Our Platform

One integrated platform.
Three interconnected layers.

Lyvup's technology stack combines a sensor layer, an intelligence layer, and a digital therapeutics layer — each independently valuable, together forming the infrastructure for continuous, personalised chronic care.

Architecture

Measure. Understand. Adjust.

01
Sensor Layer

Lyvup Ring

Non-invasive continuous glucose monitoring in a ring form factor. The raw metabolic signal that feeds every intelligence and intervention capability above it. No needles, no patches, no clinical infrastructure required.

02
Intelligence Layer

Lyvup Context & Lyvup Predict

Two AI engines that interpret sensor data. Lyvup Context understands why glucose responds as it does — modelling sleep, stress, activity, and nutrition simultaneously. Lyvup Predict determines which patients respond to which therapies before treatment begins.

03
Intervention Layer

Lyvup Recover & Lyvup Coach

Validated digital therapeutics that translate sensor data and AI predictions into personalised clinical action. Lyvup Recover targets cardiac rehabilitation and relapse prevention. Lyvup Coach delivers continuous AI-driven behavioural coaching. Lyvup Compliance monitors and restores engagement with digital therapeutics — condition-agnostic, continuously.

Coming Soon
Sensor Layer

Lyvup Ring

In development

Own-device niCGM — where no suitable sensor exists

Lyvup's platform is sensor-agnostic by design — integrating with the full range of connected devices patients already use. Lyvup Ring is Lyvup's own sensing addition for cases where continuous glucose pattern data is clinically valuable but no suitable device is available. It is a non-invasive continuous glucose monitor in ring form — and the first of its kind designed specifically for behavioural monitoring rather than clinical dosing. Where conventional CGM development targets the accuracy required for closed-loop insulin delivery, Lyvup Ring is built for a different application: tracking how glucose behaviour changes in response to sleep, stress, activity, and intervention.

This distinction is clinically and technically significant. Behavioural glucose monitoring does not require the absolute accuracy of a Freestyle Libre equivalent — it requires sensitivity to patterns and relative change. That is a more achievable specification, and a more relevant one for the behavioural health management applications at the core of Lyvup's platform.

Sensing
NIR spectroscopy + IMU + temperature
Motion-rejection filtering · Thermometric perfusion compensation
Validation target
MARD <10% vs venous reference
ISO 15197-derived · MDR Class IIa pathway · Amsterdam UMC
Intelligence Layer — 01

Lyvup Context

Research-stage

Contextual Metabolic Intelligence

Lyvup Context addresses a fundamental limitation of current metabolic monitoring: raw glucose values in isolation are clinically ambiguous. The same person eating the same meal will respond differently depending on sleep quality, stress load, activity patterns, and dozens of contextual micro-factors.

By fusing continuous glucose pattern data from Lyvup Ring with behavioural signals, HRV-derived stress indices, activity levels, and sleep architecture, Lyvup Context generates a predictive model of individual metabolic response. The output is not a glucose value — it is an explanation of why glucose is behaving as it is, and what is likely to happen next.

Input signals
Multimodal contextual fusion
Continuous glucose · Activity and movement (IMU) · Sleep stage and quality · HRV-derived autonomic stress · Nutritional logging (optional) · Location and ambient context
Output
Personalised metabolic prediction
Individual glycaemic response models · Contextual trigger identification · Prediction of upcoming excursion risk · Input to Lyvup Predict stratification · RWE data stream for pharmaceutical partners
Intelligence Layer — 02

Lyvup Predict

Research-stage · ZonMW funded

Precision Treatment Stratification

The central pharmacological challenge in metabolic disease is responder heterogeneity: why two patients with identical diagnoses and identical prescriptions achieve radically different outcomes. Lyvup Predict is built to answer that question before treatment begins.

By integrating pharmacogenomic markers, metabolic phenotype data from Lyvup Context, lifestyle and behavioural profiles, and longitudinal treatment outcome data, Lyvup Predict generates individual response probability estimates for specific therapeutic agents. The initial validation focus is GLP-1 receptor agonists, conducted with Amsterdam UMC, TU Delft, TU Eindhoven, and UvA — funded by ZonMW.

Key concept — Pharmacogenomics (PGx)

Pharmacogenomics is the study of how an individual's genetic profile influences their response to a drug — including metabolism rate, therapeutic efficacy, and risk of adverse effects. PGx variants in genes such as CYP450 enzymes determine how quickly a patient breaks down a given compound, directly affecting optimal dosing and the probability of a clinical response. Lyvup Predict integrates PGx data as one layer of a broader stratification model — providing the molecular basis that contextual and behavioural data alone cannot supply.

