1. Clinical Need
Chronic inflammatory and autoimmune diseases — rheumatoid arthritis (RA), inflammatory bowel disease (IBD), systemic lupus erythematosus, and psoriatic arthritis — affect an estimated 24 million Americans (National Institutes of Health). These conditions are managed primarily through immunosuppressive biologics (TNF inhibitors, IL-6 receptor antagonists, JAK inhibitors) that cost $30,000–$80,000 per patient per year and carry adverse effects including increased infection risk, hepatotoxicity, and malignancy.
Rheumatoid arthritis alone affects approximately 1.3 million US adults (CDC, National Health Interview Survey). Of these, 40% have moderate-to-severe disease requiring biologic therapy, and 30–40% of patients on biologics either fail to respond or lose response over time — termed secondary failure (Smolen et al., “Rheumatoid arthritis,” The Lancet, 2016). These treatment-resistant patients, numbering 156,000–208,000 in the US, have exhausted available pharmacological options.
The Agency for Healthcare Research and Quality estimates annual direct medical costs for RA exceed $19.3 billion in the US, with biologic drug costs constituting the largest component. Indirect costs — disability, lost productivity, caregiver burden — add an estimated $39.2 billion annually (Birnbaum et al., “Societal cost of RA in the US,” Current Medical Research and Opinion, 2010). The total economic burden of autoimmune disease in the US exceeds $100 billion per year.
An alternative therapeutic modality — electrical stimulation of the vagus nerve to activate the cholinergic anti-inflammatory pathway — received FDA approval in July 2025 via SetPoint Medical’s open-loop device for moderate-to-severe RA, following a pivotal trial published in Nature Medicine (2025). However, the approved device delivers fixed, open-loop stimulation irrespective of the patient’s current inflammatory state. The unmet clinical need is a closed-loop system that adapts stimulation in real time based on physiological biomarkers, reducing energy consumption, minimizing side effects (bradycardia, voice alteration), and optimizing therapeutic efficacy for individual inflammatory dynamics.
2. State of the Art
Three parallel developments have converged to define the research opportunity in adaptive bioelectronic medicine.
FDA-validated neuroimmune biology
Kevin Tracey’s laboratory at the Feinstein Institutes for Medical Research identified the inflammatory reflex — a neural circuit in which vagus nerve stimulation activates splenic T cells to release acetylcholine, suppressing TNF-alpha and other pro-inflammatory cytokines via alpha-7 nicotinic acetylcholine receptors on macrophages. SetPoint Medical translated this biology into an implantable pulse generator (IPG) that stimulates the cervical vagus nerve for 1–5 minutes daily. The pivotal trial, published in Nature Medicine (2025), demonstrated statistically significant improvement in DAS28-CRP scores versus sham stimulation in biologic-inadequate-response RA patients. FDA approval followed in July 2025 — the first approved bioelectronic medicine device for autoimmune disease. This establishes that neuroimmune modulation via VNS is a clinically effective, FDA-recognized treatment pathway.
Reinforcement learning for adaptive neurostimulation
Liu et al. (2024, IEEE Transactions on Neural Systems and Rehabilitation Engineering) implemented a Twin Delayed Deep Deterministic Policy Gradient (TD3) reinforcement learning agent for closed-loop deep brain stimulation. The RL agent achieved 67% reduction in power dissipation compared to continuous open-loop stimulation while preserving normal basal ganglia-thalamic response dynamics. Separately, Brambilla et al. (2024, PMC) demonstrated transfer learning for RL-based closed-loop VNS for cardiovascular regulation, showing improved sample efficiency when adapting stimulation policies across individual rat cardiovascular models. These results establish that RL-based adaptive control is technically feasible and quantitatively superior to fixed stimulation.
Miniaturized closed-loop hardware
Dickey et al. (2025, Stroke) reported a miniaturized closed-loop VNS system — 50 times smaller than preceding devices — delivering sensor-triggered stimulation paired with rehabilitation in chronic stroke patients. Participants showed lasting motor recovery. In parallel, Farrell et al. (2025, Scientific Reports) demonstrated a fully automated wireless VNS system with real-time dynamic adjustment of stimulation parameters to minimize bradycardia using physiological feedback, validated in vivo. These results demonstrate that the hardware engineering challenge — integrating sensors, processing, and stimulation in a compact implantable form factor — has been solved at prototype scale.
