Resonix MedicalStem Cells · Cardiac · Bioprocess

Computational infrastructure for life sciences and stem cell manufacturing.

Noise-stable ODE solvers for physiological simulation. Bioreactor digital twins that survive real sensor data. Deterministic arithmetic for regulated medical devices — with every dead zone documented.

Medical capabilities Product line Talk to our engineering team

5/5: Medtech POCs CONFIRMED
450: Validation runs
100%: SolvSRK survival rate
0%: BDF / Radau survival

Computational validation — not clinical

Every result referenced on this page is a computational validation on published ODE models. These are not wet-lab results, clinical trials, or regulatory approvals. No FDA clearance, CE marking, or clinical outcome is claimed. All five medtech proofs of concept have been validated under the current SolvSRK architecture and triaged as CONFIRMED with Grade A evidence.

Proofs of concept

Five CONFIRMED medtech benchmarks.

See all POCs

POC 02

Stem Cell Bioreactor Digital Twin

Bioreactor metabolic ODE (S1, dim=4). SolvSRK: 30/30 survived, 20ms median. BDF: 0/30. Radau: 0/30. CONFIRMED.

30/30: SolvSRK seeds survived
0/30: BDF seeds survived
20ms: SolvSRK median wall time
974: SolvSRK median nfev

CONFIRMED

POC 01

Cardiac Hemodynamics Under Sensor Noise

Windkessel-3 cardiac model (S2, dim=3). SolvSRK: 30/30 survived, 63ms median. BDF: 0/30. Radau: 0/30. CONFIRMED.

30/30: SolvSRK seeds survived
0/30: BDF seeds survived
0/30: Radau seeds survived
63ms: SolvSRK median wall time

CONFIRMED

POC 03

Neural Tissue Network Simulation

100-neuron FHN network (S1, dim=200). SolvSRK: 30/30 survived, 380ms median. BDF: 0/30. Radau: 0/30. CONFIRMED.

30/30: SolvSRK survived (200-dim)
0/30: BDF seeds survived
380ms: SolvSRK median wall time
200: State variables (dim)

CONFIRMED

Medical capabilities

Five capabilities. Every dead zone documented.

Noise-stable solvers, deterministic arithmetic, and audit-grade evidence across physiological simulation, bioprocess, spectral analysis, and regulatory traceability.

Physiological Simulation

Noise-stable integration for cardiac and neural tissue models

Standard implicit solvers depend on converged Jacobian estimates. When the right-hand side includes stochastic terms — channel noise in cardiac tissue, synaptic noise in neural networks — the Jacobian becomes noisy and convergence fails. SolvSRK uses a proprietary noise-conditioning architecture that tolerates stochastic function evaluations natively. In dedicated medtech validation (450 runs), SolvSRK achieved 100% survival on all five medtech problems — cardiovascular Windkessel-3, 100-neuron FHN network, 11-compartment PBPK, bioreactor, and single FHN neuron — while BDF and Radau achieved 0% across all problems at sigma=0.001. CONFIRMED with Grade A evidence.

SolvSRK

Bioprocess Digital Twin

Same solver on the bioreactor controller and the cloud twin

Bioreactor digital twins fail for the same reason factory twins fail: the on-device model and the cloud model run different solvers on different hardware. A metabolite divergence alert could mean a real process deviation or a numerical artifact. SolvSRK runs the same native binary on both sides, so divergence means a real change in the culture. In dedicated medtech validation, the bioreactor metabolic ODE (S1, dim=4) achieved 100% SolvSRK survival across 30 seeds (median 20ms, 974 nfev) while BDF crashed in <200ms and Radau timed out at 10s. CONFIRMED with Grade A evidence.

SolvSRKSolvNum

Regulatory Determinism

Bit-identical arithmetic across hospital hardware for FDA/IEC 62304 traceability

A clinical AI model that produces different last digits on the CT workstation vs the cloud inference server vs the edge gateway is a regulatory risk under FDA guidance and IEC 62304. SolvNum produces bit-identical arithmetic across x86, ARM, GPU, and WebAssembly. One SHA-256 receipt proves every deployment agrees. This is a computational infrastructure capability — it does not constitute FDA clearance, CE marking, or any regulatory approval.

