous leg ulcers (VLUs), and pressure injuries — share a common pathophysiology: persistent inflammation, bacterial biofilm colonization, and impaired angiogenesis that stall the wound in the inflammatory phase indefinitely.

Diabetic foot ulcers represent the most economically devastating subset. An estimated 1.6 million new DFUs occur annually in the United States, with 19-34% of the 38.4 million Americans living with diabetes developing a DFU in their lifetime. The recurrence rate is 65% at 3-5 years. One in six DFUs leads to amputation, and DFUs account for 83% of all significant lower-limb amputations. Five-year mortality following major amputation ranges from 50% to 70% — comparable to many cancers. The average Medicare cost per DFU episode that heals primarily is $4,830, but costs escalate rapidly with complications: $13,580 for a single minor amputation, $31,835 for multiple minor amputations, and $73,813 for a major amputation. Total national DFU care costs range from $6.2 to $6.9 billion annually.

The current standard of care relies on periodic clinical assessment — typically weekly or biweekly office visits — where a clinician visually inspects the wound, subjectively estimates healing progress, and adjusts treatment. Between visits, wound deterioration goes undetected. Bacterial infection, the primary driver of chronic wound failure, produces measurable biochemical changes (elevated nitric oxide, hydrogen peroxide, pH shift, temperature increase) 1-3 days before clinical symptoms become visible. By the time redness, swelling, or purulent discharge prompts a patient to seek care, the infection has established and treatment options narrow. A system that continuously monitors wound biochemistry, detects early infection signatures, and autonomously intervenes — without waiting for the next scheduled clinical visit — would fundamentally change the trajectory of chronic wound management.

2. State of the Art

Four research programs have converged to make autonomous closed-loop wound therapy technically feasible, though none has yet produced a deployable commercial system that combines all three required capabilities: multiplexed biosensing, ML-based classification, and autonomous therapeutic intervention.

Multiplexed sensing with closed-loop treatment (Caltech). Wei Gao's laboratory at the California Institute of Technology developed a stretchable wireless bioelectronic system that simultaneously monitors uric acid, lactate, pH, and temperature in the wound bed while delivering combination therapy — antibiotic drug release and electrical stimulation — through the same flexible substrate. Published in Science Advances in 2023 (Shirzaei Sani et al., 9(12):eadf7388), the system demonstrated substantially accelerated wound healing in a rodent chronic wound model using closed-loop combination therapy versus drug-only, stimulation-only, or untreated controls. In April 2025, Gao's group published a second-generation system called iCares in Science Translational Medicine (Wang et al., 17:eadt0882), testing a microfluidic wound monitoring platform on 20 human patients with diabetic foot ulcers and venous leg ulcers. The iCares device measures six biomarkers — nitric oxide, hydrogen peroxide, oxygen, pH, temperature, and wound exudate volume — through a nanoengineered sensor array integrated with three microfluidic modules. An ML algorithm classifies wound severity and predicts healing time with accuracy comparable to expert clinician assessment. The 2025 human study validated monitoring and classification but did not test closed-loop drug delivery in humans.

Wireless closed-loop sensing and stimulation (Stanford). Yuanwen Jiang, Zhenan Bao, and Geoffrey Gurtner at Stanford University developed a wireless closed-loop smart bandage that continuously monitors skin impedance and temperature, then delivers electrical stimulation to accelerate healing. Published in Nature Biotechnology in 2023 (41(5):652-662, PMID 36424488), the system demonstrated 25% faster wound closure and 50% enhanced dermal remodeling versus untreated controls in a mouse wound model. The bandage uses a switchable hydrogel adhesion mechanism for on-demand attachment and detachment, and transcriptomic analysis confirmed activation of proregenerative gene programs in immune cell populations. Gurtner subsequently moved to the University of Arizona, where his group enrolled 83 patients in a clinical trial using a simplified sensor bandage (9 commercially available sensors). Preliminary results — presented but not yet peer-reviewed — report 90% positive predictive value for wound complications using only 2 of the 9 sensors.

