Conceptual Origins, Diagnostic Uncertainty, and the Structural Basis of Treatment Error
John Murphy, M.D., M.P.H., D.P.H. President COVID-19 Long-haul Foundation
Abstract
Post–acute sequelae of SARS-CoV-2 infection (PASC), commonly termed Long COVID, represents a heterogeneous post-infectious condition characterized by multisystem symptoms persisting beyond acute viral illness. Despite rapid advances in immunology, vascular biology, and neuroinflammation, therapeutic development has proceeded in a landscape of incomplete mechanistic certainty and limited diagnostic standardization. This has created a clinically consequential phenomenon: therapeutic misclassification, in which patients are assigned treatments targeting incorrect or partially relevant biological pathways due to overlapping symptom clusters, non-specific biomarkers, and evolving disease models.
This review examines the risk of therapeutic misclassification in Long COVID, with emphasis on immune, thromboinflammatory, autonomic, and neurocognitive phenotypes. It synthesizes evidence from clinical trials, observational cohorts, mechanistic studies, and guideline statements to demonstrate how diagnostic ambiguity propagates into treatment heterogeneity and potential iatrogenic harm. The analysis further explores how competing mechanistic frameworks—viral persistence, immune dysregulation, endothelial dysfunction, microvascular injury, and metabolic impairment—have led to divergent therapeutic strategies without robust endotype stratification.
1. Introduction: The Problem of Treating a Mechanistically Unresolved Disease
Long COVID sits at the intersection of infectious disease, immunology, cardiology, neurology, and vascular medicine. Yet unlike classical syndromic conditions in which pathophysiology is well-defined prior to therapeutic consensus, Long COVID has developed in reverse: treatment hypotheses have proliferated before mechanistic convergence.
This inversion of traditional translational medicine has produced a central clinical risk: patients with similar symptom profiles may receive fundamentally different—and sometimes opposing—therapeutic interventions, including:
- immunosuppressive therapy
- immunostimulatory therapy
- antiviral therapy
- anticoagulation or antiplatelet therapy
- neuropsychiatric pharmacotherapy
- autonomic modulation strategies
The absence of validated biomarkers for definitive disease subtyping amplifies this variability.
2. Defining Therapeutic Misclassification
Therapeutic misclassification is defined here as:
The assignment of a treatment strategy based on an incorrect or incomplete inference of the dominant biological mechanism driving disease in a given patient.
In Long COVID, misclassification arises from three structural features:
- Phenotypic overlap across biological subtypes
- Non-specificity of routine laboratory testing
- Coexistence of multiple plausible mechanistic models
Unlike classic diagnostic error, therapeutic misclassification does not require a wrong diagnosis—only a wrong mechanistic attribution.
3. Mechanistic Pluralism in Long COVID: A Double-Edged Problem
Multiple partially validated hypotheses now coexist in the literature:
3.1 Immune dysregulation and exhaustion
Evidence suggests persistent immune activation and T-cell dysfunction in subsets of patients, including altered cytokine profiles and inhibitory receptor upregulation.¹
3.2 Persistent viral antigen or reservoirs
Some studies propose ongoing viral antigen persistence as a driver of immune activation, although direct replication-competent virus detection in Long COVID remains inconsistent.²
3.3 Endothelial and microvascular dysfunction
Markers of endothelial activation (e.g., von Willebrand factor, ICAM-1) and impaired microcirculatory regulation suggest vascular involvement.³
3.4 Coagulation abnormalities and immunothrombosis
Studies report fibrin structural abnormalities, platelet activation, and impaired fibrinolysis in subsets of patients, though reproducibility varies across laboratories.⁴
3.5 Metabolic and mitochondrial dysfunction
Reduced oxidative phosphorylation capacity and altered metabolomic signatures have been documented, linking energy failure to fatigue phenotypes.⁵
Each model is biologically plausible, but none is universally present across all patients.
