00
Executive Summary
The NHS spends hundreds of millions annually on diagnostic imaging for head and neck conditions — many of which could be identified faster, cheaper, and more accurately through an entirely different modality: sound.
The Acoustic Diagnostic Mesh System (ADMS) is a flexible, wearable device that listens to the body as it moves. Using an array of MEMS microphones, inertial sensors, and real-time AI analysis, it captures the acoustic signatures of bones, joints, tendons, arteries, nerves, and fluids — precisely located in 3D space and tagged to the exact movement that produced them. A standard session takes 5–10 minutes. No radiation. No specialist at point of screening. No invasive procedure.
This document is presented at concept stage to invite clinical and academic review. No clinical claims are made pending evidence development.
Primary Benefit
First-line triage tool replacing unnecessary CT/MRI referrals at primary and secondary care level
Financial Case
£2M–£8M annual saving per Trust; system-wide adoption could reclaim hundreds of millions per year
Long-Term Value
Generates a living biomechanical atlas — a population-scale dataset for predictive and preventative medicine
01
Clinical Need & Problem Definition
The NHS faces sustained pressure across diagnostic pathways for musculoskeletal, neurological, and vascular conditions of the head and neck. A fundamental diagnostic gap exists: conditions that are dynamic — only presenting under specific movement — remain invisible to static imaging technologies.
Static Imaging Blindspot
CT and MRI capture anatomy at rest. A cervical joint that misaligns only at 35° of rotation appears entirely normal — yet symptoms are real and disabling.
Disproportionate Cost
A single MRI costs £350–£800; a CT £250–£500. Radiology departments are near or at capacity. Head/neck imaging referrals have risen 23% over five years.
Subjective Symptom Reporting
Clinicians depend on patient descriptions. Vague presentations lead to repeated consultations, misclassification, and inappropriate mental health referrals.
Diagnostic Delay
Patients with hidden mechanical or vascular causes — ligament instability, arterial compression, CSF flow anomalies — can wait months or years while harm accumulates.
System-Wide Backlog
Every unnecessary imaging referral displaces a case that genuinely requires it. Diagnostic backlog reduction is a stated NHS priority requiring upstream triage innovation.
Hidden Vascular Events
Partial arterial blockages and vascular compression in the head and neck are difficult to detect — yet acoustic changes precede visible structural change.
NHS Context: The Diagnostic Economy
NHS England spends an estimated £1.3 billion annually on diagnostic imaging. Head and neck presentations account for a significant and growing proportion. Primary care GPs have limited first-line tools beyond clinical examination, creating a binary choice between watchful waiting and expensive imaging. The ADMS fills this gap as a rapid, low-cost, first-line screen.
02
Proposed Diagnostic Solution
The ADMS is grounded in a simple but profound physical principle: different biological structures transmit and reflect sound differently. This is already exploited in geophysics to map underground formations. Applied passively to the human head and neck, it becomes a precision diagnostic instrument.
The Acoustic Palette of the Human Body
Sound Signature
Structure
Clinical Indicator
Freq.
Bone · Joint
Misalignment, degeneration, bone-on-bone contact
200–2,000 Hz
Tendon · Ligament
Laxity, micro-tear, instability, stretch injury
500–5,000 Hz
Artery · Vessel
Stenosis, compression, partial blockage, vascular anomaly
20–500 Hz
Fluid · Soft Tissue
CSF flow issues, oedema, soft tissue stress
10–200 Hz
Muscle Fibre
Tone abnormalities, spasm, nerve-related tension
50–400 Hz
Cranial · Intracranial
Intracranial pressure change, CSF event, vascular collapse
5–100 Hz
Technology Precedent — This Is Not Speculative
Geophysicists map underground formations using seismic sound. Smartphone IMUs track motion to sub-degree precision. Medical ultrasound proves that sound reveals internal anatomy non-invasively. The ADMS combines these established domains in a novel configuration — passive acoustic listening, synchronised with motion data, analysed by AI trained on anatomical ground truth.
03
Clinical Safety, Risk & Governance
As a passive listening device that emits no energy into the body, the ADMS presents an inherently low direct-harm profile. Governance and clinical safety frameworks must nonetheless address diagnostic accuracy risks, device hygiene, and the consequences of false-negative outputs.