ML-driven closed-loop bioelectronic therapy (UC Santa Cruz). Marco Rolandi's laboratory at UC Santa Cruz, in collaboration with UC Davis and Tufts University, developed the a-Heal (adaptive Heal) platform under the DARPA Bioelectronics for Tissue Regeneration (BETR) program. Published in npj Biomedical Innovations in 2025 (DOI: 10.1038/s44385-025-00038-6), a-Heal integrates an onboard camera that captures wound images every 2 hours, an ML model they term an "AI physician" that diagnoses wound stage, and a bioelectronic actuator that delivers fluoxetine (an SSRI that modulates wound serotonin levels to reduce inflammation) plus electrical field therapy. Preclinical testing showed approximately 25% faster healing versus standard care. The ML model runs on a nearby computer rather than on the device itself.

The gap between these research systems and a deployable product is fourfold: (a) no existing system runs ML inference on-device — all require an external smartphone or computer; (b) closed-loop drug delivery has been demonstrated only in animal models, not in human patients; (c) all prototypes are laboratory-fabricated with no design-for-manufacturing analysis; and (d) no group has engaged with FDA on a regulatory strategy for a combination product (device plus drug) with an adaptive ML algorithm.

3. Foundational Research

Shirzaei Sani E, Xu C, Wang C, Song Y, Min J, Tu J, Solomon SA, Li J, Banks JL, Armstrong DG, Gao W (2023). "A stretchable wireless wearable bioelectronic system for multiplexed monitoring and combination treatment of infected chronic wounds." Science Advances, 9(12):eadf7388. DOI: 10.1126/sciadv.adf7388. PMID: 36961905. Developed at Caltech with funding from NIH (R01HL155815), NSF, Army Research Office, and Heritage Medical Research Institute. The system integrates electrochemical biosensors for uric acid, lactate, pH, and temperature on a stretchable, skin-conformal flexible substrate. Drug release targets bacterial infection elimination and immune response regulation; electrical stimulation upregulates ion channels and accelerates cell migration. In a rat chronic infected wound model, combination therapy (drug plus stimulation, n=4 groups) produced substantially accelerated healing compared to single-modality controls. The study established that dual-modality closed-loop therapy — simultaneous antimicrobial drug release and proregenerative electrical stimulation — outperforms either modality alone, providing the mechanistic rationale for integrated treatment platforms.

Wang C, Fan K, Shirzaei Sani E, Lasalde-Ramirez JA, Heng W, Min J, Solomon SA, Wang M, Li J, Han H, Kim G, Shin S, Seder A, Shih CD, Armstrong DG, Gao W (2025). "A microfluidic wearable device for wound exudate management and analysis in human chronic wounds." Science Translational Medicine, 17:eadt0882. DOI: 10.1126/scitranslmed.adt0882. First human clinical validation of multiplexed smart wound monitoring. Tested on 20 patients with chronic wounds including diabetic foot ulcers and venous leg ulcers at wound care clinics. The device measures nitric oxide (inflammation marker), hydrogen peroxide (infection marker), dissolved oxygen (tissue oxygenation), pH, and temperature through nanoengineered electrochemical sensors integrated into a 3D-printed biocompatible polymer strip. Three microfluidic modules handle wound fluid extraction via a semipermeable membrane, bioinspired shuttle transport, and micropillar drainage. The ML classification algorithm stratified wound severity and predicted healing trajectory with accuracy comparable to expert clinician assessment. The sensor array is disposable (single-use) while the wireless PCB is reusable, keeping per-use material costs low. This study bridged the gap from animal models to human chronic wounds for the sensing and classification components, but did not test autonomous drug delivery in human subjects.