Stratification inputs
Multi-layer phenotyping
Pharmacogenomics (PGx) — genetic variants that predict drug metabolism, efficacy, and adverse response · Continuous metabolic phenotype (Lyvup Context) · Digital behavioural phenotype · Anthropometric and clinical baseline · Historical treatment response data
Output
Pre-treatment response probability
Individual responder probability per therapeutic agent · Confidence-weighted recommendation · RWE capture for ongoing model refinement · Adherence risk stratification
Initial indication: GLP-1 RA in T2D / obesity. ZonMW-funded. Partners: Amsterdam UMC · TU Delft · TU/e · UvA.
Intervention Layer — 01

Lyvup Recover

Clinical validation · Horizon Europe

Digital Therapeutics for Rehabilitation & Relapse Prevention

Cardiac rehabilitation has a well-documented efficacy gap between supervised programme completion and long-term behavioural maintenance. Most patients who complete structured rehabilitation experience significant behavioural regression within six to twelve months — not because the programme failed, but because the transition from supervised care to unsupported self-management is unmediated.

Lyvup Recover is a validated digital therapeutic that extends the clinical effect of cardiac rehabilitation beyond the point of discharge — providing continuous monitoring, early relapse detection, and evidence-informed digital coaching grounded in behavioural psychology. Currently the subject of a Horizon Europe-funded clinical validation study, in which Lyvup serves as Principal Investigator and Project Coordinator.

Clinical function
Continuous monitoring + relapse prevention
Remote patient monitoring via wearables · Early warning signal detection · Clinician-facing dashboard · Personalised behavioural coaching adapted to individual motivational profile · Longitudinal outcome tracking for RWE
Horizon Europe-funded. Lyvup: Principal Investigator & Project Coordinator.
Regulatory status
Clinical validation in progress
MDR-aligned development pathway. Clinical evidence generation ongoing across European consortium sites. Integrates into existing hospital and rehabilitation workflows without additional clinical infrastructure.
Intervention Layer — 02

Lyvup Coach

In development

Continuous AI-Driven Behavioural Coaching

Therapeutic adherence failure is a behavioural problem — not a motivational deficit in the patient, but a failure of the intervention to meet the individual where they are, when they need it. Generic reminder systems do not solve this. Lyvup Coach is built on a different premise: that effective digital coaching requires deep contextual awareness, individual motivational modelling, and intervention delivery timed to moments of maximum receptivity.

Lyvup Coach operates as the patient-facing delivery layer — connecting sensor data from Lyvup Ring, contextual intelligence from Lyvup Context, and stratification output from Lyvup Predict into a coaching experience that adapts in real time to each individual's behavioural state, context, and progress trajectory.

Behavioural science foundation
Individual motivational modelling
Dynamic motivational profiling · Context-sensitive intervention timing · Evidence-based frameworks (self-determination theory, implementation intentions, habit formation) · Dropout risk detection and proactive re-engagement
Technical architecture
Scalable across conditions and languages
Condition-agnostic coaching engine deployable across metabolic, cardiovascular, and mental health indications · Multilingual · Accessible across age groups and digital literacy levels · White-label integration for pharmaceutical and payer programmes
Intervention Layer — 03

Lyvup Compliance

In development

Continuous Compliance Monitoring, Analysis & Recovery

Long-term compliance with digital therapeutics is the single most consequential determinant of clinical outcome — and the most commonly unmonitored. Most DTx programmes assume that a patient who has started a programme is continuing it. Lyvup Compliance is built on the opposite assumption: that compliance is dynamic, highly individual, and continuously at risk.

Lyvup Compliance continuously monitors each patient's engagement with their digital therapeutic programme — tracking behavioural signals, interaction patterns, and contextual factors to detect early-stage disengagement before it becomes dropout. When compliance risk is identified, the system does not simply alert: it analyses the underlying pattern, identifies the most probable behavioural driver, and generates a personalised recovery strategy delivered through the appropriate intervention channel.

Critically, Lyvup Compliance is condition-agnostic. The same monitoring and recovery architecture that operates within Lyvup Recover for cardiac rehabilitation operates identically for metabolic care, mental health, and any other therapeutic domain. Compliance with a digital therapeutic programme is a behavioural phenomenon — it follows the same patterns regardless of the underlying diagnosis.

What it monitors
Continuous behavioural compliance tracking
Programme interaction frequency and quality · Sensor data engagement patterns · Self-reported symptom and wellbeing signals · Deviation from established individual baseline · Contextual factors associated with disengagement (stress load, schedule disruption, life events)
What it generates
Personalised recovery strategies
Early disengagement detection — before dropout occurs · Root cause analysis of compliance failure pattern · Personalised recovery strategy matched to individual motivational profile · Automated micro-intervention delivery through Lyvup Coach · Clinician alert for cases requiring human escalation
Scope
Condition-agnostic by design
Lyvup Compliance operates across all therapeutic domains supported by the platform — cardiac, metabolic, mental health, and beyond. The underlying compliance architecture is independent of indication, making it a horizontal capability that scales across every Lyvup deployment.

Interested in the platform for research, clinical, or commercial applications?