Ancestry, genotype, phenotype, and context inputs

The calibration layer depends on context. Some of that context is clinically ordinary, such as fasting status, pregnancy status, medication exposure, recent illness, or specimen quality. Some is more sensitive, such as ancestry, genotype, or phenotype. Consensus Engine treats these inputs as clinical signals, not identity labels. Their only purpose is to improve interpretation when a documented biological mechanism changes what a biomarker means. The system does not use ancestry to generalize about a person. It uses specific, consented, clinically relevant inputs to decide whether a defined calibration may apply.

Why these inputs exist

A calibration can only be applied safely if the engine knows whether its conditions are present. The purpose is not to collect more data for its own sake; it is to prevent incorrect interpretation. If the engine does not know the context, it should not pretend that it does.

Calibration	Required context
Duffy-null / ACKR1 neutrophil	Genotype, phenotype, or clinician-reviewed lab pattern
APOL1 kidney-risk	Confirmed genotype where available
HbA1c hemoglobinopathy	Known hemoglobinopathy, genetic result, assay context, or clinical suspicion
Vitamin D context	Ancestry, altitude, geography, clinical context, MD-approved logic
Fasting-dependent metabolic	Confirmed fasting status
Pregnancy-sensitive	Pregnancy status
Exercise-confounded	Recent strenuous exercise history
Assay-confounded	Assay method, supplement use, or medication context

Input hierarchy

The engine prioritizes the most specific biological evidence available.

Input	Confidence	Engine behavior
Confirmed genotype (APOL1, Duffy/ACKR1, hemoglobinopathy)	Highest	Calibration may apply if lifecycle and evidence conditions are met
Confirmed phenotype	High to moderate	Calibration may apply or route to clinician review depending on stakes
Repeated lab pattern	Moderate	May trigger review or confirmatory testing
Self-reported ancestry	Contextual	Supports interpretation but does not override genotype
Clinical context (pregnancy, fasting, illness, exercise, medications, altitude, specimen)	Variable	May suppress, modify, or delay interpretation
Missing or conflicting signal	Low	Route to `INSUFFICIENT_DATA`, `AWAITING_REVIEW`, or `REQUIRES_CLINICAL_CORRELATION`

Genotype as the strongest signal

When a relevant genotype is confirmed, it should govern the calibration decision. APOL1 high-risk genotype may modify kidney-risk interpretation beyond standard creatinine or eGFR logic; Duffy/ACKR1 status may modify interpretation of baseline neutrophil counts; hemoglobinopathy markers may affect the reliability of HbA1c as a glycemic marker. Confirmed genotype is a high-confidence input because it directly maps to a defined biological mechanism. Even then, genotype does not create an automatic diagnosis or treatment decision. It changes interpretation, supports review, or affects routing — and remains under the same controls: evidence linkage, lifecycle status, clinician visibility, patient language control, and audit trace.

These controls prevent genotype from becoming an invisible algorithmic modifier.

Phenotype and lab patterns

When genotype is not available, phenotype or repeated lab patterns may suggest that a calibration is relevant — for example, a recurring neutrophil pattern suggesting a Duffy-null phenotype, or a glycemic mismatch raising concern that HbA1c is not accurately reflecting glucose exposure.

Situation	Engine behavior
Phenotype clinically confirmed	Apply calibration if approved and appropriate
Lab pattern suggestive but unconfirmed	Route to clinician review
Pattern may affect high-stakes interpretation	Require clinical correlation before patient-facing output
Confirmatory testing would reduce uncertainty	Suggest clinician-directed confirmation
Pattern conflicts with other facts	Route to `REQUIRES_CLINICAL_CORRELATION`

This avoids both underuse and overuse of calibration. The system can surface the possibility without pretending the mechanism has been confirmed.

Self-reported ancestry

Self-reported ancestry can be clinically useful context, but it must be handled carefully. It should not be treated as a definitive biological marker, should not override confirmed genotype, should not be used to make broad assumptions, and should not be used outside the stated clinical purpose.