The field has achieved validated biology (FDA-approved), validated algorithms (67% power reduction), and validated hardware (50x miniaturized, human-tested). What does not exist is the integration of all three: an implantable device that uses RL-optimized adaptive control to deliver personalized VNS for inflammatory disease. This integration gap is the research opportunity.
3. Foundational Research
SetPoint Medical Pivotal Trial (2025). “Vagus nerve-mediated neuroimmune modulation for rheumatoid arthritis: a pivotal randomized controlled trial.” Nature Medicine, 2025.
Multicenter randomized, double-blind, sham-controlled trial of cervical VNS in patients with moderate-to-severe RA and inadequate response to biologic therapies. Primary endpoint: change in DAS28-CRP at 12 weeks showed statistically significant improvement in the active stimulation group versus sham. Led directly to FDA approval in July 2025. This result is foundational: the underlying biology is FDA-validated. Any closed-loop system builds on this established efficacy, reducing clinical risk to the adaptive control component.
Liu et al. (2024). “Closed-Loop Deep Brain Stimulation with Reinforcement Learning.” IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2024. PubMed: 39302783.
TD3 reinforcement learning agent for closed-loop deep brain stimulation in a computational model of basal ganglia-thalamic circuits. Achieved 67% reduction in stimulation power consumption compared to continuous open-loop, while maintaining normal thalamic relay fidelity. Two-thirds less energy consumption translates to longer battery life (fewer replacement surgeries), reduced tissue heating, and lower side effect burden — critical factors for implantable device clinical viability and patient acceptance.
Brambilla et al. (2024). “Reinforcement learning for closed-loop regulation of cardiovascular system with vagus nerve stimulation: a computational study.” PMC, 2024. PMCID: PMC11145940.
Applied RL to closed-loop VNS for heart rate regulation across individualized computational models of rat cardiovascular physiology. Transfer learning between individual models improved sample efficiency, addressing the personalization challenge — each patient’s vagus nerve anatomy, fiber composition, and inflammatory dynamics differ. This result is significant for clinical translation: RL models pre-trained on population data can be efficiently fine-tuned to individual patients, reducing the calibration burden.
Dickey et al. (2025). “Closed-Loop Vagus Nerve Stimulation Delivered With a Miniaturized System Produces Lasting Recovery in Individuals With Chronic Stroke.” Stroke, 2025.
VNS system 50 times smaller than preceding devices, delivering closed-loop stimulation triggered by sensor input during rehabilitation in chronic stroke patients with lasting motor recovery. First clinical demonstration of a miniaturized closed-loop VNS system in human subjects. Establishes that the hardware engineering challenge — integrating sensors, processing, and stimulation in a compact implantable form factor — has been solved at prototype scale.
Farrell et al. (2025). “Fully automated wireless vagus nerve stimulation.” Scientific Reports, 2025.
Real-time, fully automated adjustment of VNS parameters using physiological feedback to minimize bradycardia — the most common adverse effect of vagus nerve stimulation. Dynamically modulated pulse width and frequency without human intervention based on continuous heart rate monitoring. Validated in vivo. Addresses a specific safety concern that has limited VNS adoption: the risk of excessive parasympathetic activation during unsupervised chronic use.
4. Competitive Landscape
SetPoint Medical (Tarrytown, NY). Total funding: $581M+. FDA-approved VNS for RA (July 2025). The approved device delivers open-loop stimulation — fixed daily stimulation at predetermined parameters. SetPoint has not disclosed work on closed-loop adaptive stimulation. Their approval establishes a predicate device for the VNS-for-autoimmunity device class.
Galvani Bioelectronics (Stevenage, UK). GSK/Verily joint venture, $540M commitment (2016). Targets the splenic nerve rather than cervical vagus, with a technically distinct approach facing additional anatomical challenges (small, variable nerve embedded in splenic artery adventitia). Early feasibility trials. No approved product.
No entity currently sells or has in clinical trials a closed-loop adaptive VNS system for inflammatory disease. Established neurostimulation companies (Medtronic, Abbott, Boston Scientific) have closed-loop products for other indications (responsive neurostimulation for epilepsy, adaptive DBS for Parkinson’s) but have not entered the autoimmune VNS market.