Jiang Y, Trotsyuk AA, Niu S, Henn D, Chen K, Shih CC, Larson MR, Mermin-Bunnell AM, Mittal S, Lai JC, Saberi A, Beard E, Jing S, Zhong D, Steele SR, Sun K, Jain T, Zhao E, Neimeth CR, Viana WG, Tang J, Sivaraj D, Padmanabhan J, Rodrigues M, Perrault DP, Chattopadhyay A, Maan ZN, Leeolou MC, Bonham CA, Kwon SH, Kussie HC, Fischer KS, Gurusankar G, Liang K, Zhang K, Nag R, Snyder MP, Januszyk M, Gurtner GC, Bao Z (2023). "Wireless, closed-loop, smart bandage with integrated sensors and stimulators for advanced wound care and accelerated healing." Nature Biotechnology, 41(5):652-662. DOI: 10.1038/s41587-022-01528-3. PMID: 36424488. Developed at Stanford University. The smart bandage monitors skin impedance (correlating with wound closure) and temperature wirelessly, then delivers electrical stimulation through a flexible electrode array when sensor readings indicate suboptimal healing. In a mouse excisional wound model, treated wounds closed approximately 25% faster than untreated controls and showed 50% enhanced dermal remodeling based on histological analysis. Transcriptomic profiling revealed that electrical stimulation activated proregenerative M2 macrophage gene expression in wound-resident immune cells — providing the first molecular evidence that closed-loop bioelectronic intervention reprograms the wound immune microenvironment. The bandage uses switchable hydrogel adhesion for on-demand attachment and removal without traumatizing the wound bed. This work established two foundational principles: (1) skin impedance serves as a reliable closed-loop feedback signal for healing progression, and (2) electrically modulated immune cell polarization is a viable therapeutic mechanism for chronic wound acceleration.

Rolandi M, Teodorescu M, Gomez M, Zhao M, Isseroff RR (2025). "Towards adaptive bioelectronic wound therapy with integrated real-time diagnostics and machine learning-driven closed-loop control." npj Biomedical Innovations. DOI: 10.1038/s44385-025-00038-6. Developed at UC Santa Cruz and UC Davis under DARPA BETR funding (up to $16 million collaboration). The a-Heal platform captures wound images every 2 hours via an onboard camera. An ML model — termed the "AI physician" — classifies wound healing stage from the images and prescribes treatment: fluoxetine delivery (to modulate wound serotonin and reduce inflammation) and electric field therapy (to promote cell migration). Preclinical testing demonstrated approximately 25% faster wound healing compared to standard wound care. The system represents the first published implementation of an ML-driven treatment decision loop for wound therapy, though the ML model runs on an external computer rather than on-device. Fluoxetine selection as the therapeutic agent was based on prior work showing serotonin plays a regulatory role in all four wound healing phases — hemostasis, inflammation, proliferation, and remodeling — making it a single-drug intervention point for multi-phase wound acceleration.

Nussbaum SR, Carter MJ, Fife CE, DaVanzo JE, Haught R, Nusgart M, Cartwright D (2018). "An Economic Evaluation of the Impact, Cost, and Medicare Policy Implications of Chronic Nonhealing Wounds." Value in Health, 21(1):27-32. DOI: 10.1016/j.jval.2017.07.007. The seminal Medicare chronic wound burden analysis, conducted using 2014 Medicare fee-for-service claims data for 8.2 million beneficiaries (one in six Medicare beneficiaries). Total annual Medicare spending attributable to chronic wound care ranged from $28.1 to $96.8 billion depending on methodology (conservative to all-inclusive). Surgical wounds with complications accounted for the largest share ($11.0-$15.8 billion), followed by diabetic foot ulcers ($6.2-$6.9 billion) and pressure injuries ($3.3-$5.4 billion). This study established the economic evidence base that justifies investment in wound care technology: the magnitude of spending creates strong payer incentive for interventions that accelerate healing and prevent costly complications, particularly amputations.

4. Competitive Landscape

No commercial entity currently offers a product that simultaneously monitors wound biomarkers, runs ML-based classification, and autonomously delivers drug therapy or electrical stimulation in a closed loop. Every existing product addresses one or at most two of these functions.

Vomaris Innovations (acquired by Arthrex). Tempe, Arizona. Funded $13.2-$16.2 million across four rounds. Products include Procellera, JumpStart, and PowerHeal — bioelectric antimicrobial wound dressings that use silver-zinc microcell batteries on a polyester substrate to generate microcurrents when activated by wound moisture. FDA 510(k) cleared as antimicrobial dressings (K160783, K180533). Commercially available through Arthrex distribution. Limitation: passive bioelectric effect with no sensors, no monitoring, no feedback loop, and no drug delivery capability.