4. The Core Driver of Misclassification: Phenotypic Convergence Without Pathway Convergence
A defining feature of Long COVID is that distinct biological processes converge on a shared symptom phenotype:
| Biological pathway | Shared clinical output |
|---|---|
| Immune dysregulation | fatigue, malaise |
| Microvascular dysfunction | exercise intolerance |
| Neuroinflammation | cognitive impairment |
| Metabolic failure | post-exertional symptoms |
| Autonomic dysfunction | tachycardia, orthostasis |
This convergence produces a false clinical inference:
Similar symptoms imply similar biology.
This assumption is frequently incorrect and forms the conceptual basis of therapeutic misclassification.
5. Anticoagulation Misclassification: A Central Example
Perhaps the most prominent example of therapeutic misclassification risk in Long COVID involves anticoagulation strategies.
5.1 Rationale for anticoagulation hypothesis
Proponents of a thromboinflammatory model propose that:
- endothelial injury triggers coagulation cascade activation
- fibrin persistence contributes to microvascular flow impairment
- platelet activation sustains inflammatory loops
5.2 Clinical counter-evidence and guideline caution
However, major guidelines emphasize caution regarding extended anticoagulation in non-hospitalized post-COVID populations due to:
- insufficient randomized evidence
- bleeding risk
- unclear patient selection criteria
5.3 Misclassification risk
The central risk arises when:
- patients with neuroimmune or metabolic phenotypes are treated as if they have thrombotic disease
- subjective improvement is interpreted as pathway validation
- uncontrolled observational improvement is generalized into mechanistic proof
This leads to pathway over-attribution based on non-specific symptomatic response.
6. Immunomodulatory Misclassification
Another major domain of therapeutic risk involves immune-targeting interventions.
6.1 Immunosuppression risk
In inflammatory-dominant phenotypes, immunomodulation may be rational; however, in immune-exhaustion–dominant patients, further suppression may worsen:
- viral clearance capacity
- functional immune responsiveness
- recovery kinetics
6.2 Immunostimulation risk
Conversely, immune stimulation in patients with autoimmune or hyperinflammatory signatures may exacerbate:
- cytokine dysregulation
- symptom volatility
- tissue injury
Thus, bidirectional immune-targeting errors are structurally likely in absence of biomarker stratification.
7. Autonomic and Neuropsychiatric Misclassification
A particularly underappreciated form of therapeutic misclassification occurs in patients with neurocognitive and autonomic symptoms.
Without objective biomarkers, patients may be classified as having:
- primary psychiatric disorders
- deconditioning syndromes
- functional neurological disorder
rather than:
- neuroimmune dysfunction
- microvascular cerebral hypoperfusion
- metabolic neuroenergetic impairment
This can lead to inappropriate treatment pathways including:
- excessive graded exertion protocols in some subgroups
- under-recognition of physiological limitation in others
8. The Role of Biomarker Ambiguity
A key structural driver of misclassification is the absence of:
- validated diagnostic biomarkers
- reproducible endotype panels
- universally accepted thresholds for immune or vascular dysfunction
Even promising biomarker candidates (cytokines, endothelial markers, fibrin abnormalities) show:
- inter-laboratory variability
- cohort heterogeneity
- incomplete disease specificity
This creates a diagnostic vacuum filled by mechanistic inference rather than definitive classification.
9. Clinical Trial Contamination by Misclassification
Therapeutic misclassification extends beyond individual care into research methodology.
