ICR WHITE PAPER 009
MEASURING COHERENCE
Requirements for a Valid Multidomain Index
David FischerInstitute for Coherence and Regulation (ICR)Knightdale, North Carolina, USASeptember 2026 | Publication Version 1.0
Recommended citationFischer, D. (2026). Measuring Coherence: Requirements for a Valid Multidomain Index. ICR White Paper 009 (Publication Version 1.0). Institute for Coherence and Regulation.
DOI: 10.5281/zenodo.22709724
Abstract
The Coherence & Regulation Framework (CRF) uses coherence as a conceptual description of sufficiently coordinated, timely, and context-appropriate regulatory activity that supports function, flexibility, and recovery. It does not currently possess a validated quantitative measure of whole-person coherence. This distinction is essential. Network Physiology has demonstrated that physiological systems interact dynamically across organ systems and that network topology and coupling change with physiological state. Those findings support studying multisystem coordination, but they do not establish that a single scalar 'coherence score' exists. This paper defines the evidentiary requirements that would have to be met before ICR could responsibly claim to measure coherence. A valid index would require a clearly specified construct, prespecified measurement model, reliable component measures, temporal and contextual standardization, convergent and discriminant validity, incremental predictive value, responsiveness, calibration, external replication, and explicit failure criteria. The index must also outperform simpler alternatives and avoid circular validation against outcomes used to build it. This paper proposes a staged research program and a provisional Coherence Measurement Architecture while explicitly prohibiting premature commercial or clinical scoring.
Keywords: coherence; measurement; construct validity; network physiology; multidomain index; psychometrics; biomarkers; coordination; validation; predictive validity
1. Purpose and Scientific Status
The word coherence is useful conceptually but dangerous scientifically if it is allowed to mean whatever outcome appears favorable. WP-009 therefore establishes a measurement firewall: CRF may use coherence as a defined theoretical construct, but ICR should not claim to measure whole-person coherence until a measurement system survives formal validation.
This paper does not introduce a validated Coherence Index. It specifies what would be required to create one.
The distinction protects the framework from a common failure in emerging health models: inventing a score first and searching for validation afterward.
2. Canonical CRF Definition of Coherence
Coherence is a dynamic condition in which interacting regulatory processes are sufficiently coordinated, timely, and context-appropriate to support efficient function, flexibility, recovery, and continued adaptive capacity.
Four words are essential: dynamic, coordinated, timely, and context-appropriate. Coherence is not perfect synchrony, uniformity, stillness, high HRV, relaxation, positive mood, or absence of symptoms.
The phrase 'sufficiently coordinated' also means that perfect coupling is not assumed to be desirable. Biological systems require partial independence, modularity, variability, and state-specific reconfiguration.
3. What Network Physiology Establishes
Network Physiology provides strong scientific precedent for studying the body as interacting physiological systems rather than isolated organs. Bartsch and colleagues demonstrated that organ interactions form dynamic networks whose topology and coupling characteristics reorganize across physiological states.
Subsequent work has emphasized that physiological systems interact across multiple spatial and temporal scales and that coordinated network interactions contribute to organism-level states such as wake, sleep, exercise, cognition, and consciousness.
This literature supports the CRF proposition that coordination itself can be scientifically studied. It does not establish CRF's five-layer architecture, validate ICR's definition of coherence, or justify collapsing multisystem dynamics into one number.
4. Coordination Is Not the Same as Synchrony
A serious coherence measure must distinguish coordination from simple synchrony. Two signals can correlate or synchronize for reasons unrelated to beneficial regulation, including shared external forcing, artifact, pathology, or mathematical preprocessing.
Network interactions can also change direction, strength, timing, and functional form as physiological state changes. Therefore, a fixed expectation that 'more coupling equals more coherence' would be biologically implausible.
CRF instead proposes state-appropriate coordination: the pattern should fit the task, environment, and time scale.
5. The Measurement Problem
A multidomain index faces at least five problems. First, different CRF layers contain different kinds of variables. Second, their time scales differ. Third, normal ranges can be context dependent. Fourth, relationships may be nonlinear. Fifth, an aggregate score can conceal clinically or scientifically important disagreement among components.
For example, a participant could report lower perceived stress while showing no measurable change in autonomic recovery. A valid system must preserve that disagreement rather than force both measures into a favorable composite.
