HAKILIX LABS | Hakilix · Autonomous Bio-Digital Twinning for UK Health & Social Care

// 01_PARADIGM_SHIFT

From Reactive Rescue to
Predictive Reablement.

The Limits of Current UK Models

The current ecosystem of elderly care is fundamentally reactive. Traditional telecare systems (pendants, pull-cords) only activate after a catastrophic event has occurred. A pendant does not prevent a fall; it merely reports the trauma. This “ambulance at the bottom of the cliff” approach contributes to avoidable admissions, deconditioning and long-term care placements – issues repeatedly highlighted in UK falls, frailty and reablement policy.

Reliance on active user compliance (remembering to wear or charge devices) is particularly fragile for people with cognitive impairment or for night-time risk. This sits uneasily with the intent of Home First and ageing well strategies adopted across Integrated Care Systems.

The Hakilix Reablement Protocol

Hakilix establishes a continuous “Pattern of Life” Digital Twin in the home. By passively monitoring kinematic data such as gait velocity, stride consistency, time-to-stand and sedentary duration, the system aims to detect micro-degradations in mobility that typically precede a fall by 2–3 weeks.

This predictive window can be used by Occupational Therapists, community teams and reablement services to trigger strengthening exercises, hydration support, medication review or environmental adaptations before an accident occurs. The intent is aligned with UK policy direction: maintaining independence, supporting carers and reducing avoidable pressure on acute and long-term care.

For Councils & Local Authorities

Extending independent living tenure by even 6–12 months supports duties under the Care Act 2014 to promote wellbeing and prevention, while easing demand on residential placements and enabling reinvestment in community services.

For Care & Reablement Agencies

Objective risk and recovery signals enable acuity-led rostering instead of purely time-and-task models. This supports better outcomes, safer care and structured evidence for CQC inspections under “Safe”, “Effective” and “Well-led”.

For Families & Informal Carers

A simple “Green / Amber / Red” view of activity and stability, without cameras, supports carers’ peace of mind while respecting the person’s privacy and autonomy – consistent with UK GDPR principles of data minimisation and dignity in care.

// 02_INTEROPERABILITY_LAYER (NHS HOME-BRIDGE ENGINE)

The NHS “Home-Bridge” Engine.

A persistent challenge in the UK is the “black hole” of information between acute discharge and community follow-up. Patients are discharged with standard instructions, yet clinicians and community teams have limited visibility of recovery until a planned contact – or an unplanned readmission.

The Hakilix Home-Bridge Engine is an interoperability and summarisation layer designed to convert raw, unstructured sensor fusion data into structured recovery signals suitable for NHS and local authority workflows.

It is architected to produce FHIR-aligned JSON structures and virtual ward-friendly summaries, enabling safe consumption by existing systems subject to future integration and governance agreements.

Data Flow Architecture

Edge Acquisition & Feature Extraction

4D mmWave radar and radiometric thermal sensors capture volumetric and heat signatures. The edge node performs local filtering, background subtraction and temporal smoothing, then extracts key biomarkers (velocity, posture transitions, room occupancy patterns) on-device, reducing data volume and privacy risk.

Home-Bridge Encryption & Normalisation

Biomarkers are encrypted in transit and at rest, wrapped in compact health resources. No raw images or video ever leave the home; only de-identified numerical representations travel over the wire, supporting UK GDPR and Data Protection Act 2018 obligations.

Clinical Integration & Alerting

Data is designed to be delivered via secure APIs into existing analytics layers and virtual ward tooling, surfaced as risk flags and trend lines. This aligns with NHS Digital Technology Assessment Criteria (DTAC) expectations around interoperability, clinical safety and cyber security, and conceptually maps to FHIR-based resources.

Feedback & Continuous Calibration

When clinicians or community teams act on an alert (e.g. medication review, physio referral), the system can incorporate that outcome as a feedback signal. Over time, the twin can be calibrated to local populations and care models across ICS / ICB footprints.

At TRL 4, this is an architectural and prototyping description intended to support research collaboration and structured evaluation with NHS and social care partners – not a claim of live clinical deployment or full FHIR integration today.

// 03_COGNITIVE_SAFEGUARDING

Dignified Dementia Care at Home.

