The lung,
simulated breathing.
Medlea turns a chest CT into a functional digital twin. AI agents rebuild the patient’s lungs; a physics engine on GPU makes them breathe. Airflow, ventilation and spirometry, computed on their own anatomy.
- Agentic AI workflow
- GPU multiphysics
- Patient-specific CFD
- Confidential computing
- FEV1
- 3.20 L
- FVC
- 4.38 L
- FEV1/FVC
- 72.9 %
- 1Your team defines the pattern
- 2Clinicians annotate
- 3Training on your GPU
- 4The model proposes, you correct
- Physics, not statistics
Airflow and ventilation are computed on the patient’s airways, not inferred from a population.
- One patient at a time
Every number on screen belongs to this anatomy, this scan, this breath.
- Inside the hospital
Reconstruction, training and simulation run on your GPU, in the cloud or on premises.
Working with
Clinical question, engineering answer
Imaging shows the shape. The twin shows the breath.
The same simulation methods used to design aircraft and engines, run on a single patient’s airways, inside the hospital.
- 01 · Agentic AI
From the scan
AI agents read the CT and clinical record, then segment the lobes and the airway tree, case by case.
- 02 · Physics on GPU
To the breath
A lattice-Boltzmann flow solver makes the reconstructed lung breathe: airflow in every airway, ventilation in every region.
- 03 · Morphofunctional biomarkers
To the numbers
Spirometry per lobe, airway resistance, ventilated volume: measurements no scan shows on its own, comparable over time.
Two products, one twin
LeaRes
The functional twin at work: anatomy, regional ventilation, virtual spirometry, vascular reconstruction and the Active Learning loop, in the browser.
Explore LeaRes →
Platform
Your department names the pattern, annotates its own cases, and the model is trained inside your perimeter. The product is the machine that makes a model.
Explore the platform →- 1Your team defines the pattern
- 2Clinicians annotate
- 3Training on your GPU
- 4The model proposes, you correct
Evidence your Data Protection Officer can verify
Every virtual-spirometry run can leave a signed record of what went in, what ran and what came out: the software, the hardware it ran on, and the data it touched. Built with EQTY Lab, so your team can verify it independently.
- Software The exact pipeline version, fingerprinted and signed.
- Hardware Confidential-computing evidence from the machine that ran it.
- Data Input and output fingerprints. No patient data in the record.
Who it is for
Critical care
Explore ventilation settings on the patient’s own lung model, not on an average lung.
See use cases →Thoracic surgery
See how much each lobe contributes to the breath before planning a resection.
See use cases →Research & pharma
In-silico studies of drug response on lung structure and function, across cohorts.
See use cases →Built for a hospital’s data rules
Software, hardware and data fingerprints for every attested run, built with EQTY Lab.
How attestation works →Annotation, training and simulation run on the hospital’s own GPU, in the cloud or on premises.
How the platform works →The output is quantitative. Clinical judgement stays with the clinician.
About LeaRes →See a lung breathe on a case like yours
A walkthrough with our team, on a demo case or on your own data inside your perimeter.