Longitudinal state
Replica models change through time instead of treating every observation as an isolated snapshot.
Hale-X / Biological Twin Engine
Making biology computable.
Replica transforms fragmented longitudinal data into an auditable, individual model of biological state, trajectory and uncertainty.
Individual patient Longitudinal Traceable evidence
REPLICALiving systems map
SYNTHETIC MODELSelect a system to reveal its biological relationships.
Within rangeModerate deviationHigh deviationNot documented
SYNTHETIC MODEL : NOT A CLINICAL RESULT
CORE PLATFORM OPERATIONALFIT 4 START #16 GRADUATEAWS TECHNOLOGY COLLABORATIONBUILT IN LUXEMBOURG
The missing layer
Healthcare AI is learning to predict. It still lacks a model of the patient.
Most healthcare AI models learn patterns from fragmented records. Replica adds a persistent representation of the patient: what was observed, how biological state is changing and where uncertainty remains.
01RecordsPieces of evidence
≠02PatientA changing biological system
The result is a reusable patient model layer that connects longitudinal evidence, biological relationships and uncertainty across clinical applications.
From fragments to state
A living model, not a static snapshot.
Replica aligns clinical events and measurements through time, maps them to biological systems and reconstructs a patient state linked to evidence.
01Collect
Fragmented records
02Align
One patient timeline
03Map
Biological context
04Reconstruct
Current patient state
05Explain
Trajectory + evidence
One engine, focused applications
Replica is the biological layer.
Applications can ask specific clinical questions without rebuilding the patient model each time.
Clinical teams
Researchers
Life science partners
questions
INTERACTION LAYER
FOCUSED APPLICATIONSApplications for specific clinical domains, built on one patient model
Anali-XICU research programme
contextual query
BIOLOGICAL MODEL
REPLICABiological state of the individual patient
Current stateWhat the evidence supports now
TrajectoryHow the state is changing
UncertaintyConfidence and unresolved evidence
ProvenanceSources supporting each interpretation
state reconstruction
TIME LAYER
LONGITUDINAL ALIGNMENTOne patient timeline linked to evidence
Clinical recordsevents · diagnoses
Laboratorybiomarkers · trends
Vital signscontinuous signals
Medicationsexposure · response
Imaging metadatafindings · context
Wearables & contextbaseline · behaviour
Conceptual architecture of the reusable patient model layer and the focused applications built above it.
Interpretable output
Beyond a probability score.
Replica connects predictive signals to the state of the individual patient, supporting evidence, trajectory and uncertainty.
CONVENTIONAL OUTPUT
RISK0.78
A risk signal without a patient representation.
A probability can identify risk, but it does not by itself explain the biology of the individual patient, the supporting evidence or the uncertainty behind that signal.
versus
REPLICA OUTPUT
Evidence
State
Trajectory
Uncertainty
An interpretable patient state model.
Observed evidence, modelled state, trajectory and uncertainty remain distinct, traceable and available for expert review.
Designed for scrutiny
Clinical context is part of the output.
Traceable evidence
Every interpretation retains a visible connection to the observations, sources and timing that support it.
Inspectable interpretation
Observed facts, modelled state, trajectory and uncertainty remain explicitly separated for expert review.
First clinical research programme
Replica powers Anali-X.
Anali-X applies Replica to the research challenge of earlier, explainable recognition of ICU deterioration and sepsis.
REAL-TIME ICU SYSTEMSynthetic research illustration
FROM SIGNALS TO PATIENT CONTEXT01 · BEDSIDEOne patient. Many live signals.
ICU R:0042ECGLIVE
92 bpmSpO₂LIVE
96 % CONTINUOUS MONITORING DEIDENTIFIED
Arterial pressureSTREAM
118/64 mmHgVentilationLIVE
18 rpmBlood gasnew result
Medicationexposure updated
Lactatetrend changing
02 · HALE-XConnect the whole picture.
ACTIVEONE PATIENTContinuous
contextnow · before · next
TIMELINESYSTEMSEVIDENCEGAPScontextnow · before · next
01Synchronisesignals through time
02Connectsystems and events
03Interpretchange with evidence
03 · ANALI-X RESEARCH VIEWFollow the patient trajectory.
RESEARCHPatient state through timeLIVE WINDOW
−6 h−4 h−2 hNOW
STATEWhat is happening now?
TRAJECTORYWhat is changing?
EVIDENCEWhat supports it?
UNCERTAINTYWhat remains unresolved?
Clinician remains in control
Illustrative system architecture using synthetic data. Research stage. Not approved for clinical or diagnostic use.
RESEARCH STAGE
Retrospective ICU feasibility study in preparation. Protocol, funding and data governance work are underway; study execution and data access have not yet started.
Explore the programmeExecution & validation
From platform architecture to clinical evidence.
Core longitudinal patient state architecture is operational.
Selected for and graduated from Luxembourg’s leading startup acceleration programme.
Infrastructure work supports a governed and scalable path to deployment.
Retrospective study protocol, governance and funding preparation are underway.
Build with biological context