A coherence score that always improves when any selected wellness outcome improves would be circular and scientifically invalid.
6. Construct Definition Before Instrument Construction
Before selecting sensors or questionnaires, ICR must define what coherence is expected to predict that established constructs do not already predict. Candidate targets include recovery after demand, functional readiness, response efficiency, resilience to repeated challenge, and reduced compensatory cost.
If coherence cannot be distinguished empirically from perceived well-being, allostatic load, resilience, fitness, HRV, symptom burden, or general health, a separate index may be unnecessary.
Construct validity therefore precedes branding.
7. Provisional Coherence Measurement Architecture
Domain
Candidate feature
Example measure
Role
Boundary
Context
Demand-appraisal fit
Validated stress/appraisal measure
Context modifier
Not physiology
Autonomic
Response/recovery dynamics
HR, selected HRV, respiration
Dynamic regulation
Not whole-body coherence
Behavioral
State transition and recovery behavior
Sleep/activity/task data
Functional expression
Wearables are imperfect proxies
Functional
Performance under defined demand
Cognitive/physical task
Outcome capacity
Task specific
Recovery
Time, slope, residual carryover
Repeated post-demand measures
Core dynamic feature
Domain-specific kinetics
Cross-system
Coupling/reconfiguration
Synchronous multichannel analysis
Coordination feature
Requires advanced methods
Reserve/cost
Output relative to effort
Task + cost measures
Capacity margin
Not yet validated as CRF reserve
8. Why the Five CRF Layers Should Not Simply Be Averaged
The five layers are organizing categories, not five interchangeable subscales. Layer 1 may be measured through appraisal and context, while Layer 5 may require laboratory assays. Averaging standardized scores from fundamentally different domains can create a mathematically neat but biologically meaningless result.
A future index may prove hierarchical, network-based, profile-based, or state-space based rather than additive. The measurement model should be selected from empirical structure, not imposed because five layers exist.
ICR should therefore begin with a multidomain profile rather than a global score.
9. Minimum Psychometric Requirements
Content validity: selected measures must represent the intended construct rather than merely what is easy to collect.
Reliability: repeated measurement under comparable conditions must produce sufficiently stable estimates.
Measurement error: expected noise and smallest detectable change must be quantified.
Structural validity: the proposed factor, network, or hierarchical structure must fit observed data.
Convergent validity: related constructs should associate in prespecified directions.
Discriminant validity: coherence must remain distinguishable from neighboring constructs.
Known-groups validity: groups expected to differ under a justified hypothesis should show predicted differences.
Responsiveness: the measure should detect meaningful change when change truly occurs.
Predictive validity: the index should predict prespecified future outcomes.
External validity: findings must replicate outside the development sample.
10. Incremental Validity Is Mandatory
A new coherence index is scientifically worthwhile only if it improves prediction, explanation, or decision-making beyond simpler established measures.
For example, if perceived stress, sleep duration, resting heart rate, and functional testing predict recovery just as well as a complex Coherence Index, the complex score adds little value.
Future studies should therefore compare the full index with parsimonious baselines and report out-of-sample performance, not only in-sample correlations.
11. Avoiding Circular Validation
An index cannot be validated against the same variables used to construct it without independent outcomes. If sleep quality is one component of the score, showing that the score correlates with sleep quality does not establish external validity.
Likewise, if participant well-being ratings determine the weighting of components, subsequent correlation with well-being is partly built into the score.
Primary validation outcomes should therefore be independent and prespecified, such as standardized recovery performance, repeated-challenge readiness, or prospective functional outcomes.
12. Context and State Standardization
Physiological coordination changes with sleep stage, posture, exercise, cognitive demand, breathing, meals, circadian phase, medication, and environment. A valid measure must specify the state in which it is collected.
This means ICR may ultimately need several state-specific coherence profiles rather than one universal number—for example, resting coordination, challenge response, recovery coordination, and repeated-demand coordination.
State dependence is not measurement failure; it is a property of physiology.
13. Time Scale
Coordination exists across milliseconds, seconds, minutes, hours, and days. Measures sampled at different scales cannot be combined casually.
HRV may capture beat-to-beat cardiac dynamics; sleep/activity data may require days; endocrine rhythms may require repeated sampling across hours; contextual demand can change within minutes.
