158、超越传统营养指标:元律医学营养—代谢状态与功能健康的多维计算研究(Beyond Conventional Nutritional Biomarkers: Multidimensional Computation of Nutritional–Metabolic States and Functional Health in Prim-Lex Medicine)

Nutritional assessment faces a structural tension: the global burden of disease simultaneously carries undernutrition and overnutrition, while the dominant diagnostic instruments of clinical nutrition—the GLIM criteria, NRS-2002, MUST, and MNA-SF—are organized around a “screen–diagnose–grade” logic rather than around the question of how multi-system information can be integrated into a testable predictive structure. The GLIM criteria show acceptable diagnostic specificity in hospitalized older adults (pooled specificity 0.88, 95% CI 0.81–0.93), but validation evidence in community-dwelling and nursing-home populations remains insufficient; the choice of screening tool materially affects GLIM-based sensitivity and agreement; and serum albumin and prealbumin, historically used as nutrition markers, have been explicitly characterized by an ASPEN position paper as inflammation-related indicators rather than valid proxies of total body protein or muscle mass. Meanwhile, the evidence for nutritional interventions is highly heterogeneous: NAD⁺ precursors (NMN/NR) have not shown consistent improvements in glycemic or lipid outcomes in human randomized trials, probiotic effects are strictly strain-specific, and multidimensional integration in clinical practice still relies predominantly on clinicians’ synthesis rather than reproducible statistical models. This paper develops a Prim-Lex Multidimensional Nutritional Assessment Framework. The framework uses the Fivefold Prim-Fire of Prim-Lex Theory as a cross-scale theoretical background, the Eight-Dimensional Coherence structure as the state-description layer, and the Prim-Nexus Guidance as an evidence-constrained decision-support layer, mapping energy and metabolic status, anabolic–catabolic dynamics, the food–gut–tissue cross-scale process, nutritional risk stages, five functional modules, structural integrity and transport, eating–sleep–circadian rhythm, and multimodal information integration onto observable or operationalizable nutrition-related variables. The paper does not claim that Prim-Lex Theory has identified a “unified root cause” of nutritional imbalance, nor does it propose that any eight-dimensional score should replace GLIM diagnosis or clinical nutrition support decisions. Its core contribution is threefold: (1) theory development—specifying the mapping between the Fivefold Prim-Fire, the eight-dimensional structure, and nutritional status, together with explicit epistemological boundaries; (2) clinical evidence mapping—using PRISMA 2020 and TRIPOD+AI as methodological anchors to distinguish evidence supported by guidelines or high-quality systematic reviews (E1), evidence from human observational studies not yet recommendation-grade (E2), evidence from human non-target-disease or mechanistic studies without validated nutritional-assessment value (E3), and hypotheses supported only by animal/in vitro or theoretical reasoning (E4); and (3) a falsifiable validation protocol—formulating five pre-registered hypotheses, with incremental predictive value (H1) as the primary test and structural validity (H2), longitudinal value (H3), reproducibility (H4), and clinical net benefit (H5) as supporting tests, specifying benchmark models, calibration, decision curve analysis, external validation, and failure conditions. All mathematical expressions are constrained by explicit definitions, units, directionality conventions, missing-data rules, and evidence grading. Existing thresholds (0.85, 0.70, 0.55) are re-designated as exploratory parameters requiring independent calibration and are not used as clinical diagnostic standards. Phase models are not applied to cross-sectional data. The Prim-Nexus Guidance does not enter the formal testing of H1–H4; its decision-support value is examined through H5 in pre-specified decision scenarios. The anticipated contribution is not a new nutritional therapy, but an operationalizable, falsifiable, and methodologically auditable theory–measurement–validation loop for multidimensional nutritional assessment.