For people living with dementia, home can be both sanctuary and source of risk. Traditional tracking methods (GPS watches, lock-based approaches) can be restrictive and distressing. Hakilix is positioned as a low-friction, privacy-preserving adjunct to community dementia care, with design principles consistent with UK best practice on restraint, autonomy and safeguarding.

Exit-Seeking Prediction

The Digital Twin looks for early patterns such as pacing near exits or unusual restlessness, enabling supportive interventions before a person reaches a point of high risk – consistent with least-restrictive practice and safeguarding duties.

Circadian Rhythm & Night-Time Risk

Sleep disruption, night-time wandering and frequent bathroom visits can be surfaced as patterns to inform GP, memory clinic or community mental health reviews, supporting personalised, evidence-informed care plans.

Zero-Stress Monitoring

No cameras. Nothing to wear. The person can simply live in their home, while risk is monitored in the background. This supports CQC expectations of dignity and privacy and UK GDPR requirements around necessity and proportionality.

// 04_SYSTEM_ARCHITECTURE_OVERVIEW

Theoretical Model: Neuromorphic Predictive Edge AI

Hakilix moves beyond standard cloud-centric deep learning. The system uses a bio-inspired, temporal computing approach at the edge. Instead of streaming high-volume raw data to the cloud, 4D mmWave radar and radiometric thermal signals are transformed into compact spatio-temporal signatures directly on the device.

At the core is a lightweight, neuromorphic-style model that maintains a continuously updated baseline of “normal” activity for each room and individual. Deviations – such as subtle slowing of gait, increased time-to-stand or fragmented sleep are modelled as changes in risk over time rather than isolated alarms. This enables:

Self-supervised adaptation to each home without manual labelling.
Probabilistic risk scoring rather than binary “alarm / no alarm”.
Explainable summaries aligned with clinical decision-making.

The edge logic draws on Spiking Neural Networks (SNNs), including Leaky Integrate-and-Fire (LIF) dynamics, and is informed by Active Inference principles for handling uncertainty and context over time. The result is a neuromorphic, edge-first pipeline that is energy efficient and suitable for UK housing stock, including properties with variable connectivity.

System Components

>>
SENSORS: 4D mmWave radar and radiometric thermal sensing, tuned for indoor environments and capable of operating in darkness, steam or smoke, providing volumetric and heat signatures without revealing identity.
>>
EDGE COMPUTE: Dedicated neural processing on the device for real-time inference, with design targeting low-power edge accelerators (including edge TPUs), suitable for running LIF-style SNN models and temporal feature extraction.
>>
COMMS LAYER: Resilient connectivity with optional cellular fallback and encrypted telemetry, designed with NHS DSP Toolkit and cyber security principles in mind.
>>
POWER & FAILSAFE: Mains-powered with local health monitoring and graceful degradation patterns, ensuring that failure states are visible rather than silent – important for safety case development (e.g. DCB 0129 / 0160 style thinking).
>>
PLATFORM SERVICES: Secure fleet management for remote updates, onboarding and decommissioning, enabling councils, housing providers and care organisations to manage thousands of nodes with appropriate role-based access controls and auditability.

Strategic Tech Principles · Edge-First & Privacy-by-Design

The architecture is explicitly edge-first and privacy-by-design, supporting UK GDPR, NHS DTAC expectations and CQC’s focus on safe, effective and well-governed use of technology in care.

Edge Layer

On-device neuromorphic inference for falls and mobility risk.
Configurable sensitivity profiles for dementia, reablement and virtual ward cohorts.
Robust behaviour in low-bandwidth or intermittent connectivity scenarios.

Privacy & Interop Layer

Data minimisation: only abstract motion biomarkers leave the home.
Encryption in transit and at rest, with pseudonymised identifiers.
API patterns designed to align with FHIR-aligned resources and NHS interoperability guidance.

Operations & Scale-Out (UK)

Multi-organisation tenancy (ICS, Trust, Local Authority, provider).
Configurable governance and consent flows per locality.
Audit-ready logging to support inspections, incident review and research governance.