Appropriate (contextual) role	Never
Supporting a calibration possibility when evidence and clinician review allow	Automatically assigning biological traits
Helping decide whether confirmatory testing may be useful	Inferring risk from appearance or identity
Informing clinician review	Replacing genotype or phenotype
Improving transparency around lab interpretation	Segmenting patients for non-clinical purposes

This is one of the most sensitive parts of the system. The patient must understand why the question is being asked.

Context inputs beyond ancestry

Not all calibration inputs are ancestry-related. Many are routine clinical context variables that can change interpretation: fasting status, pregnancy status, recent acute illness, recent strenuous exercise, medication use, biotin use, dehydration at draw, time of draw, altitude, and assay method. The engine treats context as part of the clinical input, not as an afterthought. Pregnancy, acute illness, recent strenuous exercise, fasting confirmation, and assay confounding are examples of safety conditions that can suppress, harden, or route interpretation to review. Ancestry and genotype data are sensitive. The system collects and uses them only with express consent, clear purpose, and heightened protection. Ancestry and genotype data are sensitive under Colombian Ley 1581, requiring express consent and heightened protection, and the patient must be told why the information is requested and how it is used. A patient-facing explanation should be simple: “Some lab values can be interpreted more accurately when we know certain biological or genetic context. We ask for this information only to improve clinical interpretation and only with your consent.” A safe consent model includes:

Purpose — the data is used to interpret labs more accurately.
Scope — the data is used only for clinical calibration and related review.
Voluntary nature — the patient understands what is optional.
Impact of missing data — some calibrations may not apply without the input.
Protection — sensitive data receives heightened access controls.
Review — clinicians can inspect how the input affected interpretation.
Withdrawal or correction — the patient can update or correct information.

The goal is not only compliance. The goal is trust.

Purpose limitation

Sensitive inputs are used only for their stated clinical purpose: calibration and interpretation.

Permitted	Prohibited
Applying a documented biomarker calibration	Marketing segmentation
Routing uncertain results to clinician review	Non-clinical profiling
Suggesting confirmatory testing when appropriate	Inferring unrelated traits
Improving de-identified calibration research with consent	Unconsented model training
Supporting clinical audit	Broad identity classification

Ancestry is never used for any purpose other than clinical calibration. This should remain a hard product, legal, and clinical boundary.

Conservative handling of uncertainty

The engine should never force a calibration when the input signal is weak, missing, or contradictory.

Condition	Output
Required input missing	`INSUFFICIENT_DATA`
Signal suggestive but not confirmed	`AWAITING_REVIEW`
Signals conflict	`REQUIRES_CLINICAL_CORRELATION`
Calibration high-stakes	Clinician review required
Calibration evidence incomplete	Keep in draft, clinician-only, or research mode
Patient consent absent	Do not use ancestry or genotype input

This preserves the preventive tone of the system: act early when justified, but escalate when uncertain.

Avoiding misuse

The system must avoid broad racial generalizations. A calibration applies to a defined mechanism, not to appearance, identity, or unsupported assumptions.

Unsafe generalization	Correct encoding
”Patient appears to be Black, apply neutrophil calibration"	"Confirmed Duffy-null status or clinician-reviewed phenotype supports calibration"
"Patient reports African ancestry, assume APOL1 risk"	"APOL1 risk requires genotype confirmation or clinician-directed testing"
"HbA1c is unreliable for this ancestry group"	"HbA1c may be affected if hemoglobinopathy or assay-specific concern is present"
"Vitamin D threshold changes automatically by ethnicity"	"Apply only the approved calibration logic based on defined context and evidence”

The system encodes specific, documented biological mechanisms, not broad racial generalizations. This is both scientifically and ethically necessary.