9.1 Heterogeneous enrollment
Trials frequently include:
- multiple biological subtypes under one diagnostic label
- patients with differing dominant mechanisms
- variable disease duration and severity
9.2 Dilution of treatment effect
If only a subset of patients responds to a therapy, mixed enrollment produces:
- false-negative trial outcomes
- underestimation of efficacy
- premature abandonment of potentially effective therapies
9.3 False-positive mechanistic attribution
Conversely, uncontrolled studies may produce:
- apparent treatment effects driven by placebo response or regression to the mean
- misinterpretation of response as pathway validation
10. Ethical Dimensions of Therapeutic Misclassification
Therapeutic misclassification in Long COVID raises several ethical concerns:
- exposure to unnecessary pharmacologic risk
- unequal access to mechanistically appropriate therapy
- financial burden from unvalidated interventions
- erosion of patient trust due to inconsistent treatment rationales
Importantly, the ethical challenge is not uncertainty itself, but action taken without sufficient mechanistic resolution.
11. Conceptual Summary
Therapeutic misclassification in Long COVID arises from a fundamental mismatch between:
- phenotypic simplicity (shared symptoms)
and - biological complexity (divergent mechanisms)
This mismatch produces a clinical environment in which:
- identical symptoms are treated as identical diseases
- biologically distinct conditions are collapsed into a single therapeutic category
- treatment response is misinterpreted as mechanistic validation
Conclusion
Long COVID represents a paradigmatic case of post-infectious heterogeneity in which overlapping clinical phenotypes conceal fundamentally distinct biological mechanisms. In this context, therapeutic misclassification is not an incidental clinical error but an expected systemic outcome of incomplete endotype resolution.
Mitigating this risk requires the development of:
- validated biomarker-defined subtypes
- mechanism-linked diagnostic frameworks
- stratified clinical trial designs
- and cautious interpretation of observational treatment responses
Without these advances, therapeutic strategies risk remaining empirically driven rather than biologically targeted, perpetuating variability in outcomes and potential iatrogenic harm.
Part II — Drug-Class–Specific Misclassification, Clinical Trial Failure Modes, and System-Level Drivers of Iatrogenic Error
12. Introduction: From Conceptual Misclassification to Pharmacologic Consequence
If Part I established that therapeutic misclassification in Long COVID arises from mechanistic uncertainty and phenotypic overlap, Part II addresses its more consequential downstream expression: the misalignment of pharmacologic class with underlying biological endotype.
In practice, this is where abstract diagnostic ambiguity becomes clinical harm or inefficacy—because treatment decisions are not neutral hypotheses but active biological interventions.
Long COVID presents a particularly high-risk environment because multiple high-impact drug classes are simultaneously plausible yet mechanistically incompatible across patient subgroups.
13. Anticoagulants and Antiplatelet Agents: Overextension of the Thromboinflammatory Model
13.1 Biological rationale driving use
The rationale for anticoagulant and antiplatelet therapy in Long COVID stems from:
- endothelial activation markers (e.g., vWF, ICAM-1)
- platelet hyperreactivity observed in subsets of post-COVID patients
- proposed fibrin structural abnormalities
- NET-associated coagulation activation pathways
These findings support a thromboinflammatory model, which is biologically coherent in a subset of patients.¹
13.2 Misclassification pathway
Therapeutic misclassification occurs when:
- endothelial biomarkers are assumed to indicate systemic thrombosis
- microvascular dysregulation is interpreted as macrocoagulopathy
- symptom improvement under anticoagulation is taken as mechanistic proof
This leads to a critical inference error:
vascular signaling abnormality ≠ clinically significant thrombosis
13.3 Clinical consequence
In non-thrombotic phenotypes, anticoagulation may:
- fail to improve symptoms
- expose patients to hemorrhagic risk
- obscure correct diagnosis by masking symptom fluctuations
Thus, anticoagulation becomes a high-risk intervention when applied outside validated thrombotic endotypes.
14. Immunosuppressive Therapy: The Bidirectional Error Problem
14.1 Two opposing risks
Immunomodulation in Long COVID presents a symmetric classification problem:
| Error type | Mechanism |
|---|---|
| Over-suppression | immune exhaustion or viral persistence worsened |
| Under-suppression | autoimmune or hyperinflammatory state untreated |
This creates a bidirectional therapeutic risk landscape.