SolvNum

Solver Selection

Pick the right solver for your computational biology model — with evidence

Every simulation package ships with a default solver, and most researchers never change it. When a physiological model fails to converge or produces suspicious results, the first question is whether the solver was appropriate for the system's stiffness profile. SolvScout fingerprints the ODE system and benchmarks every candidate solver against it — dead zones included, not hidden. SolvBench archives the decision for regulatory or peer review.

SolvScoutSolvBench

Health Claims Enrichment

CMS/X12-aligned bounded-signal enrichment for clinical claims

MoridainHIPAA is an in-path bounded-signal enrichment layer for CMS-aligned health claims (X12 837). It applies deterministic fraud and anomaly signals to each claim in transit — with signed provenance on every enrichment decision. The enrichment is bounded: every signal has a documented upper bound on false-positive rate, and every decision is attestable. This is claims-processing infrastructure, not a clinical decision tool.

Moridain

Want to see the numbers on your own model?

We offer a bounded-scope benchmark against your physiological model or bioprocess ODE. Two weeks, fully credited toward a Year-1 license. Every numerical claim traces back to a test ID and a verdict.

Talk to us

The product line

The medical product line on top of one computational engine.

Each product brief below opens with a life-sciences scenario, then walks through what it is, who uses it, and how the claims survive regulatory and peer review.

Solver

SolvSRK for Physiological Simulation and Bioprocess Control

The ODE solver that survives stochastic physiology.

Cardiac tissue models with channel noise. Bioreactor metabolic ODEs with noisy dissolved-oxygen sensors. Neural network models with synaptic noise. Standard implicit solvers crash when the right-hand side is stochastic — SolvSRK's proprietary noise-conditioning architecture tolerates it natively. CONFIRMED on 5 medtech problems (450 runs): 100% SolvSRK survival vs 0% BDF/Radau on cardiac hemodynamics, stem cell bioreactors, neural tissue networks, pharmacokinetics, and single-neuron dynamics.

Open product brief

Four deployment targets — workstation, edge, cloud, audit — hashing to the same SHA-256.

Solver

SolvNum — Deterministic Arithmetic for Regulated Devices

Same input, same output, every hospital, one hash.

A clinical AI model that produces different results on different hardware is a regulatory liability. SolvNum is a deterministic real-number primitive that produces bit-identical arithmetic across x86, ARM, GPU, and WebAssembly. One SHA-256 receipt proves every deployment agrees — a computational infrastructure capability for FDA/IEC 62304 traceability.

Open product brief

System fingerprint and a ranked benchmark table comparing every candidate solver on the user's actual physiological model.

Developer tool

SolvScout

Pick the right solver for your physiology model — with evidence.

SolvScout fingerprints your ODE system's stiffness, dimension, and dynamics profile, benchmarks every candidate solver against it, and produces a ranked, dead-zone-honest comparison report. Replaces the 'we used the default solver' conversation that does not survive peer review or regulatory audit.

Open product brief

Claims pipeline with bounded enrichment signals and signed provenance at each hop.

Library

Moridain — Bounded Intelligence for Health Claims

Deterministic enrichment on CMS/X12 rails.

MoridainHIPAA applies bounded fraud and anomaly signals to CMS-aligned health claims (X12 837) in transit. Every enrichment decision carries signed provenance and a documented false-positive bound. Claims-processing infrastructure — not a clinical decision tool.

Open product brief

Engagement model

Built for teams that have to survive regulatory review.

Bounded-scope benchmark on your actual physiological model in two weeks. Shadow run alongside your existing solver for one month. Every option credits in full toward a Year-1 license. Every numerical claim traces back to a test ID, a seed, and a verdict in a live evidence register.

Talk to our engineering team