Clinician-facing visibility

Clinicians should see exactly how ancestry, genotype, phenotype, or context influenced an interpretation. The clinician view shows the input source, confidence level, calibration applied, applies-when condition, evidence source, version, review requirement, patient-visible effect, and a link to the decision record. This protects the physician from black-box interpretation. The clinician should never have to guess why the engine changed a result.

Patient-facing language

Patient-facing language should be accurate, simple, and non-alarming. The patient does not need every technical detail, but the explanation should not hide the existence of context-based interpretation.

Scenario	Example language
Calibration applied with confidence	”Your result was interpreted using additional biological context that may affect this marker.”
Calibration requires review	”This result may need additional clinical context before we can explain it accurately.”
Missing context	”We need one more piece of information before interpreting this marker safely.”
Conflicting signals	”This result is being reviewed because some inputs point in different directions.”
Confirmatory testing suggested	”Your clinician may recommend a confirmatory test to interpret this result more accurately.”

The tone should be preventive and calm. The system avoids language that sounds deterministic when the input is uncertain.

Data protection

Sensitive inputs require stronger protection than ordinary product data. Clinical data and decision records are sensitive data, with least-privilege access, segregation, and audit logging, and de-identified data may improve calibrations only with separate consent. For ancestry, genotype, and phenotype inputs, the protection model includes express consent, access restriction to authorized clinical or technical roles, purpose restriction to clinical interpretation, audit logging, data minimization, segregation from general product data, consented de-identification for learning, and a retention policy that avoids indefinite storage without clinical or legal basis. Traceability and privacy must work together. The system should be explainable without exposing sensitive information unnecessarily.

Runtime example

Duffy-null neutrophil count
Uncertain HbA1c

Step	Value
Raw biomarker	Absolute neutrophil count below standard range
Context available	Confirmed Duffy-null / ACKR1 status
Input confidence	High
Calibration check	Duffy-null calibration applies
Evidence trace	Approved source linked through evidence library
Lifecycle check	Calibration is `ACTIVE`
Output behavior	Avoid automatic abnormal labeling if calibrated interpretation supports it
Review routing	Clinician sees calibration trace; patient receives cautious explanation
Decision record	Stores input, calibration, rule, evidence, state, and version

Step	Value
Raw biomarker	HbA1c appears elevated
Context available	Possible hemoglobinopathy, not confirmed
Input confidence	Moderate to uncertain
Calibration check	HbA1c reliability concern detected
Output behavior	Do not overstate glycemic conclusion
Review routing	`REQUIRES_CLINICAL_CORRELATION`
Next step	Clinician may consider glucose, CGM, fructosamine, repeat testing, or confirmatory evaluation

These examples show how the engine prevents both misclassification and overconfidence.

Design principle

Use sensitive context only when it improves clinical interpretation, only with consent, only through defined mechanisms, and only under medical governance.

This principle protects the patient, the clinician, and the company.

Introduction

Architecture

Clinical methodology

Governance & safety

For stakeholders

Ancestry, genotype, phenotype, and context inputs

Why these inputs exist

Input hierarchy

Genotype as the strongest signal

Phenotype and lab patterns

Self-reported ancestry

Context inputs beyond ancestry

Purpose limitation

Conservative handling of uncertainty

Avoiding misuse

Clinician-facing visibility

Patient-facing language

Data protection

Runtime example

Design principle

​Why these inputs exist

​Input hierarchy

​Genotype as the strongest signal

​Phenotype and lab patterns

​Self-reported ancestry

​Context inputs beyond ancestry

​Consent and transparency

​Purpose limitation

​Conservative handling of uncertainty

​Avoiding misuse

​Clinician-facing visibility

​Patient-facing language

​Data protection

​Runtime example

​Design principle

Why these inputs exist

Input hierarchy

Genotype as the strongest signal

Phenotype and lab patterns

Self-reported ancestry

Context inputs beyond ancestry

Consent and transparency

Purpose limitation

Conservative handling of uncertainty

Avoiding misuse

Clinician-facing visibility

Patient-facing language

Data protection

Runtime example

Design principle