Engineering case study · Clinical Informatics / Educational Tool

LipidPath.

Explainable guideline pathways, with uncertainty made visible.

An educational dyslipidemia guideline explorer with deterministic adult pathways, transparent evidence and explicit clinical limits.

  • React
  • TypeScript
  • Vite

Making clinical pathways inspectable

LipidPath turns supported adult recommendations from the 2026 multisociety dyslipidemia guideline into an interactive educational workflow. It is a software project under review, not a clinically validated prescribing system. Independent qualified clinical review is pending.

A deterministic engine with a visible source trail

Pure TypeScript rules evaluate eligibility, simultaneous indications, contraindication review, missing data and laboratory values separately from React. Structured results connect each decision to an original evidence summary, recommendation number, evidence grade and journal page.

The guided interface includes a keyboard-accessible expandable pathway, measured-versus-target comparisons and an evidence explorer. Static prerendering and route-based code splitting fit the existing GitHub Pages architecture.

From inputs to explainable output
  1. Check applicabilityValidate clinical facts and stop unsupported presentations.
  2. Evaluate pathwaysPreserve concurrent indications and evidence references.
  3. Inspect reasoningCompare goals, expand decision facts and review source limitations.

Original design; no source guideline figure is reproduced.

Clinical inputs stay in memory

No account or patient identifiers are requested. Inputs are not written to URLs, browser storage, analytics or a backend. Reset or reload clears the assessment. The authoritative PREVENT calculator opens separately without receiving the entered data.

The software does not claim HIPAA compliance, ACC/AHA endorsement or institutional certification. The copyrighted source PDF, tables, figures and organizational marks are not redistributed.

Software verification and clinical review are separate

Automated scenarios exercise numerical boundaries, eligibility, competing indications, missing values and review-required states. Browser checks cover the guided workflow, hydration, privacy, responsive layout and accessibility. Detailed validation records accompany the implementation.

The core release supports adult prevention, severe LDL elevation, diabetes age 40–75, eligible CAC interpretation, triglyceride assessment, Lp(a), apoB and selected CKD/HIV considerations. Complex populations and detailed medication rechallenge are deferred. PREVENT uses a confirmed manual result from the official AHA tool; no equation is approximated.

Qualified independent clinical review remains a release requirement. No development history, adoption or medical-review credentials are implied.