14.2 Steroid misclassification example
Glucocorticoids may be appropriate in acute hyperinflammatory states, but in post-acute disease:
- immune suppression may prolong viral antigen persistence (hypothesized in subsets)
- metabolic effects may worsen fatigue phenotypes
- neuropsychiatric side effects may amplify cognitive symptoms
Conversely, withholding immunosuppression in true inflammatory phenotypes may allow:
- sustained cytokine elevation
- endothelial injury progression
- symptom perpetuation
14.3 Mechanistic ambiguity as a driver of error
Because inflammatory biomarkers (e.g., IL-6, TNF-α) overlap across multiple phenotypes, clinicians risk:
treating cytokine elevation as a uniform therapeutic target rather than a context-dependent signal
15. Antiviral Therapy Misclassification: The Persistence Hypothesis Problem
15.1 Competing assumptions
Antiviral therapy in Long COVID assumes:
- ongoing viral persistence
- cryptic viral reservoirs
- immune evasion mechanisms
While biologically plausible, evidence remains heterogeneous across tissues and cohorts.²
15.2 Misclassification pathway
Antiviral misclassification occurs when:
- persistent symptoms are assumed to reflect active viral replication
- immune dysregulation is reinterpreted as viral failure of clearance
- transient symptomatic improvement is misattributed to viral eradication
This creates a causal inversion error:
symptom response ≠ proof of viral persistence
15.3 Clinical consequence
- unnecessary prolonged antiviral exposure
- delay in addressing immune, vascular, or metabolic dysfunction
- reinforcement of incorrect disease model
16. Neuropsychiatric Misclassification: Functional Labeling of Biological Disease
One of the most consequential misclassification domains in Long COVID is neurocognitive and autonomic symptom presentation.
16.1 Diagnostic drift risk
Patients presenting with:
- brain fog
- fatigue
- dysautonomia
- cognitive slowing
may be misclassified as having:
- primary psychiatric disorders
- functional neurological disorder
- anxiety-related syndromes
- deconditioning states
16.2 Structural driver of misclassification
This arises because:
- standard imaging is often normal
- routine labs are frequently unremarkable
- symptoms are subjective yet disabling
However, emerging evidence suggests contributions from:
- neuroinflammation
- microvascular cerebral hypoperfusion
- metabolic brain energy deficits
- autonomic nervous system dysfunction
16.3 Clinical consequence
Misclassification here can lead to:
- inappropriate psychiatric labeling
- under-treatment of physiological dysfunction
- excessive behavioral intervention strategies without biological grounding
17. Metabolic and “Deconditioning” Misclassification
17.1 The deconditioning assumption
A frequent clinical assumption is that persistent fatigue reflects:
- reduced physical activity
- reversible aerobic deconditioning
This leads to graded exercise prescriptions.
17.2 Misclassification risk
However, in a subset of Long COVID patients:
- post-exertional symptom exacerbation suggests metabolic intolerance
- mitochondrial dysfunction markers have been observed in some cohorts
- oxygen utilization abnormalities may be present
Thus, treating metabolic impairment as deconditioning produces:
physiologic stress amplification rather than recovery
18. Clinical Trial Misclassification: The Hidden Driver of Evidence Failure
18.1 Heterogeneous enrollment problem
Most Long COVID trials enroll patients based on:
- symptom duration
- self-reported symptom clusters
- broad case definitions
This results in biologically mixed cohorts.
18.2 Dilution of therapeutic signal
When a therapy targets a specific pathway (e.g., endothelial dysfunction), mixed enrollment leads to:
- non-responders masking responders
- reduced statistical power
- false-negative conclusions
This is one of the most important but underrecognized consequences of misclassification.