Hakilix Core · v4.5 Platinum · Key Features

Neuromorphic Edge Inference

Leaky Integrate-and-Fire (LIF)-style models and spiking neural network concepts running at the edge, with design targeting low-power accelerators (including edge TPUs). The current TRL 4 prototype demonstrates this via an interactive neuromorphic simulation, illustrating the behaviour of LIF-style neurons under noise and thresholds.

Privacy-by-Design

No optical cameras. No raw audio or video leaves the home. Only encrypted telemetry and abstracted biomarkers are exported, supporting UK GDPR, Data Protection Act 2018 and ICO privacy-by-design expectations.

NHS Home-Bridge Engine

An interoperability concept that converts edge biomarkers into FHIR-aligned JSON structures and virtual ward-friendly summaries. At TRL 4 this is a prototyping layer, designed so the platform can be taken through NHS DTAC and formal integration work with virtual ward and community systems in the future.

Care Command Center

A real-time Care Command Center dashboard for agencies (domiciliary, reablement, intermediate care), surfaced in this prototype as the Tactical Radar Dashboard with working login, stateful UI and simulated caseload risk triage evidencing product maturity beyond a static landing page.

Technical Stack · Prototype & Research

Frontend

HTML5 single-page interface.
Tailwind CSS for responsive, utility-first styling.
Canvas-based twin visualisation in this prototype.
Three.js reserved for future holographic Digital Twin visualisation (design-aligned, planned for subsequent iterations).

Backend Simulation & Data

Python-based mock services (design and prototyping layer) for simulating edge events and recovery signals.
FHIR-style JSON structures defined for potential future integration with NHS virtual ward and analytics systems.
Browser-based telemetry simulation for this public prototype.

Edge Logic & AI

Spiking Neural Networks (SNN) concepts with LIF dynamics for neuromorphic inference.
Active Inference-inspired framing for handling uncertainty and context over time.
Edge-first deployment model to minimise cloud dependence and support UK privacy and latency requirements.

The stack is described at a TRL 4 prototype level: it captures the architectural intent and current prototype capabilities without implying live integration into NHS systems today.

// 05_EVIDENCE_PROFILE_UK

Innovation, Product Maturity & Scalability – for a UK Context.

Hakilix is articulated as a UK research-led platform, with a roadmap that supports academic evaluation, NHS / local authority pilots and professional endorsement processes.

Innovation (Neuromorphic & Edge-First)

Innovation is evidenced by the neuromorphic, edge-first architecture and LIVE_TWIN_FEED v2.1 telemetry: temporal bio-digital twinning, LIF-style spiking simulation and camera-free sensing – beyond traditional telecare or camera-based solutions.

Product Maturity (Working Login & Dashboard)

Product maturity is demonstrated by a working Care Command Center login and Tactical Radar Dashboard, along with the LIVE_TWIN_FEED and interactive calibration sandbox. This is more than a static site – it is a TRL 4 prototype suitable for research and stakeholder review.

Scalability (NHS & UK Social Care)

Scalability is framed in terms of NHS virtual wards, community services, reablement and local authority housing. Architecture and governance are designed for deployment across Integrated Care Systems and provider networks within the UK.

Product & Evaluation Trajectory (UK)

PHASE 01 Concept & Early Prototyping Formalising architecture, validating 4D mmWave + thermal sensing, building simulation tooling.

PHASE 02 Edge Intelligence & Twin Model Refining LIF-style SNN inference, risk scoring, explainability and safety case thinking.

PHASE 03 Controlled UK Pilots Working with NHS and local authority partners on small-scale, ethically governed deployments.

PHASE 04 UK Scale-Up Scaling across ICS / ICBs and providers, with policy-aligned governance and formal evaluations.

This site provides a concise technical narrative of a TRL 4 prototype. A structured Evidence Pack (PDF) – including detailed technical notes, governance artefacts, and roadmap material can be downloaded or shared with assessment bodies, research groups and innovation programmes.

Strategic Impact Analysis (UK)

£16k

Potential Saved Per Hip Fracture Avoided

Early detection of gait degradation may prevent avoidable falls, preserving independence while relieving acute and post-acute spend associated with hip fracture pathways.

Capacity

Unlocking Community & Virtual Ward Capacity

Reliable, passive monitoring supports safer early discharge and remote follow-up, reducing avoidable readmissions and supporting urgent and emergency care recovery plans.