Any future index must define its temporal resolution and justify how cross-scale information is integrated.
14. Cross-System Coupling
Network Physiology uses simultaneous recordings to infer time-varying interactions among systems. CRF can learn from this methodology by treating coupling as a measurable property rather than assuming it.
Candidate features could include coupling strength, directionality, time delay, network topology, modularity, reconfiguration during state transition, and recovery of network organization after perturbation.
These analyses require specialized signal processing and large enough datasets to avoid unstable estimates. They should not be approximated by visually comparing two consumer-device graphs.
15. HRV: Useful Component, Invalid Shortcut
HRV can provide useful information about cardiac autonomic dynamics when collected and interpreted appropriately. It should not be relabeled as 'coherence' by itself.
A whole-person CRF construct necessarily exceeds a single cardiac signal. HRV can be one candidate feature in Layer 2 or in a cross-system model, but a high or aesthetically smooth HRV pattern does not validate cross-layer coordination.
ICR materials should consistently distinguish 'HRV measure' from 'coherence measure.'
16. The Coherence Profile Before the Coherence Score
The recommended first-generation output is a profile. A profile preserves the measured domains and allows disagreement among them.
For example, an individual might show strong task performance, delayed autonomic recovery, good subjective recovery, and high compensatory effort. That pattern is scientifically more informative than averaging the components into 72/100.
Only after profiles demonstrate stable, replicable relationships with outcomes should ICR investigate whether a scalar or small set of summary indices is justified.
17. Candidate Validation Outcomes
Outcome
Why useful
Limitation
Recovery time after standardized demand
Directly tests dynamic regulation
System specific
Second-challenge performance
Tests retained capacity
Requires safe repeatable challenge
Cost-to-output ratio
Tests efficiency/compensation
Cost metric must be valid
Prospective functional decline or improvement
Tests prediction
Requires longitudinal study
Daily-life recovery
Tests ecological validity
High confounding
Independent clinician/research assessment
External comparator
Not a gold standard for coherence
18. Development Stages
Stage 0 — Construct freeze
Freeze the conceptual definition, boundaries, intended use, and prespecified outcomes before scoring.
Stage 1 — Component reliability
Test each candidate measure independently under standardized conditions.
Stage 2 — Multidomain observational study
Collect synchronized data without creating a global score; characterize relationships and redundancy.
Stage 3 — Challenge-recovery modeling
Determine which features predict response, recovery, and second-challenge readiness.
Stage 4 — Model derivation
Develop candidate profile or index using a training dataset with penalization and prespecified feature handling.
Stage 5 — Internal validation
Use cross-validation or bootstrap methods and evaluate calibration and overfitting.
Stage 6 — External validation
Test the locked model in a separate population and setting.
Stage 7 — Responsiveness
Determine whether change in the index corresponds to independent meaningful change.
Stage 8 — Independent replication
Require investigators outside ICR to test the model.
Stage 9 — Intended-use evaluation
Only then determine whether the measure is suitable for research, wellness tracking, or any higher-stakes application.
19. Ten Falsifiable Hypotheses
H1. A prespecified multidomain coordination profile will predict recovery after standardized demand better than any single component alone.
H2. Cross-system dynamic features will add predictive value beyond resting HRV and subjective stress measures.
H3. The optimal coordination pattern will differ by physiological state, rejecting a universal 'more synchrony is better' model.
H4. Coherence-related features will show partial test-retest reliability under matched conditions.
H5. A profile-based model will initially outperform a simple additive five-layer score.
H6. Coherence measures will remain empirically distinguishable from general well-being, symptom burden, fitness, and perceived stress.
H7. Recovery and second-challenge outcomes will provide stronger validation targets than same-time self-report alone.
H8. Some component disagreement will be informative rather than measurement error.
H9. A locked model will retain meaningful predictive performance in an external cohort.
H10. If the multidomain model does not outperform simpler established measures, ICR should not market or publish a global Coherence Index.
20. Sample Size and Overfitting
Multidomain indices can easily contain more candidate features than a modest study can support. This creates overfitting: a model appears impressive in the development sample but fails elsewhere.
ICR should prelimit features, use appropriate penalization, separate training and validation data, and obtain statistical consultation before model development. Sample size should be justified from the intended model and outcome rather than chosen by convenience.
