Rationale, aims and objectives. This paper addresses the journal’s concern with the conceptual foundations of clinical decision making and the roles of evidence and guidelines. It argues that asking whether artificial intelligence can replace physician judgment is misdirected. The relevant variable is not machine capability alone, but whether a clinical domain treats its categories as settled at the point of care. The paper examines how evidence-based medicine, quality measurement, and liability make some forms of reasoning available for automation. Method. The argument uses conceptual analysis and historically grounded cases from sepsis, psychiatry, opioid prescribing, pulse oximetry, and clinical artificial intelligence. Cases were selected to vary in category stability, enforcement, and information architecture, not to estimate prevalence. Results. Evidence-based practice requires categories stable enough to define trial populations, guideline eligibility, and quality denominators. Guideline construction behaves as latent-state estimation indexed to an objective, so category boundaries move when objectives change. Once a category is operationally fixed, reasoning within it becomes a reproducible mapping from recorded features to permitted actions. Two variables organize the pattern: ontological settlement, the degree to which a category is treated as fixed in practice, and liability salience, the visibility and personal attribution of departure from it. Settlement affects whether a decision function is learnable; liability salience influences whether displacement arrives through enforcement or economic substitution. The function least represented is judgment about the category itself. Conclusion. The clinician’s residual advantage is positional, not mystical or permanent. Clinicians often receive encoded measurements together with information absent from the record, allowing them to detect discordance between the patient and the category. Medicine should cultivate category-failure detection in postgraduate training, assess it, and create a structured route through which recurrent judgments of misfit become reviewable research questions.
Background/Objectives: We compared reported complementary health approach (CHA) use by self-reported clinician diagnosis of chronic fatigue syndrome (CFS), with favorability and expected helpfulness as additional exploratory outcomes. Methods: This secondary analysis included 3022 adults from a 2024 US nonprobability online quota sample. Diagnoses were not clinically verified, and formal ME/CFS criteria were not assessed. Logistic models adjusted for ten covariates, using Firth regression for sparse or unstable estimates. Revision-time Benjamini–Hochberg correction defined nine question-based families by outcome and approach grouping. Results: Among 167 respondents reporting CFS, adjusted odds of use were higher for physical (aOR 2.22, 95% CI 1.60–3.07), nutritional (2.25, 1.34–3.80), combined (1.89, 1.24–2.88), and systems approaches (1.99, 1.32–3.01); all q < 0.05. Nineteen individual CHAs and all seven conventional approaches met q < 0.05 for use. Any-CHA use was 98.2% versus 94.6% (aOR 2.46, 95% CI 0.97–8.91; q = 0.065). Adding the count of other self-reported clinician diagnoses attenuated category-use associations; their CIs included 1. Psychological-category favorability was lower using a mean >3.5 cutoff, but its FDR threshold status depended on midpoint handling. No expectancy association met q < 0.05. Conclusions: Respondents reporting CFS reported greater use of multiple approaches. Additional-diagnosis adjustment changed the estimates, but the count cannot distinguish morbidity from healthcare contact or diagnosis ascertainment. The findings do not establish CFS-specific treatment use, substitution, or effectiveness.
Highlights • A theoretical synthesis reframes chronic pain as maladaptive precision weighting • PAG columnar architecture is proposed as a coping-specific precision gate • Subanesthetic ketamine is proposed as one candidate dual cortical-thalamic opener • Open-Shape-Stabilize yields timing-dependent, falsifiable predictions across pain, depression, and anxiety • Shared precision dysregulation may help explain pain-depression-anxiety comorbidity Abstract Chronic pain frequently co-occurs with depression and anxiety, and these syndromes may partly share disturbances in how the brain assigns precision to predictions and prediction errors. This theoretical synthesis proposes that chronic pain-particularly neurogenic, neuropathic, and nociplastic phenotypes-can be understood as a maladaptive sampling regime in which interoceptive prediction errors are persistently assigned excessive precision, trapping the system in a hypervigilant attractor. We situate this proposal within hierarchical oscillatory architecture: slower rhythms define temporal windows for top-down predictions, faster rhythms express bottom-up prediction errors, and phase-amplitude coupling coordinates their interaction. We interpret thalamocortical dysrhythmia as a candidate timing-level substrate for this rigidity and propose that the periaqueductal gray (PAG) may function as a coping-specific gate for interoceptive precision through its columnar organization. We further identify dopaminergic precision modulation as a plausible bridge between affective state and nociceptive gain. On this basis, we advance an Open-Shape-Stabilize intervention hypothesis. In the Open phase, subanesthetic ketamine is proposed as one candidate means of transiently destabilizing maladaptive precision weighting through complementary cortical and thalamic mechanisms. In the Shape phase, cortical neuromodulation and short-course peripheral afferent reduction are hypothesized to bias the system toward a safer latent state. In the Stabilize phase, synaptic, homeostatic, computational, and cognitive supports are proposed to reduce relapse. The OSS sequence is a hypothesis-generating translational framework; its named components represent separable experimental operations, not a required treatment package. The framework generates specific, falsifiable predictions concerning timing, convergence, comorbidity, and descending modulation.