HIGH
False Negative: Missed Serious Pathology
ADMS must not be used to rule out serious conditions. It is a triage supplement, not a diagnostic replacement. Clinical protocols must mandate escalation thresholds.
HIGH
AI Misclassification
Machine learning outputs carry inherent uncertainty. Clinician override must always be possible. Confidence intervals and explainability are embedded requirements.
MED
Infection Control
Single-use liners or validated decontamination protocols required. Materials must be compatible with standard NHS cleaning agents. IPC compliance assessed before deployment.
MED
Patient Harm During Movement Replication
Patients with acute or unstable cervical pathology should not reproduce provocative movements without clinical supervision. Contraindication screening protocol required.
LOW
Skin and Comfort
Hypoallergenic materials required. Session duration is short; prolonged pressure unlikely to cause harm. Paediatric and elderly variants may require modified sizing.
LOW
Electromagnetic Interference
EMC compliance testing required in clinical environments, particularly near MRI suites or active medical equipment.
Governance Framework
ADMS deployment must be governed by a Clinical Oversight Committee (radiologists, neurologists, ENT surgeons, patient safety lead). A formal Clinical Safety Case under DCB0129/DCB0160 is required before any clinical deployment. Adverse event reporting must integrate with the NHS NRLS.
04
Evidence Requirements
Before clinical commissioning, ADMS must demonstrate diagnostic accuracy against established reference standards. A phased evidence programme is proposed, moving from laboratory validation to large-scale comparative trials.
1
Acoustic Signature Laboratory Validation
Controlled cadaveric and phantom studies to characterise frequency-specific signatures for each tissue type. Target: coefficient of variation <10% across 3 prototype units.
2
Healthy Volunteer Baseline Study (n=100+)
Establish population-level acoustic baselines stratified by age, sex, BMI, and anatomical variation. Required to train AI classification models and define normal reference ranges.
3
Prospective Comparative Study vs. MRI/CT (n=500+)
ADMS assessment before confirmed radiological diagnosis. Primary endpoints: sensitivity ≥85%, specificity ≥80% for major diagnostic categories. Pre-registered on ISRCTN.
4
Inter-Rater Reliability Study
Multiple trained clinicians interpret identical ADMS outputs independently. Agreement coefficient (Cohen's κ) target >0.75. Informs standardisation of output interpretation.
5
Longitudinal Monitoring Study (12 months)
Patients with known conditions assessed monthly. Demonstrates ADMS utility for disease progression monitoring and treatment response tracking.
6
Health Technology Assessment (NICE)
Full HTA submission. Cost-effectiveness modelling (ICER per QALY). Submission through Medical Technologies Evaluation Programme (MTEP).
≥85%
Sensitivity
Protects against false negatives in serious pathology
≥75%
Specificity
Prevents unnecessary escalation from false positives
≥90%
NPV
For vascular and neurological categories specifically
All
Subgroups
Age >65, high BMI, paediatric populations required
05
Integration into NHS Workflow
The ADMS is designed as a first-line triage adjunct — not a replacement for specialist assessment or confirmed imaging. Integration is envisioned across three care settings, with clear escalation pathways and full EPR interoperability.
Primary Care (GP / PCN)
First-line screen when a patient presents with unexplained head or neck symptoms. Replaces the binary choice between watchful waiting and immediate imaging referral.
Secondary Care (Outpatient / A&E)
ENT, neurology, and vascular surgery outpatient clinics use ADMS to characterise presentations before ordering specialist imaging.
Pathology & Radiology Interface
ADMS outputs exported as HL7 FHIR-compatible data. Radiology referrals include ADMS findings as pre-test context, improving request quality.
Digital Systems Integration
Full integration with NHS Spine, EMIS, SystmOne, and Lorenzo EPR. SNOMED CT coding, NHS number linkage, DSPT compliance.
Clinical Pathway: ADMS-Guided Triage
01
Patient Presents
GP / ENT / ED head or neck complaint
02
ADMS Session
5–10 min acoustic-motion map
03
AI Analysis
Tissue type · source location · confidence
04
Triage Decision
Low / Moderate / High complexity flag
05
Targeted Action
Discharge · referral · urgent imaging
06
Regulatory Pathway
The ADMS is classified as an active medical device under UK Medical Devices Regulations 2002 (as amended). As a diagnostic device that does not emit energy into the body but processes and presents diagnostic information, it is expected to qualify as a Class IIa device under UKCA marking requirements.
