For cell-therapy modeling teams

Choose the mechanism you can defend.

Denspath compares plausible kinetic mechanisms, generates challenger models, and produces a reviewable evidence package before structural model decisions are locked.

Designed for CAR-T and adjacent cell-therapy studies
Human-in-the-loop, not autonomous replacement
Exports to NONMEM / Monolix / Pumas workflows
Fixed-scope pilot, agreed in kickoff
Review Package PreviewCAR-T CD19 pilot
Top Candidate
Biphasic expansion-contraction with persistence

Best overall fit with interpretable persistence behavior and cleaner late-phase diagnostics.

Challenger
Target-mediated disposition plus T-cell exhaustion

Retained because it explains early contraction differently and remains plausible on current data.

Key Disagreement
The leading candidates differ on persistence mechanism, contraction order, and target-mediated effect.
Recommended Next Experiment

Add late-phase sampling at day 60-90 to separate persistence mechanisms before locking the structural model.

Deliverables
ranked_candidates.pdfassumptions.csvdiagnostics.pdfnext_experiment.mdnonmem_export.ctl
Workflow Fit

Designed to slot into incumbent MIDD workflows rather than replace them.

Cell-therapy teams still hand-build mechanism choices

This isn't generic model selection. It's mechanism discrimination for CAR-T and adjacent cell-therapy workflows — where the wrong structural choice is expensive.

Cell-therapy kinetics do not fit the usual template

Expansion, contraction, persistence, exhaustion, and target effects can all matter. The hard part is often deciding which mechanism family is even plausible.

Teams still compare bespoke structures by hand

Senior modelers test a few mechanism variants manually, usually under time pressure. The result is often “best among what we tried,” not a systematic discrimination workflow.

The real burden is trust and review

Reviewers ask why one mechanism was selected over another. Diagnostics exist, but assumption differences, challenger models, and next-step evidence are still assembled manually.

Why this fits

Model-informed drug development is review-heavy by design. The harder part is upstream: deciding which mechanism family is even plausible, and explaining why one structural choice is defensible. That's where Denspath sits.

Discriminate. Compare. Recommend. Package.

One workflow, one buyer story, one output: a defendable mechanism package.

Generate plausible candidate mechanisms, not just one model

Denspath ingests the study data and generates candidate mechanism families plus challenger models that test alternative structural assumptions. The goal is not a magical single answer. The goal is a credible comparison set.

Discriminate
Compare
Recommend
Package
Input
CAR-T CK study — 2,847 observations, 24 patients
Output
7 candidate mechanism families ranked, including 2 challenger structures outside the initial starting set

What the team actually gets

A reviewable package that fits into the workflow already used for MIDD work.

Ranked Candidates
PDF / JSON

Candidate mechanism families with explicit assumptions, ranked by fit and diagnostic quality.

Challenger Models
PDF / JSON

Alternative structural forms generated to stress the leading explanation, not just confirm it.

Comparative Diagnostics
PNG / PDF

Side-by-side evidence showing where candidates agree, where they diverge, and where each one breaks.

Next Experiment Memo
Markdown / PDF

A concrete recommendation for the experiment most likely to resolve the remaining mechanism ambiguity.

Justification Narrative
PDF

A reviewable document explaining what was tested, what changed across models, and why the selected mechanism is defensible.

Workflow Export
.ctl / .mlxtran / .jl

Exportable package for the tools the team already trusts rather than a forced platform migration.

Example pilot outcome

A CAR-T cellular-kinetics program with ambiguous persistence behavior.

1

A modeling team has Phase I CAR-T kinetic data with rapid expansion, sharp contraction, and slow persistence. Standard small-molecule PK templates are not enough.

2

Today the senior modeler hand-tests a few structural variants, usually under time pressure. The comparison set is limited by what the team had time to propose, not by a systematic mechanism-discrimination workflow.

3

With Denspath, the team gets ranked candidate mechanisms, challenger models that stress the leading explanation, explicit assumption differences, and a recommendation for the next measurement that would actually separate the top two candidates.

Illustrative top candidate summary
dC/dt = r(t)·C·(1 - C/K) - δ·C + p·C·e^(-λt)
Biphasic expansion plus contraction plus persistence term. Challenger model adds target-mediated effect and exhaustion saturation.

Why this is different from the current workflow

Denspath is not a replacement for existing modeling tools. It is a better upstream mechanism-comparison workflow.

DenspathManual + template tools
Search spaceCandidate mechanisms plus challenger structuresTemplate variants inside a predefined family
Model comparisonExplicit structural assumption diffsMostly implicit in equations and notes
Uncertainty handlingCan recommend what to test nextUsually manual follow-up planning
Review packageDiagnostics plus assumptions plus justificationDiagnostics plus manual narrative
Workflow fitExports into incumbent toolsNative to one toolchain at a time

Sits upstream of NONMEM, Monolix, and Pumas

Denspath doesn't replace the estimation and review tools your team already trusts. It helps you lock a defensible structural model, then exports it into the workflow you use for fitting, identifiability, and simulation.

Frequently asked questions

Start with a pilot

The first engagement should prove value on real data before any broader deployment discussion.

Fixed-scope pilot

Side-by-side against the current workflow on a problem you scope with us. Success means the package changes either model choice, review clarity, or experimental planning. Scope, duration, and pricing are agreed in the kickoff call.

Data from a live program
Ranked candidate mechanisms and challengers
Assumption comparison and diagnostics
Next-experiment recommendation when needed
Export package for incumbent workflow
Success Criteria

Better mechanism comparison than the current process.

Clearer review packet for internal or regulatory discussion.

Evidence that the output changes either model choice or the next experiment.

Stop defending “best among what we tried.”

Get a mechanism package that makes the alternatives explicit, surfaces the real disagreement, and shows what to test next.

Request a pilot