24/7

Autonomous Coverage

Always-on monitoring without the constraints of rota patterns providing a safety net that can complement human observation and professional judgement.

Original Technical Approach

Hakilix combines 4D mmWave radar, radiometric thermal and behavioural signals into a unified, edge-native Digital Twin. The goal is not “more data”, but better signals – sparse, clinically interpretable biomarkers that can be trusted in safety-critical UK settings.

Designed for UK Ageing & Reablement

The platform is built around the realities of UK ageing, multimorbidity, workforce pressure and constrained budgets – with deployment models for social housing, extra care, reablement and virtual ward pathways.

Policy- & Governance-Ready

Privacy-by-design, explainable risk signals and explicit mapping to NHS, UK GDPR and CQC expectations position Hakilix Core as a candidate for responsible evaluation within UK health and social care.

// 06_SYSTEM_SUPERIORITY

Feature	Legacy Pendants	Computer Vision	HAKILIX CORE
Activation	Manual (must press button)	Passive	Passive & Predictive (Digital Twin)
Privacy	High	Low (video inside home)	Camera-free (4D radar / thermal only)
Bathroom Use	Often removed (not waterproof)	Ethically constrained	Full coverage with dignity
Dementia Suitability	Very low (forgotten/discarded)	Medium	High (zero interaction)
Data Insight	Binary alarms only	Rich but heavy data	Continuous biomarkers & Digital Twin trends

// 07_SYSTEM_CALIBRATION · NEUROMORPHIC DEMO

Interactive LIF-Style Tuning.

Adjust the system sensitivity parameters to see how a simplified, LIF-style spiking neuron responds to noise, thresholds and baseline drift. This illustrates how Hakilix Core’s neuromorphic edge logic can remain stable and informative in realistic UK home environments.

SIGNAL_STABILITY (DECAY) 0.95

ALERT_THRESHOLD (LIF SPIKE) 1.0

This sandbox shows a simplified Leaky Integrate-and-Fire neuron running in the browser as a conceptual analogue of the SNN / Active Inference edge logic planned for hardware-accelerated deployments.

REAL-TIME NEUROMORPHIC TRACE

LIVE

// 08_SAFETY_GOVERNANCE (UK POLICY)

Privacy, Safety & Responsible AI – for UK Health & Social Care.

Privacy-by-Design Architecture

Hakilix Core is architected around data minimisation: only abstracted motion biomarkers leave the home. No raw video, audio or identifiable images are transmitted. This supports UK GDPR and the Data Protection Act 2018, alongside ICO expectations for privacy-by-design and default.

Safety-Critical Thinking

The system is being designed with safety envelopes, graceful degradation and explicit “safe failure” modes. This is consistent with the clinical safety mindset in DCB 0129 / 0160 and NHS DTAC, even at this early TRL 4 prototyping stage.

Clinical & Carer Co-Design

The intent is to build with the ecosystem, not just for it – embedding feedback from clinicians, carers, local authorities and families so that the twin remains clinically relevant, usable and humane within UK regulatory and professional frameworks.

Indicative Alignment with UK Policy & Regulation

NHS & ICS / ICB

Supports virtual ward & ageing-well objectives.
Designed for integration with local clinical governance.
Intended to be assessed via NHS DTAC.

UK GDPR & ICO

Data minimisation and purpose limitation.
Encryption, pseudonymisation and access control.
DPIA and information governance as core artefacts.

CQC & Social Care

Evidence for “Safe, Effective, Caring, Responsive, Well-led”.
Supports person-centred, least-restrictive practice.
Can underpin quality improvement narratives.

Research & Ethics

Structured protocols for academic and NHS ethics review.
Clear separation between R&D, pilot and live service use.
Suitable for inclusion in endorsement and innovation dossiers.

Alignment refers to design intent and governance planning at TRL 4, not current certification or deployment status. Formal assessments would be undertaken with partners as part of pilots and scale-up.

// 00_PRIMARY_USE_CASES (UK)

Designed for Real UK Care Pathways.

Dementia Client Living Independently

Reablement & Intermediate Care Client

New Hospital Discharge & Virtual Ward Patient