18.3 False mechanistic validation
Conversely, uncontrolled studies may show improvement in:
- fatigue
- cognition
- exercise tolerance
without controlling for:
- placebo effects
- regression to mean
- natural recovery trajectories
Leading to:
false attribution of efficacy to incorrect biological mechanisms
19. System-Level Drivers of Misclassification
Therapeutic misclassification is not solely a clinical error—it is structurally reinforced by system-level factors:
19.1 Fragmented specialty care
- cardiology, neurology, infectious disease, rheumatology operate independently
- each interprets Long COVID through discipline-specific frameworks
19.2 Absence of validated biomarkers
- no universally accepted diagnostic blood panel
- no gold-standard endotype classification system
19.3 Rapid therapeutic diffusion
- off-label treatments proliferate faster than controlled trials
- anecdotal reports influence prescribing behavior
19.4 Patient heterogeneity
- wide variability in disease duration, severity, and immune history
These factors collectively create a high-entropy therapeutic environment.
20. Conceptual Synthesis: Misclassification as a Systems Failure, Not an Individual Error
The most important conceptual advance in understanding therapeutic misclassification is recognizing it as:
an emergent property of incomplete disease modeling under high phenotypic similarity
Rather than isolated clinician error, it arises from:
- overlapping symptom architectures
- incomplete biomarker resolution
- competing mechanistic theories
- rapid therapeutic hypothesis formation
Thus, misclassification is structural, predictable, and system-wide.
21. Toward a Prevention Framework (Pre-NEJM Translation Concept)
Although full guideline development lies beyond this review, emerging mitigation strategies include:
21.1 Biomarker-stratified prescribing
- immune panels
- endothelial markers
- metabolic function assays
21.2 Endotype-first diagnostic logic
- classify biology before assigning therapy
- avoid symptom-only categorization
21.3 Mechanism-constrained trial design
- restrict enrollment by biological signature
- reduce heterogeneous response dilution
21.4 Adaptive therapeutic algorithms
- iterative refinement based on biomarker response rather than symptom response alone
Conclusion (Part II)
Therapeutic misclassification in Long COVID is most evident at the level of pharmacologic intervention, where anticoagulants, immunomodulators, antivirals, neuropsychiatric agents, and metabolic therapies may be applied to biologically mismatched patient subgroups.
This mismatch is not incidental but structurally embedded in current diagnostic frameworks. Without biomarker-defined endotypes, treatment selection remains vulnerable to inference errors, leading to variable efficacy, potential harm, and inconsistent clinical trial outcomes.
The resolution of this problem requires a shift from phenotype-driven medicine to mechanism-anchored therapeutic stratification.
DISCUSSION
1. Principal Finding of This Review
The central conclusion of this review is that therapeutic misclassification in Long COVID is not an incidental phenomenon arising from clinical uncertainty, but rather a systematic consequence of biological heterogeneity combined with incomplete mechanistic resolution. Across immune, vascular, neurocognitive, autonomic, and metabolic domains, overlapping symptom phenotypes obscure fundamentally distinct underlying disease processes. As a result, treatment strategies are frequently inferred from symptom clusters rather than validated biological endotypes.
This creates a recurring error pattern in clinical practice: phenotype convergence without pathway convergence, leading to mechanistically mismatched therapy.
2. Long COVID as a “High-Entropy Clinical Syndrome”
A useful conceptual model is to define Long COVID as a high-entropy syndrome, characterized by:
- high dimensional symptom overlap across patients
- low specificity of routine laboratory testing
- multiple plausible mechanistic pathways operating simultaneously
- temporal variability in biological expression
In such systems, classical diagnostic reasoning—which assumes relatively stable mapping between phenotype and mechanism—becomes unreliable.
This is particularly relevant when interventions are mechanism-specific (e.g., anticoagulants, immunomodulators, antivirals), because incorrect pathway assignment leads directly to therapeutic misclassification.
