HelixCore
Public Documentation

HelixCore: System-Level Documentation

HelixCore™ is a deterministic, longitudinal signal engine built to surface patterns of stability and drift within an individual’s own data. These docs describe system contracts and boundaries without exposing internal tuning, formulas, or clinical interpretation.

Deterministic Longitudinal Read-Only Surfaces Explainable by Design
Overview

Signals over answers

HelixCore™ produces signals, stability summaries, and drift indicators. It does not diagnose, predict outcomes, or provide recommendations. Every output is tied to a specific observation window and engine version.

The system is built for longitudinal integrity: patterns across time carry more weight than individual measurements. Outputs remain conservative when data coverage is uneven, and are suppressed rather than inferred when confidence is insufficient.

System Architecture

Layered, deterministic infrastructure

HelixCore™ runs as a nightly batch system with strict separation between data ingestion, feature windowing, signal computation, composite scoring, drift detection, explainability, and API delivery. Each layer has exclusive responsibilities to prevent interpretive leakage.

Append-only ingestion preserves raw observations without mutation or smoothing.
Seven-day rolling windows align data before any signal computation.
Signal computation is deterministic, versioned, and non-adaptive.
Drift detection is comparative, conservative, and never predictive.
Canonical Signals

Five fixed semantic contracts

Each signal measures within-person patterns over a 7-day window. Signals are descriptive and do not imply diagnosis or health status. Scores are accompanied by coverage and confidence metadata.

S1 · Waking Stability

Day-to-day stability of basal glucose patterns after waking.

S2 · Meal Response Predictability

Consistency of post-meal response patterns within the individual.

S3 · Recovery Curve

Consistency and timing of return toward baseline after perturbations.

S4 · Variability Range

Within-day dispersion independent of mean level.

S5 · Stress / Circadian Stability

Temporal regularity of glucose patterns across the day.

Explainability Philosophy

Constrained, longitudinal explanations

Explanations are templated, versioned, and deliberately narrow. They describe what a signal measures, what changed relative to prior windows, and the limits of interpretation without advice, diagnosis, or prediction.

Coverage and missingness are always disclosed.
Uncertainty is surfaced when data quality is uneven.
Interpretation limits are explicit and non-prescriptive.
Platform Surfaces

Read-only, versioned access

HelixCore™ surfaces are designed to be observational and conservative. They deliver versioned signal outputs with confidence and limits, while preventing exposure of raw data or internal tuning.

Mobile

Read-only snapshots and signal explanations with explicit engine versions, coverage, and limits. Missing values are returned as null rather than inferred.

Clinician

Authenticated access to signal history, drift, and explainability outputs with coverage, confidence, and auditability embedded.

Research

Population-level, windowed access with pseudonymized identifiers, strict read-only scopes, and reproducibility metadata.

Boundaries

What HelixCore™ is not

Not a diagnostic system, risk model, or predictive system.
Not a source of recommendations, coaching, or clinical directives.
Not a real-time or adaptive personalization loop.
Not an exposure of raw data, thresholds, or tuning parameters.

Limits: Outputs remain observational and include explicit limits such as “Not diagnosis” and “Not medical advice.”

MetaboLink (company) and HelixCore (engine) are unrelated to similarly named academic metabolomics methods.