Skip to main content

Quality Control, Quality of Code & Regulatory Requirements

Nova In Silico applies a rigorous, fully traceable Quality of Code and Quality Control framework to ensure that all Modeling & Simulation (M&S) projects making use of its technology - notably leveraging the modeling & simulation platform jinkō - meet the highest scientific, technical, and regulatory standards. This document provides an overview of how quality is embedded throughout our in silico modeling lifecycle.

Our approach ensures that:

  • Scientific foundations are transparent, evidence-based, and traceable.

  • Software engineering practices guarantee robustness, reproducibility, and security.

  • Regulatory expectations (ICH M15 recommandations, FDA/EMA best practices) are met through built-in auditability.

This "white-box" philosophy underpins our ability to deliver credible, defendable modeling results to clients and regulatory agencies.

Quality Control via Jinkō's Software Development Lifecycle (SDLC)​

Nova applies a controlled and auditable Software Development Lifecycle (SDLC) to all modeling-related software components and supporting infrastructure. This ensures that the modeling environment itself—jinkō and its underlying engines—adheres to the same standards of quality expected for regulated digital systems.

Key SDLC pillars​

  • Version control & protected branches ensure all changes are traceable and cannot bypass review.

  • Mandatory peer reviews validate correctness, clarity, and maintainability before merging.

  • Security controls (SAST/DAST) detect vulnerabilities early and enforce secure coding standards.

  • Documented release gates require validation of acceptance criteria and evidence before deployment.

  • Full traceability links requirements → implementation → tests → release.

Our rigorous SDLC ensures:

  • Reliability of all model execution engines and analytical components,

  • Consistency and stability of the computational environment used to run models,

  • Reproducibility of simulations across releases,

  • Auditability for partners and regulators,

  • Minimized risk of regression or unintended behavioral change.

Built-In Quality of Code in Jinkō​

Jinkō operationalizes Quality of Code practices through built-in, platform-level capabilities that make versioning, collaboration, and scientific peer review seamless and fully traceable.

Core jinkō features supporting QoC include:

Version-controlled computational models (and any item in jinkō)​

Every model, submodel, parameter, equation, assertion and results are versioned automatically. Model evolution is fully auditable and can be navigated historically at any depth.

Unit consistency checks​

Jinkō’s core engine ensures that equations, variables, and parameters are dimensionally consistent and adhere to defined unit constraints.

Numerical verification & stability​

To ensure numerical integrity, jinkō integrates the following:

  • Code generation: Randomly generated ODE systems are tested against analytical solutions to validate solver correctness.

  • Calculation verification: Models are run under decreasing solver tolerances until global error falls below a predefined threshold.

These processes ensure stability, accuracy, and reproducibility of simulations.

Code diff for model edition​

Jinkō provides a visual diff interface for computational models, enabling users to:

  • Compare changes between versions,

  • Inspect modifications to equations, reactions, parameters, and metadata,

  • Review added or removed components with line-by-line clarity.

Collaborative peer review​

Model changes can be reviewed by collaborators through inline comments, annotations, and threaded discussions directly attached to model components. jinkō supports:

  • Comment threads tied to specific variables or equations,

  • Approval workflows,

  • Transparent review history.

Models in jinkō expose:

  • Human-readable equations,

  • Structured metadata,

  • Graphical submodel diagrams,

  • Links to biological assertions and reference (see below)

Non-programmers can also validate correctness through jinkō’s user interface, enabling notably domain experts to assess model structure and assumptions.

  • Run models and inspect outputs,

  • Compare versions side-by-side,

  • Review automated checks (units, dimensions, warnings),

  • Evaluate impacts of modifications.

jinkō furthermore enables external collaborators—such as partners, clients, or regulatory reviewers—to perform:

  • Independent inspection of the model,

  • Transparent evaluation of changes,

  • Assessment of biological plausibility,

  • Confirmation of consistency with data and literature.

This democratizes rigorous model review, making high-quality scientific validation accessible to any organization using jinkō.

Full scientific traceability​

Each project can be documented directly in jinkō with a structured literature and data review. Each item in a jinkō model links back to its knowledge source:

  • Literature assertions,

  • Extracted data,

  • Experimental references,

  • Prior versions.

This documentation forms the basis for all subsequent modeling steps and ensures that every computational element is linked to a documented source. It stores structured biological and experimental evidence used to build and justify all model components.

Encouraging Modern Regulatory Expectations​

Via its features, jinkō enables key regulatory expectations, notably:

Uncertainty management​

Jinkō enables industry-standard uncertainty management principles:

  • Systematic identification of knowledge gaps,

  • Explicit documentation of hypotheses and simplifications,

  • Sensitivity analyses for key parameters,

This methodology ensures meaningful interpretation of simulation outcomes and strengthens regulatory readiness.

Documentation templates to foster Regulatory interactions and alignment​

Jinkō interactive documentation modules incorporates default templates that can be used by project managers and their teams to document their projects and links their assets such as data, models, protocols and results to facilitate Regulatory Interactions.

It notably includes ICH M15 templates, such as a Key Elements Assessment + MIDD planning template.

Summary​

Nova’s Quality of Code and QC processes ensure:

  • Transparency from knowledge to equations

  • Traceability through versioned assets

  • Scientific rigor through structured documentation & evidence

  • Technical robustness through automated testing & reviews

  • Regulatory readiness via white-box documentation and auditable workflows

Together, these foundations ensure that Nova In Silico delivers credible, reliable, and defensible M&S results designed for high-stakes decision making.