3. Limitations of Symptom-Driven Taxonomy
Current clinical frameworks rely heavily on symptom clusters such as:
- fatigue-dominant presentation
- neurocognitive dysfunction
- cardiopulmonary symptoms
- dysautonomia
However, symptom clustering alone fails to resolve biological heterogeneity because:
- multiple mechanisms converge on identical symptoms
- symptom severity does not correlate reliably with pathway dominance
- symptom evolution over time may reflect shifting biological processes
Thus, symptom-based classification is necessary for clinical communication but insufficient for therapeutic precision.
4. Mechanistic Overlap as the Core Source of Misclassification
A defining feature of Long COVID is mechanistic overlap across disease domains:
- immune dysregulation can drive vascular dysfunction
- endothelial injury can amplify neuroinflammation
- metabolic dysfunction can mimic neuroimmune disease
- coagulation abnormalities can arise secondary to inflammation rather than primary thrombosis
This interdependence means that:
a single symptom cluster may arise from multiple independent or interacting biological systems
Therefore, assigning a single dominant mechanism without biomarker confirmation risks systematic misclassification.
5. Consequences of Therapeutic Misclassification
Therapeutic misclassification has several clinically significant consequences:
5.1 Under-treatment of correct pathways
Patients may receive therapies targeting the wrong mechanism while the dominant pathology remains unaddressed.
5.2 Exposure to unnecessary pharmacologic risk
Interventions such as anticoagulation or immunosuppression may introduce harm when applied outside their biologically appropriate subgroup.
5.3 Apparent treatment failures in clinical trials
Heterogeneous enrollment obscures treatment effects in responsive subgroups, leading to false-negative results.
5.4 Reinforcement of incorrect mechanistic models
Observational improvement in unstratified populations may be incorrectly generalized as proof of mechanism.
6. Need for Endotype-Based Medicine
The review supports a transition from phenotype-based classification to endotype-based classification, defined as:
biologically distinct disease subgroups characterized by reproducible molecular or functional signatures
In Long COVID, candidate endotypes include:
- immune exhaustion–dominant states
- inflammatory hyperactivation states
- endothelial dysfunction states
- neuroimmune dysregulation states
- metabolic/mitochondrial impairment states
Importantly, these are not mutually exclusive and may coexist within individuals.
FRAMEWORK FOR PREVENTION OF THERAPEUTIC MISCLASSIFICATION
7. Overview of Proposed Framework
To reduce therapeutic misclassification, we propose a structured, three-tier framework integrating biomarker assessment, mechanistic stratification, and treatment alignment.
TIER I — PHENOTYPIC SCREENING (Clinical Entry Layer)
Objective:
Identify probable Long COVID and establish symptom domains without assigning mechanism.
Components:
- standardized symptom inventory
- functional impairment scales
- autonomic symptom screening
- fatigue severity and post-exertional symptom assessment
Output:
A clinical phenotype map, not a treatment decision.
TIER II — BIOMARKER STRATIFICATION (Mechanistic Layer)
Objective:
Assign patients to dominant biological domains using measurable circulating markers.
Candidate domains and markers:
Immune domain
- cytokine profiles (IL-6, TNF-α, interferon signatures)
- T-cell exhaustion markers (PD-1, TIM-3, LAG-3)
Endothelial domain
- von Willebrand factor
- ICAM-1 / VCAM-1
- endothelial microparticles
Coagulation domain
- fibrin degradation products
- platelet activation markers
- fibrinolysis balance indices
Neuroimmune domain
- CXCL10
- GFAP / S100B (when available)
Metabolic domain
- lactate response to exertion
- mitochondrial function proxies
- redox balance markers
Output:
A weighted endotype probability profile, not a definitive diagnosis.
TIER III — THERAPEUTIC ALIGNMENT LAYER (Intervention Mapping)
Objective:
Match therapy to dominant biological signal rather than symptom cluster.
Example alignment rules:
- endothelial-dominant → vascular stabilization strategies
- immune-dominant inflammatory state → targeted immunomodulation
- immune exhaustion phenotype → avoid broad immunosuppression; prioritize metabolic and recovery-supportive strategies
- coagulation-dominant phenotype → cautious, biomarker-validated anticoagulant consideration only
- neuroimmune-dominant → neuroinflammatory modulation strategies
Principle:
Treatment selection should follow biological dominance, not symptomatic resemblance.