5. Addressable Scope
Bottom-up calculation (US, rheumatoid arthritis)
- US adults with RA: 1.3 million (CDC, National Health Interview Survey)
- Moderate-to-severe requiring biologic therapy: 40% = 520,000
- Biologic-inadequate-response (treatment-resistant): 30% = 156,000
- VNS system cost (device + implant): $40,000 (comparable to spinal cord stimulators)
- US addressable population for RA: 156,000 × $40,000 = $6.24 billion annually
- Expansion to broader autoimmune (Crohn’s: 780K; UC: 910K; lupus: 204K): additional ~750,000 biologic-eligible patients
- Expanded US scope: (156K + 225K) × $40K = $15.24 billion annually
Top-down cross-check
Global bioelectronic medicine market: $25.8 billion (2023), projected $35.7 billion by 2028 at 6.7% CAGR (Grand View Research, 2023). Vagus nerve stimulation: $2.3 billion (2023), projected $5.1 billion by 2031 at 10.4% CAGR (Data Bridge Market Research, 2024). Closed-loop adaptive VNS for autoimmunity capturing 20–30% of the VNS market by 2031 yields $1.0–$1.5 billion — conservative relative to the bottom-up estimate, reflecting initial penetration.
Initial deployment
Constrained by implanting physician capacity: ~500 qualified centers in the US, 100 implants/center/year: 50,000 procedures × $40,000 = $2.0 billion initial annual scope.
6. Research Gaps and HHA Contribution
Three specific gaps separate published results from a deployable closed-loop bioelectronic medicine platform. Each maps to specific HHA team expertise.
Gap 1: Adaptive control algorithms validated in vivo for inflammatory biomarkers
Current RL demonstrations (Liu et al., 2024; Brambilla et al., 2024) use computational models, not live inflammatory biomarkers. The transition from simulated signals to real-time physiological proxies (C-reactive protein, vagus nerve compound action potential amplitude, heart rate variability as an inflammatory marker) requires sensor integration and algorithm validation in animal models of collagen-induced arthritis or DSS colitis.
HHA contribution: Haedar Hadi (Lead PI) designs the RL control architecture — constrained policy optimization with safety bounds, transfer learning for patient-specific adaptation, and evaluation methodology for comparing adaptive versus open-loop efficacy. Hass Dhia (Co-PI) defines clinically meaningful control endpoints based on inflammatory biomarker dynamics and designs the preclinical validation protocol using established autoimmune animal models.
Gap 2: Integrated miniaturized hardware with onboard inference
Dickey et al. (2025) demonstrated 50x miniaturization. Farrell et al. (2025) demonstrated automated parameter adjustment. No system combines both with onboard ML inference for inflammatory disease. The engineering challenge is integrating sensing (vagus nerve electroneurography, accelerometry, impedance), processing (low-power edge inference running a policy network), and stimulation in a hermetically sealed implantable package smaller than 5 cm³.
HHA contribution: Ahmed (Key Team Member, Director of Manufacturing) leads hardware integration and Design for Manufacturability from day one. ASIC design selection, biocompatible encapsulation (titanium housing, ceramic feedthroughs), electrode-tissue interface engineering, and process validation for hermetic sealing. Most research proposals end at “it works in the lab.” This proposal includes explicit DFM milestones at every phase, ensuring that prototype decisions consider production scaling, tolerance analysis, and quality systems from project inception. This addresses the valley of death between TRL 4–5 prototypes and TRL 7+ deployable systems — the gap where most funded bioelectronic research stalls.
Gap 3: Manufacturing at clinical volumes with medical device quality systems
Current prototypes are hand-assembled in academic labs. Clinical deployment at scale (50,000–156,000 devices/year) requires automated assembly, component sourcing for medical-grade materials, ISO 13485-certified manufacturing processes, incoming quality inspection for sensors and electrodes, and supply chain qualification. No academic lab has addressed these challenges because they are manufacturing engineering problems, not research questions.
HHA contribution: Ahmed’s manufacturing expertise directly addresses this gap. His experience in production scaling, quality systems, and process optimization applies to the specific challenges of miniaturized bioelectronic device production: component placement tolerances (<50 μm for electrode arrays), batch consistency testing, accelerated aging protocols for hermetic seal validation, and supplier qualification for medical-grade titanium, platinum-iridium electrodes, and ceramic feedthroughs.
Why originating labs have not closed these gaps
The RL groups (Liu, Brambilla) are computational neuroscience labs without manufacturing infrastructure or FDA regulatory experience. The hardware groups (Dickey) focus on specific clinical applications (stroke) without inflammatory biomarker integration. SetPoint Medical, with FDA approval, has no disclosed adaptive control program. The gaps persist because they require integration across disciplines — adaptive algorithms, biomedical domain expertise, and manufacturing engineering — that do not coexist in any single academic lab or current commercial entity.