A four-person or small case series can test feasibility of data collection. It cannot derive or validate a multidomain index.
21. Missing Data and Device Quality
Wearables and wellness devices can produce missing, proprietary, or algorithmically transformed data. A scientifically useful index should prefer raw or well-characterized signals where possible and document firmware, device model, sampling rate, artifact handling, and missing-data rules.
A score should not depend critically on a proprietary black-box metric that can change without notice.
If consumer devices are used, their role should be explicitly described as research-grade, validated for the intended measure, or exploratory.
22. Clinical Versus Wellness Use
Even a statistically validated research index would not automatically become a diagnostic tool. Clinical use requires additional evidence concerning intended population, reference ranges, decision thresholds, harms, benefits, reproducibility, and regulatory considerations.
ICR's near-term appropriate target is research measurement and educational tracking—not diagnosis, disease screening, or treatment selection.
The word 'coherence' should never be used to imply that a low score identifies hidden disease.
23. Claims Discipline
Use 'CRF conceptual coherence' when referring to the theory.
Use the actual physiological or behavioral measure when reporting data.
Do not call HRV alone a whole-person coherence measure.
Do not create a 0–100 Coherence Score before validation.
Do not infer cellular or endocrine coherence without direct measures.
Do not treat high synchrony as universally desirable.
Do not validate a score against variables used to construct it.
Do not publish only favorable cut points discovered after examining outcomes.
Report null and contradictory findings.
Retire the index if it fails external validation.
24. Application to Current ICR Studies
Current ICR case studies can contribute to Stage 0 and Stage 1 work by clarifying feasible outcomes, timing, participant burden, and data-collection procedures. They should not be used to calculate or advertise a Coherence Index.
The existing subjective measures—perceived stress, sleep quality, energy, tension/discomfort, ability to relax, mental clarity, emotional steadiness, well-being, and ability to recover after stress—can be analyzed as separate exploratory outcomes.
Later studies can add standardized challenge-recovery measurements and validated physiological measures. The transition from a wellness outcome dashboard to a coherence measurement system should occur only after construct and measurement validation.
25. Falsification and Retirement Criteria
A proposed Coherence Index should be rejected or substantially revised if its component measures are unreliable; its structure fails replication; it is indistinguishable from general health or well-being; it fails to predict independent outcomes; it adds no value beyond simpler measures; performance collapses in external validation; or interpretation depends on post hoc changes.
A scientifically credible Institute must be willing to conclude that coherence is useful as a conceptual framework but not reducible to one number. That outcome would not invalidate CRF.
Harmonization With the Mature CRF
WP-009 now treats a whole-person Coherence Index as a measurement-development problem rather than an assumed score. The mature CRF defines coherence conceptually, but no single instrument, biomarker, device output, questionnaire, HRV metric, or mathematical combination has yet been validated as a whole-person measure of coherence.
Canonical Measurement Definition
For measurement research, coherence is the degree to which prespecified regulatory variables demonstrate context-appropriate coordination, timing, flexibility, and recovery under defined conditions. This operational definition is narrower than the full conceptual definition and must always identify the population, context, variables, and time scale being studied.
Profiles Before Scores
The preferred early product is a Coherence Profile: an interpretable set of domain-specific measures displayed without assuming that they form one latent quantity. A composite Coherence Score should be developed only if empirical evidence demonstrates that aggregation is meaningful, reliable, valid, responsive, interpretable, and useful.
Reflective Versus Formative Models
Before constructing an index, investigators must specify whether coherence is modeled reflectively or formatively. In a reflective model, observed variables are manifestations of an underlying construct and are expected to show an appropriate internal structure. In a formative model, different indicators jointly define the construct and need not be highly correlated. Using internal-consistency statistics on a formative index can therefore be conceptually inappropriate.
Measurement Development Standard
Instrument development should address content validity, structural validity where applicable, reliability, measurement error, construct validity, criterion validity when a defensible criterion exists, responsiveness, interpretability, and measurement invariance. COSMIN provides an established measurement-science framework for many of these properties and should be used where applicable rather than inventing ICR-specific psychometric standards.
No Gold Standard Yet
There is currently no accepted gold standard for whole-person coherence as defined by CRF. Criterion validity therefore cannot be claimed merely because a proposed index correlates with HRV, stress scores, sleep, well-being, or another favored measure. Validation should rely initially on prespecified construct hypotheses, known-groups comparisons, longitudinal prediction, responsiveness, and incremental validity.