UKCA / MHRA Regulatory Route — Four-Stage Pathway
01
MHRA Innovation Office
Pre-submission meeting to confirm classification and identify applicable standards: IEC 62304, IEC 60601-1, ISO 14971
02
Technical Documentation
Clinical evaluation, risk management file, QMS (ISO 13485), software lifecycle documentation
03
UK Approved Body Review
Conformity assessment; quality system audit; technical file review by MHRA-designated Approved Body
04
UKCA Mark & UDI
UKCA mark affixed; device registered on MHRA system; post-market surveillance plan activated
Software as a Medical Device (SaMD)
The AI analysis component constitutes SaMD and requires separate MHRA/UKCA assessment under the IMDRF SaMD framework. The AI model must provide explainable outputs that clinicians can interrogate and override. NICE Evidence Standards Framework for Digital Health Technologies (Tier D) will apply.
07
Data Protection & GDPR Compliance
ADMS generates novel biometric data — acoustic signatures of internal anatomy — constituting special category health data under UK GDPR Article 9. A comprehensive data governance framework is a prerequisite for any clinical deployment.
Lawful Basis
Article 9(2)(h) — medical diagnosis and treatment — with explicit patient consent. Research use governed by separate ethics approval and data sharing agreements.
Data Architecture
Raw audio never transmitted externally. Only processed feature vectors exported for AI training. NHS-approved cloud infrastructure (NHS Digital / NHSX compliant).
DPIA Requirement
Mandatory before deployment. Novel biometric data type triggers high-risk processing classification under ICO guidance. DPO sign-off required at each Trust.
NHS DSPT Compliance
Full compliance with NHS Data Security and Protection Toolkit. Annual certification required for any system accessing NHS patient data or networks.
Research Biomechanics Atlas
Anonymised acoustic data governed by REC approval, Data Access Agreement, and published governance policy. NHS-owned dataset with defined public benefit purpose.
Retention & Deletion
Clinical data retained per NHS Records Management Code of Practice (8 years post-episode). Patient right to erasure applied where technically feasible.
08
Health Economics
The financial case for ADMS is built on avoided costs. Conservative modelling suggests compelling ROI even at modest adoption rates. All figures are indicative; formal cost-effectiveness analysis will be required as part of NICE MTEP/HTA submission.
£250–800
Cost per CT/MRI scan avoided
£2M–8M
Annual saving per Trust (conservative)
30–50%
Projected unnecessary imaging reduction
<18 mo
Estimated payback period per deployment
| Cost Category | Current NHS Cost | ADMS Impact | Projected Saving / Trust |
|---|
| CT scan (head/neck) | £250–£500 / scan | Reduce by 35–50% through pre-imaging triage | £180k–£600k |
| MRI scan (head/neck) | £350–£800 / scan | Reduce by 30–40% — targeted referral only | £240k–£900k |
| Specialist outpatient consultation | £200–£500 / visit | Fewer repeat consultations; better pre-referral information | £120k–£400k |
| Mental health misreferral | £1,000–£5,000 / case | Eliminated where physical root cause is objectively identified | £80k–£350k |
| Delayed diagnosis complications | £3,000–£20,000+ / event | Early detection prevents high-cost acute episodes | £200k–£1.5M |
| Medicolegal costs (misdiagnosis) | £25k–£500k+ / claim | Reduced through objective diagnostic documentation | Significant |
Long-Term Strategic Value: The Biomechanics Atlas
Beyond direct cost savings, systematic ADMS deployment generates an NHS-owned population-scale acoustic dataset. Within 3–5 years of deployment, this dataset would contain millions of validated diagnostic recordings — enabling AI models that pre-empt conditions before symptoms become serious, and permanently reducing the long-term burden of chronic musculoskeletal and vascular disease on NHS resources.
09
Pilot Deployment Plan
A phased pilot approach is proposed, beginning in Primary Care Networks and ascending to multi-Trust regional deployment. Each phase has defined success metrics and go/no-go criteria.