7. Comparable Funded Projects
| Source | PI / Entity | Amount | Focus |
|---|---|---|---|
| DARPA ElectRx | Multiple (7 teams) | ~$78M | Miniaturized closed-loop neuromodulation for inflammatory, metabolic, and neurological conditions (2014–2019). Established scientific and engineering foundations for closed-loop bioelectronic medicine. |
| NIH SPARC | Multiple PIs | >$238M | Stimulating Peripheral Activity to Relieve Conditions. Cumulative NIH Common Fund investment since 2015. Peripheral nerve mapping, closed-loop neuromodulation, vagus nerve stimulation for inflammation. |
| SetPoint Medical (private) | K. Tracey / SetPoint | $581M+ | VNS for autoimmune disease. FDA-approved July 2025. Validates device class and therapeutic mechanism at the highest confidence level. |
| Galvani Bioelectronics (private) | GSK / Verily JV | $540M | Splenic nerve stimulation for RA. Early feasibility trials. Different nerve target but same disease indication. |
| ARPA-H (2024) | Vanderbilt University | $12M | Autonomous surgical systems with enhanced sensing. Adjacent miniaturized implantable sensing technology. |
Total identified investment in bioelectronic medicine for inflammation exceeds $1.4 billion across government and private sources over the past decade. This validates sustained funder interest in the approach and signals that the closed-loop adaptive segment — the next logical step beyond SetPoint’s open-loop approval — is a fundable research direction.
8. Opportunity Assessment
TRL evidence chain: TRL 4 for the integrated closed-loop adaptive system. SetPoint’s open-loop device: TRL 9 (FDA-approved). RL algorithms for neurostimulation: TRL 3–4 (computational models, early animal studies). Miniaturized closed-loop VNS hardware: TRL 5 (human stroke patients, Dickey et al., 2025). Integration of adaptive RL with miniaturized hardware for inflammatory biomarkers: not yet demonstrated — integrated system at TRL 4.
Risk 1: Real-time inflammatory biomarker sensing in vivo
Cytokine levels cannot be measured directly by implanted sensors. Proxy biomarkers — heart rate variability (HRV), vagus nerve compound action potential (CNAP) amplitude, skin conductance — correlate with inflammatory state but have not been validated as closed-loop control signals for VNS in autoimmune models.
Mitigation: HRV is measured by cardiac implantable devices with decades of validation. CNAP recording from vagus nerve cuff electrodes is established. Initial systems use HRV as primary feedback with CNAP as secondary input.
ModerateRisk 2: RL policy safety in autonomous implanted system
An RL agent increasing stimulation in response to perceived inflammation could cause excessive parasympathetic activation (bradycardia, asystole).
Mitigation: Safety-constrained RL (constrained policy optimization) with hard physiological limits on amplitude, frequency, and duty cycle implemented in firmware. Farrell et al. (2025) automated safety controller demonstrates real-time bradycardia prevention via physiological feedback.
ModerateRisk 3: Onboard inference power consumption
Running a neural network policy on an implanted ASIC must consume microwatts to preserve 5–10 year battery life.
Mitigation: TD3 policies (Liu et al., 2024) use small networks (two hidden layers). Quantized and pruned, these run on sub-milliwatt neuromorphic ASICs. TinyML techniques demonstrate inference under 1 mW on microcontrollers with 240 KB RAM.
ModerateRegulatory pathway
FDA De Novo or 510(k) with SetPoint’s approved VNS for RA as predicate. The adaptive AI/ML component falls under FDA’s 2023 guidance on predetermined change control plans (PCCP) for AI/ML-enabled device software. Classification: Class II or III medical device. Estimated timeline: 12–18 months for pre-submission meetings and classification, 18–24 months for pivotal trial, 6–12 months for FDA review. Total: 3–4.5 years to market authorization.
9. Team Capabilities
Hass Dhia
MS Biomedical Sciences with medical school background (anatomy TA). Deep understanding of vagus nerve anatomy — cervical trunk fascicular organization, branch points to cardiac, pulmonary, and abdominal viscera, and the cholinergic anti-inflammatory pathway (splenic nerve, alpha-7 nAChR). AI infrastructure architect with experience in multi-agent orchestration and evaluation framework design. Maps to: clinical problem identification, definition of closed-loop control endpoints based on inflammatory biomarker dynamics, preclinical validation protocol design using collagen-induced arthritis and DSS colitis models, and framing regulatory submissions in clinical language expected by CDRH reviewers.