Candidate Multidomain Architecture
A research Coherence Profile may sample Meaning & Context, Nervous System Regulation, Metabolic & Endocrine Coordination, Structural & Tissue Organization, and Cellular & Biochemical Function, but every included measure must have a defensible rationale and adequate measurement properties for the target population and context. The five-layer framework does not require equal weighting or one measure from every layer.
Dynamic Coherence
Because CRF coherence is context-sensitive, dynamic measurements may be more informative than resting snapshots. Candidate designs include baseline → standardized demand → transition → recovery → second matched demand. Coordination can then be studied in response timing, cross-variable coupling, recovery trajectories, output-cost relationships, and retained subsequent-demand capability.
Cross-Layer Coupling Is Not Automatically Coherence
Correlation, synchrony, or network coupling between variables is not inherently beneficial. Strong coupling can be adaptive, neutral, or maladaptive depending on context and function. A valid coherence measure must therefore relate observed coordination to prespecified functional consequences rather than reward synchrony for its own sake.
Weighting and Aggregation
Equal weights should not be assumed. Data-driven weights can overfit, while expert-derived weights can encode subjective preferences. Any weighting scheme should be prespecified or transparently derived, cross-validated, sensitivity-tested, and externally replicated. Results should remain inspectable at the component level.
Normalization and Directionality
Measures with different units require transparent scaling before aggregation. Investigators must also specify directionality: higher values are not universally better for HRV, cortisol, arousal, variability, muscle activity, glucose, inflammatory markers, or other physiological variables. Context-specific target ranges or response patterns may be more defensible than simple maximization.
Reliability Before Change Scores
An index cannot meaningfully detect change if its measurement error is too large. Test-retest reliability, within-person variability, smallest detectable change, and other error characteristics should be established before interpreting individual longitudinal changes as improved or reduced coherence.
Responsiveness and Meaningful Change
Responsiveness asks whether a measure can detect change in the construct when change has occurred. A statistically significant score change is not automatically meaningful. Thresholds for meaningful change should be empirically justified rather than chosen for marketing, program grading, or convenience.
Measurement Invariance
A proposed index should be tested for relevant invariance across groups and contexts before comparisons are interpreted. Age, sex, culture, language, health status, medication, fitness, device platform, measurement setting, and other factors may change distributions or measurement relationships.
External Validation and Locking
Once a candidate model is selected, its scoring algorithm should be locked before confirmatory testing in new data. Performance in the development sample is insufficient. Independent replication should precede claims that a Coherence Index is validated.
Incremental Validity
A Coherence Index should outperform or add information beyond simpler established measures for the intended use. If a stress questionnaire, functional test, HRV measure, sleep metric, resilience measure, or conventional risk model performs equally well, the additional complexity of a multidomain index may not be justified.
Device Firewall
No proprietary wellness device should define the CRF construct it is purported to measure. Biofeedback, HRV, PEMF, scalar, frequency, red-light, wearable, or other device outputs may be candidate variables only when their measurement properties and interpretation are independently defensible.
Clinical Boundary
A Coherence Profile or future index must not be presented as diagnosing disease, identifying hidden pathology, determining treatment need, or replacing clinical assessment unless separately validated and authorized for such use. Early ICR measures are research and wellness-evaluation tools.
Emerging Scientific Competition
Recent systems-level proposals are also exploring multidomain indices of biological state and adaptive capacity. Their existence increases the burden on CRF to demonstrate distinct construct validity and incremental value. Similar terminology in another framework should be cited and compared, not treated as confirmation of CRF.
Minimum Validation Program
A credible program should progress through construct specification; expert and participant content review where relevant; feasibility; reliability and measurement-error studies; structural testing where appropriate; prespecified construct validation; responsiveness; longitudinal prediction; incremental validity; external validation; and independent replication. Clinical utility should be evaluated only after the measurement foundation is adequate.
Canonical Public Definition
ICR does not currently have a validated whole-person Coherence Score. Coherence should first be studied as a profile of clearly defined, independently measured variables and dynamic relationships. A single score should be created only if rigorous measurement research shows that combining those variables is scientifically justified.