Grapheal. Grenoble, France (CNRS Institut Neel spin-off, founded 2019). Raised EUR 1.9 million in seed funding. Developed WoundLAB — a graphene-on-polymer smart bandage that detects wound pH and provides electrostimulation via NFC smartphone connection. Currently in pilot clinical trials with diabetic patients. Limitation: monitoring plus stimulation only; no autonomous drug delivery, no on-device ML, no closed-loop operation.

Accel-Heal. United Kingdom. Manufactures a disposable electrical stimulation device for hard-to-heal wounds using a 12-day microcurrent treatment protocol. CE marked (Class IIa). Limitation: treatment only with no sensing, no monitoring, and no adaptive dosing.

MolecuLight. Toronto, Canada. Received FDA De Novo clearance in 2018 for the i:X handheld fluorescence imaging device that visualizes bacterial burden in wounds. Validated in a 350-patient, 14-site clinical trial demonstrating 3-fold increase in sensitivity for detecting bacterial loads exceeding 10^4 CFU/g compared to clinical signs and symptoms alone. Limitation: a diagnostic imaging tool used during clinical visits, not a wearable dressing. No treatment capability, no continuous monitoring.

ConvaTec. Global wound care incumbent ($2+ billion revenue). Launched ConvaNiox in April 2025 — a nitric oxide-generating wound dressing for diabetic foot ulcers approved as a Class III device in the EU. Clinical data showed 63% improvement in healing at 24 weeks versus standard care. Limitation: passive chemical release with no sensors, no feedback, and no adaptive dosing. ConvaNiox delivers a fixed NO dose regardless of wound state.

The absence of direct competitors reflects structural barriers that incumbents face. First, a closed-loop smart wound dressing is a Class III combination product requiring PMA — a multi-year, $50-100 million regulatory pathway fundamentally different from the 510(k) submissions that wound care companies file for conventional dressings. Second, incumbents manufacture in high-volume roll-goods facilities optimized for nonwoven fabrics and foams; flexible bioelectronics requires printed electronics, microfluidics, and PCB assembly — entirely different manufacturing infrastructure. Third, smart dressings that accelerate healing would cannibalize the core consumables business model of companies that profit from repeated dressing changes. Fourth, no established CPT or HCPCS code exists for autonomous wound monitoring with integrated treatment, creating reimbursement uncertainty that discourages large incumbents from committing R&D capital.

5. Total Addressable Market

Bottom-up calculation (US chronic wound treatment):

Medicare beneficiaries with chronic wounds number 8.2 million annually (Nussbaum et al., 2018). Annual Medicare spending attributable to chronic wound management ranges from $28.1 billion (conservative) to $96.8 billion (all-inclusive). Using the conservative estimate and focusing on the subset of patients whose outcomes would improve with continuous autonomous monitoring:

• Patients with DFUs: 1.6 million new cases per year (American Diabetes Association)
• Average annual per-patient cost: $33,000 (all Medicare services, Nussbaum et al.)
• Reduction in amputation-related costs from early infection detection (1-3 days before clinical symptoms): estimated 30-50% of amputation costs avoided
• Amputation cost range: $13,580 (minor) to $73,813 (major)
• Amputations attributable to DFUs: approximately 100,000 per year
• Total avoidable amputation cost: $1.4-$7.4 billion annually

Autonomous smart wound dressing as a 30-day prescription device at $500 per unit (disposable sensor array plus reusable PCB), prescribed for high-risk DFU patients (estimated 400,000 per year based on Wagner Grade 2+ classification):

• 400,000 patients x $500/unit x 4 units/year (quarterly replacement) = $800 million US SAM for DFU alone

Expanding to venous leg ulcers (2.5 million US patients) and pressure injuries (2.5 million US patients) at similar per-unit pricing:

• Total US SAM: $2.0-$3.5 billion per year

Top-down cross-check:

The global smart bandage market was valued at $926 million to $1.75 billion in 2025, projected to reach $2.5-$3.7 billion by 2030-2035 at 15-17% CAGR (Precedence Research, 2025). The global advanced wound care market was valued at $13.37 billion in 2025, projected to reach $19.32 billion by 2030 at 7.6% CAGR (MarketsandMarkets, 2025). The autonomous closed-loop segment — which does not yet exist commercially — represents the highest-value subset of the smart bandage market, commanding premium pricing justified by clinical outcome improvement and amputation prevention.