Haedar Hadi
MS Computer Science (Boston University, Information Systems focus) with cloud and database architecture expertise. The RL optimization problem described in this brief — constrained policy optimization for safe neurostimulation, transfer learning for patient-specific adaptation, evaluation methodology for comparing adaptive versus open-loop efficacy — maps directly to his training. The TD3 architecture demonstrated by Liu et al. (2024) is a standard deep RL algorithm; Haedar’s contribution is in evaluation methodology (how to rigorously compare adaptive vs. open-loop in preclinical models with proper statistical controls) and scalable compute infrastructure for training RL policies on simulated patient populations derived from physiological models.
Ahmed
Director of Manufacturing with expertise in Design for Manufacturability (DFM), production scaling, quality systems, and process optimization. The single most critical gap between funded bioelectronic research and clinical deployment is manufacturing: academic labs produce hand-assembled prototypes that cannot be produced at clinical volumes. Ahmed’s contribution begins at month 1, not as an afterthought. DFM milestones include: component placement tolerance analysis for electrode arrays (<50 μm), hermetic seal validation via accelerated aging (ASTM F2095), supplier qualification for medical-grade titanium and platinum-iridium, batch consistency testing protocols, and ISO 13485 quality system implementation. Most research proposals end at “it works in the lab.” This proposal includes explicit manufacturing milestones at every phase, ensuring prototype decisions consider production scaling from day one. This addresses the valley of death between TRL 4–5 prototypes and TRL 7+ deployable systems — the gap where most funded bioelectronic research stalls.
10. Recommended Next Steps
Target funder programs
- NIH NIBIB R01 — Biomedical Imaging and Bioengineering. Closed-loop neural interfaces for chronic disease management. Typical award: $250K–$500K/year for 5 years ($1.25M–$2.5M total).
- NIH SPARC OT2 — Stimulating Peripheral Activity to Relieve Conditions. Next-generation closed-loop neuromodulation. OT mechanism allows flexible milestones. Typical award: $500K–$2M.
- ARPA-H — Advanced Research Projects Agency for Health. Bioelectronic medicine and precision neuromodulation. Larger awards ($5M–$15M) with milestone-driven structure. Comparable: $12M Vanderbilt autonomous surgical systems (2024).
- NSF CBET — Biomedical Engineering and Technology. Neural engineering and closed-loop bioelectronic devices. Typical R01-equivalent: $300K–$500K over 3–4 years.
- Schmidt Sciences — Trustworthy AI — Deadline May 17, 2026. Safety-critical AI for medical devices. Closed-loop RL for neurostimulation is a direct fit: trustworthiness = calibrated safety bounds on autonomous implanted algorithms. Tier 2: $1M–$5M+.
Estimated funding range
Based on comparable awards: $2M–$5M for initial preclinical program (R01/SPARC scale). $10M–$15M for full translation through first-in-human (ARPA-H scale). Comparable: DARPA ElectRx funded 7 teams at ~$11M each; SetPoint raised $581M total across all stages from preclinical through FDA approval.
Proposed 24-month milestone timeline
- M1–3 R&D: Computational model of vagus nerve inflammatory reflex for RL training environment. DFM: Component sourcing audit, electrode array tolerance specification. Regulatory: FDA pre-submission meeting request.
- M4–6 R&D: RL agent training on computational model; constrained policy optimization with safety bounds. Sensor integration bench testing (HRV + CNAP). DFM: First prototype with production-compatible component selection.
- M7–9 R&D: Transfer learning validation across simulated patient variation. DFM: Hermetic encapsulation prototyping and accelerated aging testing. Regulatory: Pre-submission meeting with FDA CDRH.
- M10–14 R&D: In vivo validation in collagen-induced arthritis rat model. Closed-loop vs. open-loop comparison on DAS28-equivalent inflammatory endpoints. DFM: Benchtop manufacturing process validation for 10-unit pilot batch.
- M15–18 R&D: Large animal (porcine) vagus nerve stimulation with closed-loop adaptive control. Safety and efficacy data collection. DFM: ISO 13485 quality system documentation initiation. Regulatory: Device classification strategy finalized with FDA.
- M19–24 R&D: Preclinical data package compilation. IND-enabling study design. DFM: Process validation for 100-unit manufacturing capability. Supply chain qualification complete. Regulatory: De Novo or 510(k) submission preparation.