Phase 1
Proof of Concept — Academic & Industry Partnership
Months 0–12
- Prototype flexible MEMS mesh with IMU sensor array — target unit cost <£200
- Acoustic classification algorithm development using cadaveric and volunteer data
- Partner engagement: University of Birmingham Biomedical Engineering, Birmingham Health Partners
- MHRA Innovation Office pre-submission meeting; ISO 13485 QMS established
Phase 2
Clinical Validation — NHS Trust Research Partnership
Months 12–30
- Comparative trial: ADMS vs. MRI/CT in 500+ patients across 2 NHS Trust sites
- Target sites: University Hospitals Birmingham / Leeds Teaching Hospitals
- AI model training on validated acoustic-outcome pairs
- Clinical Safety Case produced under DCB0129; DPIA completed
- UKCA conformity assessment initiated with UK Approved Body
Phase 3
NHS Pilot Deployment — PCN & Secondary Care
Months 30–48
- Deploy in 3 PCNs and 5 outpatient departments across 2 NHS regions
- Full EPR integration (EMIS, SystmOne) and NHS Spine connectivity
- Prospective measurement of imaging referral rates, diagnostic accuracy, patient satisfaction
- Health economic analysis prepared for NICE MTEP submission
Phase 4
National Commissioning & Scale
Months 48+
- NICE Diagnostics Guidance published; NHS England commissioning recommendation sought
- National procurement framework established; manufacturing at scale
- Biomechanics data atlas established under NHS Research governance
10
Scalability Across the NHS
The ADMS is designed from first principles to be mass-deployable. Unlike capital-intensive imaging equipment, it requires no dedicated room, no radiation safety infrastructure, and no specialist operator at the point of use.
Low Capital Cost
Target unit cost of £150–£300 at manufacturing scale. Compared to £800k–£2M for an MRI suite, a single Trust could deploy 50+ units for the cost of one scan room.
Cloud-Enabled Analytics
AI analysis runs via secure NHS-hosted cloud. Device transmits compressed feature data — not raw audio — enabling deployment in rural and community settings.
Minimal Training
Device operation designed for trained HCAs and practice nurses. A half-day training module covers application, session management, output interpretation, and escalation.
Equity & Access
Deployable in GP surgeries, community health centres, remote clinics, and domiciliary settings. Directly addresses NHS health inequalities agenda.
Improving Over Time
Each deployment contributes anonymised data to the central AI model. As the training set grows, diagnostic accuracy improves — a virtuous cycle where scale directly enhances performance.
Modular Extension
The acoustic-motion diagnostic principle extends beyond head and neck. Future applications include spinal assessment, cardiac auscultation enhancement, and joint monitoring.
11
Barriers, Risks & Mitigation
| Barrier / Risk | Impact | Likelihood | Mitigation |
|---|
| Clinical scepticism — new modality without established evidence base | High | High | Rigorous comparative trial design; respected clinical champions; NICE guidance as anchor |
| Acoustic noise contamination — environmental interference in clinical settings | Medium | Medium | Multi-layer active noise cancellation; validated noise floor thresholds per setting type |
| Anatomical variability — individual differences in tissue acoustic properties | High | High | Large, diverse training datasets; patient-specific baseline calibration at first session |
| AI transparency — black-box outputs unacceptable in clinical governance | High | Medium | Explainable AI architecture; confidence scoring for all outputs; mandatory clinician override |
| Regulatory timeline — UKCA process may take 18–36 months | Medium | High | Early MHRA Innovation Office engagement; Innovate UK / SBRI to fund regulatory work |
| Data security — novel biometric data type and patient trust | High | Low–Med | PPI from design stage; transparent consent; NHS-hosted infrastructure only |
| EPR integration resistance — NHS IT complexity | Medium | Medium | HL7 FHIR-compliant API from outset; NHS Digital engagement; phased integration |
Patient and Public Involvement
PPI must be embedded from the earliest stage. Patients who have experienced delayed or missed diagnosis due to dynamic head and neck conditions are uniquely qualified to shape device design, consent processes, and output presentation. A PPI advisory panel is recommended as a standing governance body throughout all development phases.
"The body generates diagnostic information that current technology does not capture. This proposal invites conversation about whether it should."
ADMS · NHS Innovation Proposal · 2026 · Concept Stage · All financial figures are indicative modelled estimates. No clinical efficacy claims are made. For innovation review purposes only.