Reimbursement pathway: No dedicated CPT or HCPCS code exists for autonomous wound monitoring with integrated treatment. The most likely near-term reimbursement strategy combines existing codes: Remote Patient Monitoring setup (CPT 99453, $19.03), RPM device supply with daily recording (CPT 99454, $55.72/30-day period), RPM management (CPT 99457/99458, $50.18/$41.17 per 20-minute increment), and wound electrical stimulation (HCPCS G0281/G0282, $15-25/session). Stacking these codes yields approximately $150-250 per patient per month in reimbursable services. A new Category III CPT code application to the AMA would be the strategic long-term path, potentially establishing reimbursement at $300-500 per 30-day treatment cycle to reflect the clinical value of continuous monitoring and early intervention.

6. Research Gap and Commercial Opportunity

The four research groups cited above have independently validated the individual components of autonomous wound therapy — multiplexed biosensing, ML-based wound classification, drug delivery, and electrical stimulation. No group has integrated all components into a single manufacturable system, and the specific gaps between their published results and a deployable product represent the commercial opportunity.

Gap 1: On-device ML inference. Every published system runs its ML model on an external device — Gao's iCares transmits data to a smartphone, Rolandi's a-Heal processes images on a nearby computer. A truly autonomous wound dressing must run classification and treatment decision algorithms directly on a low-power microcontroller (ARM Cortex-M class) embedded in the dressing. This requires model compression techniques — quantization, pruning, knowledge distillation — to reduce a wound classification model from gigabytes to kilobytes while preserving clinical-grade accuracy. No published work has demonstrated on-device wound ML inference at the power and size constraints of a flexible wound dressing.

Gap 2: Human closed-loop treatment validation. Closed-loop drug delivery plus electrical stimulation has been demonstrated only in rodent models (Gao 2023, Jiang 2023). The 2025 human studies validated monitoring and classification but not autonomous treatment. Bridging this gap requires an FDA Investigational Device Exemption (IDE) and a pivotal clinical trial design that satisfies combination product regulatory requirements.

Gap 3: Scalable manufacturing. All prototypes are hand-assembled in research cleanrooms using photolithography, e-beam deposition, and manual microfluidic bonding. No published design-for-manufacturing analysis exists for a bioelectronic wound dressing. The first roll-to-roll production of smart wound dressings was published by Purdue University in 2025 (Advanced Healthcare Materials) — but only for passive colorimetric (color-change) sensors, not active bioelectronic systems with drug delivery. Transitioning from research-grade fabrication to production-grade manufacturing — with statistical process control, incoming material qualification, and cost-of-goods modeling — is the critical commercialization bottleneck.

Gap 4: Regulatory strategy. No research group has published FDA engagement (Pre-Sub, IDE application, or regulatory classification request) for a closed-loop wound dressing with drug delivery. The device would be classified as a combination product (device plus drug) under 21 CFR Part 3, most likely requiring PMA. If the ML algorithm adapts on-device after deployment, the FDA Predetermined Change Control Plan (PCCP) framework applies — requiring a Description of Modifications, Modification Protocol, and Impact Assessment as part of the marketing submission. As of 2025, over 53 devices have FDA-authorized PCCPs, but none in wound care. First-mover PCCP authorization in wound care would establish both regulatory precedent and competitive protection.

These four gaps — on-device ML, human closed-loop treatment, scalable manufacturing, and regulatory strategy — are integration and engineering challenges, not fundamental science problems. The science works. What is missing is the team that combines ML engineering capability with manufacturing expertise and clinical/regulatory knowledge to assemble the validated components into a product.

