TECHNOLOGY & INNOVATION

Applied, not invented. Engineered where it matters.

Uber didn't invent GPS. RONKE doesn't invent voice AI. The innovation is the clinical application: turning a telephone conversation into structured, explainable, NHS-grade clinical assessment.

THE PIPELINE

From scheduled call to clinical action, in five stages.

STAGE 01

Scheduled outreach

A discharge coordinator registers the patient at discharge, about two minutes. The system then schedules regular calls automatically across the post-discharge window.

STAGE 02

Natural conversation

Conversational voice AI conducts the 7-question assessment in plain speech, engineered for older patients: slower speech, short sentences, confirmation loops, and adaptations for hearing loss.

STAGE 03

Clinical extraction

Natural speech becomes structured data: symptoms, weight, adherence, red-flag phrases, mapped to validated clinical indicators.

STAGE 04

Risk stratification

Base risk + current symptoms + temporal weighting + trend across calls: an explainable Green / Amber / Red score.

STAGE 05

Intelligent triage

A structured report lands with the nurse within five minutes of the call ending. Green continues on schedule; amber queues for review; red alerts immediately.

THE 7-QUESTION FRAMEWORK

Seven questions. Every one earns its place.

Mapped to NICE Guideline NG106, with every change to the clinical logic reviewed and signed off by our Clinical Safety Officer before it can go live. Not "how are you feeling?", specific probes for the signals that predict readmission.

If it works for an 85-year-old with arthritis and hearing difficulties, it works for everyone.

Breathing

Breathlessness at rest and on exertion, orthopnoea, paroxysmal nocturnal dyspnoea: the earliest predictors of decompensation.

Swelling

Peripheral oedema in ankles, legs and feet: its extent, and whether it is progressing.

Weight change

Change since discharge. Rapid gain of 2–3kg over a few days is the red flag for fluid retention.

Energy levels

Functional capacity and daily activity tolerance, compared with the discharge baseline.

Chest symptoms

Chest pain, pressure or tightness, at rest or on exertion, screened on every call.

Medication adherence

Missed doses, confusion, side effects, supply issues: a leading, fixable cause of readmission.

Overall wellbeing

An open question, deliberately last: it surfaces what structured questions miss, and preserves dignity.

RISK STRATIFICATION

Explainable by design. Try it.

The score is multi-factor: baseline risk from clinical history, current symptoms, how many days since discharge, and the trend across calls. Move the inputs and watch the reasoning, not just the result.

TEMPORAL WEIGHTING: the same symptom scores higher on day 4 than day 24.

LIVE MODEL · ILLUSTRATIVE

Weight change (3 days)+1.5 kg

Breathlessness changeslightly worse

Days since dischargeday 4

RED

Immediate clinical escalation

Pattern consistent with decompensation. Clinical team alerted for same-day intervention.

PLATFORM

NHS-grade by architecture, not afterthought.

INTEGRATION

Speaks FHIR R4, fits the record

Standards-based FHIR R4 export, with findings encoded using NHS-number identifiers and SNOMED CT so they slot into the patient record clinicians already use. Every export is role-controlled and audit-logged.

SECURITY

Encrypted end to end

TLS 1.3 in transit, AES-256 at rest, role-based access, full audit trail, UK-only data storage. UK GDPR aligned; developed to DCB0129 principles, with DSPT and a DPIA in preparation.

ROADMAP · NEXT

Beyond heart failure

The same voice-first pathway extends to other long-term conditions where post-discharge monitoring is just as scarce: COPD, diabetes and kidney disease.

See the evidence behind the engineering.

NICE-derived protocols, validation strategy, and the clinical case.

Clinical evidence →