Abstract & Purpose
Burak Multiomics is a deterministic, explainable bio-computational engine designed to detect sub-clinical metabolic drift before it manifests as late-stage disease. The platform converts heterogeneous lab and microbiome inputs into standardized, auditable signals, then computes risk using time-series logic and a rules-based cohort mapping system.
Scientific Authority: The scientific logic below is grounded in Burak’s internal “Sovereign Source of Truth (SSOT)” specification (v2.0.0, March 25, 2026), which defines ingestion hygiene, units, privacy constraints, marker thresholds, and cohort routing rules.
Data Lake (Ingestion) & Hygiene
Before any scoring, incoming reports pass a strict ingestion and hygiene gate designed to preserve true pathology while rejecting impossible values.
Unit Normalization (VNRA)
All incoming laboratory values are normalized into master units prior to any computation (e.g., glucose/lipids in mg/dL, hs‑CRP in mg/L, creatinine in mg/dL, HbA1c in %, insulin in μIU/mL, zonulin in ng/mL, microbiome taxa in % relative abundance).
“Limits of Life” Guard (Keep & Flag vs Reject)
To prevent deleting acute pathology as “noise,” extreme-but-possible values are preserved and flagged. Values outside biological possibility are rejected as pre-analytical error. Examples of keep/flag ranges include glucose up to 1,500 mg/dL, ALT/AST up to 10,000 U/L, triglycerides up to 5,000 mg/dL, hs‑CRP up to 350 mg/L, and ferritin up to 15,000 ng/mL.
Adaptive Two-Step Hygiene
Step A (Physiological hard stop): auto-reject biologically impossible values. Step B (adaptive statistical filter): strict Z-filtering is used only for markers under tight homeostasis; for wide dynamic-range markers (e.g., CRP, AST/ALT, glucose), statistical truncation is disabled so acute signals are preserved.
Privacy Protocol (Zero‑PII)
No names, emails, or phone numbers are allowed in science processing. Longitudinal linkage uses a SHA‑256 hash identifier derived from (National ID + salt).
Biological Demographics (Reference Logic)
Sex and age are required inputs for correct reference logic (e.g., creatinine thresholds and CKD‑EPI 2021 eGFR constants). Phase 1 logic is scoped to adults (age ≥ 18); pediatric reports are flagged out-of-scope. Ethnicity/race coefficients are not used in eGFR (CKD‑EPI 2021 removed race coefficient).
Classification Switch (Routing)
Reports are routed by signature marker presence: Type A (non‑omics) → blood engine; Type B (omics) → microbiome census engine; Type C (hybrid) → parallel processing with merged outputs.
Acute Filter, Time-Series & Noise Logic
Acute Management System (AMS)
AMS is a mandatory gatekeeper that runs before cohort routing and velocity math to protect longitudinal scoring from acute contamination.
Trigger (acute spike detection): WBC > 11.0 and neutrophils > 70% (with a sudden delta from baseline) or CRP > 10.0 mg/L (with a sudden delta).
Action: route to an acute infection pathway and quarantine those values (do not feed into chronic cohorts or velocity calculations); recommend retest post‑recovery.
Temporal Anchoring
Baseline (T‑0): the first validated report with > 70% marker density. Delta time (dt): computed in fractional years (e.g., 6 months = 0.5). If the time between tests is < 30 days, velocity is ignored as noise unless the event is AMS‑triggered.
Reference Change Value (RCV)
To avoid reacting to lab “chatter,” changes smaller than RCV are treated as non-meaningful variation.
Index of Individuality (IoI) — Personal Baselines
For markers with IoI < 0.6, the engine may shift from population ranges to a personal baseline when at least 3 readings within 12 months exist. Personal safe zone is Personal Mean ± RCV, with overrides that prevent “normalizing” clearly pathological baselines.
Marker Matrix & Derived Fields
Phase 1 logic accepts a defined marker set across metabolic, inflammation, gut integrity, renal/liver, cardiovascular, thyroid, nutrient, and microbiome domains (plus derived fields). Examples of critical thresholds include glucose > 126 mg/dL, HbA1c > 6.5%, insulin > 25 μIU/mL, triglycerides > 200 mg/dL, hs‑CRP > 3.0 mg/L, ferritin > 500 ng/mL, zonulin > 60 ng/mL, eGFR < 60 mL/min, ApoB > 90 mg/dL, Lp(a) > 50 mg/dL, and UACR > 30 mg/g.
Derived examples: HOMA‑IR = (Glucose × Insulin) / 405, Trig/HDL ratio, Firmicutes/Bacteroidetes ratio, BMI, CKD‑EPI 2021 eGFR, and VAI (Amato et al.).
Cohort Mapping & Routing Rules
Burak applies deterministic trigger logic to route profiles into cohort pathways while preventing false reassurance and double counting.
- Sub-clinical drift pathway: if no “red” markers exist but there are ≥ 4 “yellow” markers across categories, route to early warning (do not classify as “healthy”).
- Shared-marker hierarchy: shared markers (e.g., CRP, zonulin) have a primary/secondary hierarchy to prevent inflating risk; inflammatory markers only contribute to secondary cohorts when primary cohort markers are also elevated.
- Proxy logic (when gut data is missing): selected blood proxies can estimate missing gut signals with a confidence penalty (e.g., Firmicutes proxy via insulin + high triglycerides; Proteobacteria proxy via neutrophil/lymphocyte ratio).