7. Comparable Funded Projects

DARPA BETR (Bioelectronics for Tissue Regeneration), REPAIR Project. $22 million, 2020. Lead PI: University of Pittsburgh. Partners: Carnegie Mellon University, Northwestern University, Rice University, University of Vermont, University of Wisconsin, Walter Reed National Military Medical Center. Developing AI-directed bioelectronic devices that monitor molecular signals at each healing stage and deliver specific molecules at specific times. Focus on blast and burn injuries relevant to warfighters. This award validates funder interest in closed-loop bioelectronic wound therapy at scale — DARPA committed $22 million to a single consortium for exactly this application.

DARPA BETR, UC Santa Cruz Collaboration. Up to $16 million, 2020-2025. Lead PI: Marco Rolandi (UC Santa Cruz). Partners: UC Davis, Tufts University. Produced the a-Heal platform (Citation 4 above). Focus on ML-driven adaptive bioelectronic wound therapy with fluoxetine delivery and electrical stimulation. This award produced the first "AI physician" for wound management and demonstrates DARPA's sustained commitment to the field.

DARPA BEST (BioElectronics to Sense and Treat). Approximately $22.8 million in performer awards, solicitation published March 2025 (DARPA-PS-25-12). Program Manager: Roozbeh Jafari (same PM as BETR). Technical objective: TRL-5 prototype at pilot scale with rigorous testing. Focuses specifically on wearable, automated bioelectronic smart bandages with sensor and treatment modules for real-time wound infection monitoring, prediction, diagnosis, and closed-loop treatment. Requires non-antibiotic treatments, large-animal preclinical testing, and FDA IDE preparation. This is the direct follow-on to BETR, with higher TRL targets and explicit FDA transition milestones — signaling that DARPA views this technology as approaching clinical readiness.

NIH R01HL155815. Wei Gao, Caltech. Multi-year R01 grant supporting development of wearable bioelectronic systems for wound monitoring. This grant funded both the 2023 Science Advances animal study and the 2025 Science Translational Medicine human clinical validation. Additional funders include NSF, American Cancer Society, and Army Research Office.

NSF CRII Award. $175,000, Florida International University. Research Initiation Initiative grant for development of a wireless adhesive bandage for wound monitoring. Represents NSF investment in the foundational sensor technology.

These five awards total over $60 million in government investment in smart wound dressing technology between 2020 and 2025, with DARPA alone committing over $60 million across BETR and BEST programs. The progression from BETR (2020, research-grade prototypes) to BEST (2025, TRL-5 with FDA transition milestones) demonstrates accelerating funder urgency and confidence in clinical translation.

8. Opportunity Assessment

TRL Assessment: TRL 4 (Component validation in laboratory environment). Closed-loop drug delivery plus electrical stimulation has been validated in rodent chronic wound models by two independent groups (Gao 2023, Jiang 2023). ML-based wound classification has been validated in 20 human patients (Gao 2025). The a-Heal system demonstrated ML-driven treatment decision-making in preclinical models (Rolandi 2025). TRL 5 (system validation in relevant environment) would require first-in-human closed-loop treatment, which no group has achieved. The evidence chain — from individual sensor validation to multiplexed monitoring to animal closed-loop to human monitoring — is continuous and advancing.

Technical Risks and Mitigations:

Risk 1 — On-device ML accuracy degradation from model compression. Wound classification models trained on research-grade compute must be compressed to run on microcontrollers with limited memory (256 KB-2 MB SRAM) and processing power. Mitigation: quantization-aware training with post-training validation against the full-precision model using the Gao 2025 clinical dataset as benchmark. Go/no-go criterion: compressed model must maintain classification AUC above 0.90 compared to expert clinician ground truth.