26. Conclusion
Coherence is measurable only if the construct is defined narrowly enough to fail. Network Physiology demonstrates that multisystem coordination is real, dynamic, state dependent, and technically measurable. It does not provide a ready-made whole-person coherence score.
ICR should therefore proceed in stages: define the construct, validate components, collect synchronized multidomain data, test challenge and recovery, derive models cautiously, compare them with simpler alternatives, and require external replication.
Until those steps are complete, the scientifically correct position is explicit: CRF proposes coherence as a conceptual property of coordinated regulation; ICR does not yet possess a validated measure of whole-person coherence.
Declarations
Author and originator: David Fischer. Institutional affiliation: Institute for Coherence and Regulation (ICR), Knightdale, North Carolina, USA.
Competing interests: The author is associated with an organization developing educational materials, practitioner training, wellness services, and intellectual property related to CRF. Any future coherence instrument should be independently evaluated, with full disclosure of intellectual and commercial interests.
Ethics: This conceptual white paper reports no human-subject research. Data availability: No dataset was generated.
Canonical designation: ICR-WP-009, Publication Version 1.0, September 2026.
Publication note: Publication Version 1.0 aligns WP-009 with WP-001, WP-023, WP-024, and WP-025; makes Profiles Before Scores the default; adds reflective-versus-formative model requirements; adopts established measurement-science properties; and strengthens reliability, invariance, external-validation, incremental-validity, and device boundaries.
References
Bartsch, R. P., Liu, K. K. L., Bashan, A., & Ivanov, P. C. (2015). Network Physiology: How Organ Systems Dynamically Interact. PLOS ONE, 10(11), e0142143. https://doi.org/10.1371/journal.pone.0142143
Ivanov, P. C. (2021). The New Field of Network Physiology: Building the Human Physiolome. Frontiers in Network Physiology, 1, 711778. https://doi.org/10.3389/fnetp.2021.711778
Lehnertz, K., Bröhl, T., & Rings, T. (2020). The Human Organism as an Integrated Interaction Network: Recent Conceptual and Methodological Challenges. Frontiers in Physiology, 11, 598694. https://doi.org/10.3389/fphys.2020.598694
Munson, R., Sands, M., Srinivasan, S., Charalambous, C., & Bhogal, A. S. (2026). Decrypting network physiology for clinical practice. Frontiers in Network Physiology, 6, 1854895. https://doi.org/10.3389/fnetp.2026.1854895
Smith, R., Thayer, J. F., Khalsa, S. S., & Lane, R. D. (2017). The hierarchical basis of neurovisceral integration. Neuroscience & Biobehavioral Reviews, 75, 274–296. https://doi.org/10.1016/j.neubiorev.2017.02.003
Thayer, J. F., Hansen, A. L., Saus-Rose, E., & Johnsen, B. H. (2009). Heart rate variability, prefrontal neural function, and cognitive performance: The neurovisceral integration perspective. Annals of Behavioral Medicine, 37(2), 141–153. https://doi.org/10.1007/s12160-009-9101-z
Mokkink, L. B., Terwee, C. B., Patrick, D. L., Alonso, J., Stratford, P. W., Knol, D. L., Bouter, L. M., & de Vet, H. C. W. (2010). The COSMIN study reached international consensus on taxonomy, terminology, and definitions of measurement properties for health-related patient-reported outcomes. Journal of Clinical Epidemiology, 63(7), 737-745. https://doi.org/10.1016/j.jclinepi.2010.02.006
Furr, R. M. (2021). Psychometrics: An introduction (4th ed.). SAGE Publications.
Appendix A — Minimum Requirements Before Naming a Coherence Index
Frozen construct definition and intended use.
Prespecified domains and rationale.
Reliable component measurements.
Defined collection state and time scale.
Artifact and missing-data rules.
Independent validation outcomes.
Comparator models.
Structural/construct validity.
Out-of-sample predictive performance.
Calibration and uncertainty.
External validation.
Responsiveness testing.
Independent replication.
Public scoring algorithm and version control.
Explicit retirement criteria.
Appendix B — Canonical Public Statement
The Coherence & Regulation Framework uses coherence as a conceptual description of coordinated, timely, context-appropriate regulation. ICR does not currently claim to possess a validated whole-person Coherence Score. Any future measurement instrument will require formal development, independent validation, and transparent publication before it is represented as a measure of coherence.