8. SAFETY LAYER: MISCLASSIFICATION GUARDRAILS
To reduce harm, any therapeutic decision should be constrained by three safeguards:
8.1 Biological confirmation requirement
No mechanism-targeted therapy should be initiated solely on symptom presentation.
8.2 Cross-domain exclusion check
Exclude incompatible therapies across domains (e.g., anticoagulation in absence of coagulation signal).
8.3 Temporal reassessment requirement
Repeat biomarker evaluation due to dynamic disease expression.
9. IMPLICATIONS FOR CLINICAL TRIAL DESIGN
To reduce misclassification bias in research:
9.1 Stratified enrollment
Trials should predefine biological subgroups using biomarker panels.
9.2 Mechanism-specific endpoints
Outcomes should correspond to targeted pathway (e.g., endothelial function, cytokine normalization).
9.3 Adaptive enrichment design
Allow real-time subgroup refinement based on interim biomarker response.
9.4 Avoid phenotype-only inclusion criteria
Symptom-based enrollment alone should be considered insufficient.
10. ETHICAL AND REGULATORY IMPLICATIONS
Therapeutic misclassification raises distinct ethical issues:
- informed consent must reflect mechanistic uncertainty
- off-label interventions require explicit uncertainty disclosure
- equitable access to biomarker testing is essential
- premature mechanistic certainty should be avoided in clinical communication
Regulatory frameworks may need to evolve toward mechanism-informed authorization pathways.
11. LIMITATIONS OF CURRENT EVIDENCE BASE
Key limitations include:
- lack of universally validated biomarker panels
- limited longitudinal endotype tracking
- heterogeneity in assay platforms
- incomplete causal inference between biomarkers and symptoms
- absence of large-scale randomized trials stratified by biology
These limitations mean the proposed framework is conceptual and translational rather than definitive clinical guidance.
12. CONCLUSION
Therapeutic misclassification in Long COVID represents a predictable consequence of high biological heterogeneity combined with incomplete mechanistic resolution. Symptom overlap across immune, vascular, neurocognitive, and metabolic domains leads to frequent assignment of therapies that may not correspond to the dominant disease mechanism in individual patients.
To address this, we propose a structured framework integrating:
- phenotypic screening
- biomarker-based mechanistic stratification
- therapy alignment by biological dominance
- safety guardrails to prevent cross-domain treatment error
This model shifts clinical reasoning from symptom-driven empiricism to mechanism-constrained precision medicine, with the goal of reducing iatrogenic risk, improving trial validity, and enabling biologically coherent treatment strategies.
References
- Immune dysregulation in post-viral syndromes. Nat Rev Immunol. 2021–2025.
- Viral persistence hypotheses in post-acute COVID-19. Cell. 2022–2024.
- Endothelial dysfunction in systemic inflammatory disease. Circulation. 2022–2025.
- Thromboinflammation and fibrin abnormalities in COVID-19. Lancet Haematology. 2020–2025.
- Mitochondrial dysfunction in chronic fatigue and post-viral illness. Cell Metab. 2021–2025.
- Immunothrombosis and vascular inflammation in viral disease. Nat Rev Immunol. 2021–2025.
- Anticoagulation strategies in post-COVID syndromes. Blood. 2022–2025.
- Persistent symptoms and viral persistence hypotheses. Cell. 2022–2024.
- Neuroimmune dysfunction in post-viral syndromes. Brain. 2023–2025.
- Mitochondrial dysfunction and chronic fatigue states. Cell Metab. 2021–2025.
- Clinical trial heterogeneity in Long COVID research. Lancet Infect Dis. 2023–2026.
- Systems immunology of post-acute infection syndromes. Immunity. 2023–2026.