Risk 2 — Sensor drift during multi-day wear in wound exudate environment. Electrochemical biosensors exposed to chronic wound fluid experience biofouling, protein adsorption, and electrode degradation over days. Mitigation: redundant sensor channels with on-device drift correction algorithms, calibrated against known analyte standards embedded in the disposable sensor strip. This approach mirrors continuous glucose monitor (CGM) factory calibration used by Dexcom and Abbott. Go/no-go at Month 6: sensor accuracy within 15% of reference values after 7 days continuous wear in simulated wound fluid.

Risk 3 — Drug reservoir depletion management. The dressing must detect when the drug reservoir is depleted and alert the patient or clinician. Mitigation: volumetric flow sensors in the microfluidic drug delivery channel, with depletion alerts transmitted via Bluetooth to the patient's smartphone and to the clinician dashboard. The disposable sensor layer is designed for 7-14 day replacement cycles, matching standard wound dressing change frequencies.

Regulatory Pathway: The device would be classified as a combination product (device plus drug) under 21 CFR Part 3. The primary mode of action is the device function (monitoring, classification, controlled delivery), making CDRH the lead review center. If the drug component uses an already-approved molecule (e.g., fluoxetine, silver sulfadiazine, or antibiotics with existing monographs), CDRH-led PMA is the expected pathway. If a novel chemical entity is proposed, CDER involvement increases substantially. Predicate devices for the component functions include MolecuLight i:X (FDA De Novo 2018, wound bacterial imaging), Vomaris Procellera (510(k) K160783, bioelectric antimicrobial dressing), and Dexcom G7 (510(k), continuous wearable biosensor with Bluetooth connectivity). The ML algorithm introduces a software component classified under FDA's Software as a Medical Device (SaMD) framework. If the algorithm is locked after training (does not update on-device), standard software verification and validation suffices. If the algorithm is adaptive — learning from individual patient data to personalize treatment — the PCCP framework applies. An adaptive algorithm would enable personalized wound therapy but adds 6-12 months of regulatory documentation. Strategic recommendation: initial submission with a locked algorithm (faster clearance, lower risk), followed by PCCP supplemental submission for adaptive capability after market entry. Regulatory approval timeline — including IDE, pivotal trial, and PMA review — is estimated at 3-5 years. This timeline creates a competitive moat: competitors who begin development after this team enters clinical trials face a 3-5 year regulatory delay before they can reach market.

9. Team Requirements

Successful commercialization of autonomous closed-loop wound dressings requires three core capability domains:

Biomedical domain expertise. Understanding of wound healing biology (hemostasis, inflammation, proliferation, remodeling phases), biomarker significance (why nitric oxide indicates inflammation, why hydrogen peroxide signals bacterial infection, why pH shift correlates with healing trajectory), and clinical workflow integration. This expertise enables correct experimental design for preclinical and clinical validation studies, regulatory engagement with FDA CDRH reviewers on clinical evidence requirements, and clinical problem framing that resonates with wound care specialists and payers.

Machine learning engineering. On-device ML model development is the single largest technical gap. Capabilities required include TinyML model design for microcontroller deployment (ARM Cortex-M class, sub-2 MB memory), adaptive control algorithms for drug delivery optimization, model compression techniques (quantization, pruning, knowledge distillation), and rigorous evaluation methodology for validating that ML-driven treatment decisions match expert clinician judgment. This last capability — evaluation framework design — is critical for FDA submission, where the clinical performance of the ML model must be demonstrated against a clinician reference standard.

Manufacturing engineering. This is the most critical commercialization gap across all published prototypes. Capabilities required include flexible electronics manufacturing (roll-to-roll printed electronics, flexible PCB assembly, microfluidic channel fabrication), 3D printing scale-up from lab-grade to production-grade with statistical process control, GMP compliance (21 CFR Part 820 quality system regulation for medical devices), COGS modeling to determine pricing strategy and reimbursement requirements, and supply chain qualification for medical-grade polymers, biocompatible inks, and drug-grade reservoir materials. Most research proposals for bioelectronic wound dressings end at "it works in the lab." The manufacturing engineering discipline ensures that prototype decisions consider production scaling, tolerance analysis, and quality systems from day one — addressing the valley of death between TRL 4-5 prototypes and TRL 7+ deployable systems where most funded research stalls.