Adherence to a high-quality dietary pattern consistently demonstrates a stronger association with long-term health outcomes than any isolated nutrient. A higher PURE Healthy Diet Score, characterized by abundant intake of fruits, vegetables, nuts, legumes, fish, and whole-fat dairy, is significantly associated with a lower risk of incident CKD. Notably, Genetic risk of CKD, DASH, aMed, AHEI-2010, and hPDI did not significantly modify the association between PURE score and incident CKD^[6]. Conversely, pro-inflammatory and pro-oxidant diets synergistically increase CKD risk, particularly in the early Cardiovascular-Kidney-Metabolic (CKM) stages, while anti-inflammatory/antioxidant diets confer protection via multi-omics pathways linked to lipid metabolism and the extracellular matrix^[7]. These findings advocate for CKM stage-specific dietary interventions and integration of multi-omics biomarkers into CKD prevention.

The health impact of widely consumed beverages such as coffee and tea is highly nuanced and depends critically on the presence or absence of added sweeteners.

For coffee, regardless of preparation type (instant or ground) or the addition of milk, moderate consumption of unsweetened coffee (1-4 drinks per day) is associated with a lower risk of new-onset hypertension. Conversely, the same protective association is not observed for sweetened coffee^[8]. Likewise, moderate intake (2-3 drinks per day) of unsweetened coffee, with or without milk, is linked to a reduced risk of new-onset acute kidney injury (AKI), an effect that persists across different coffee types and is not significantly modified by genetic variation in caffeine metabolism^[9].

A similar pattern emerges for tea: consumption of unsweetened tea, whether plain or with milk, is associated with a lower risk of venous thromboembolism (VTE). However, tea with added sweeteners shows no significant protective association, and genetic differences in caffeine metabolism do not substantially alter these relationships^[10].

These findings underscore a key public health insight: the health effects of coffee and tea appear to be strongly modulated by the addition of sugar or artificial sweeteners, independent of genetic background or specific beverage subtype.

• Fish oil and marine omega-3 fatty acids: Habitual fish oil supplementation is associated with a lower risk of new-onset CKD and kidney stones. For kidney stones, this protective association appears most pronounced in individuals with a low or intermediate genetic predisposition, suggesting a targeted benefit^[11,12]. These findings are corroborated by consistent inverse relationships between higher fish consumption, higher circulating omega-3 polyunsaturated fatty acid (PUFA) concentrations, and reduced CKD risk^[12].

• Fruit intake: Both the quantity and diversity of fruit consumption confer benefits for kidney health. Higher intake of total fruits, citrus fruits, and berries is significantly associated with a reduced risk of incident CKD. Metabolomic studies have identified specific plasma metabolite profiles - including levels of 3-hydroxybutyrate, fatty acid desaturation indices, and omega-3 fatty acid percentages - as significant mediators of these protective associations, linking dietary intake to downstream metabolic changes^[13]. Regarding kidney stones, higher intake of citrus fruits, pome fruits (e.g., apples), and tropical fruits, as well as greater diversity across these three categories, are associated with a lower risk of new-onset kidney stones^[14].

• Ultra-processed foods and added salt: High intake of ultra-processed foods (UPFs) is linked to an increased risk of CKD, with a more pronounced detrimental effect observed in individuals with diabetes^[15]. Separately, the seemingly simple behavior of adding salt to food at the table exhibits a direct, dose-dependent association with the long-term risk of microvascular, cerebrovascular, and cardiovascular diseases, reinforcing the importance of discretionary sodium reduction^[16].

• Food group diversity: Extending beyond the effects of any single food or beverage, a broader principle emerges: greater dietary diversity across food groups itself serves as an independent protective factor. Epidemiological evidence consistently demonstrates that a higher variety within overall food intake is independently associated with a reduced risk of new-onset hypertension and type 2 diabetes^[17,18]. This suggests that the synergistic interactions among a wide array of nutrients and bioactive compounds, inherent in a varied diet, contribute to metabolic resilience beyond what can be achieved by focusing on isolated dietary components.

The research focus has shifted from merely total intake to source and quality. Recent evidence indicates that consuming a greater variety of protein sources is associated with a lower risk of new-onset hypertension^[19] and new-onset diabetes^[20]. Specific circulating amino acid profiles have also been found to be closely associated with the presence and progression of CKD^[21].

Particularly insoluble dietary fiber, both the variety and total amount of its intake have been shown to be independent protective factors against new-onset hypertension^[22] and new-onset diabetes^[23].

The association of carbohydrates with chronic diseases exhibits a clear dual regulatory characteristic involving both “quantity” and “quality.” The percentage of energy from total carbohydrate intake shows significant U-shaped associations with the risk of new-onset diabetes^[24] and new-onset hypertension^[25], with the lowest risk observed at 49%-56% and 50%-55% of energy from carbohydrates, respectively. Simultaneously, the quality of carbohydrates profoundly influences their health effects. Intake of low-quality carbohydrates is not only associated with an increased risk of the aforementioned metabolic diseases but also shows a significant positive correlation with accelerated cognitive decline in the elderly. In contrast, increasing the intake of high-quality carbohydrates demonstrates a protective effect^[26]. These findings collectively establish a core principle: based on maintaining total carbohydrate intake within an appropriate range (approximately 50%-55% of energy), prioritizing high-quality sources and limiting low-quality carbohydrates is a key dietary strategy for preventing metabolic diseases and maintaining cognitive health.

As the core cofactor of one-carbon metabolism, folate’s role extends beyond simply lowering homocysteine (Hcy). Research indicates that natural food folate and synthetic folic acid may have differential long-term health outcomes; for example, their intake shows different relationships with all-cause mortality in individuals with CKD^[27]. This suggests that the “form” of a nutrient is also important. More crucially, folate’s metabolic efficacy does not exist in isolation but is tightly regulated by both genetic and nutritional synergy. Vitamin B₁₂ is not only a cofactor for methionine synthase; its deficiency can lead to a “methyl folate trap,” rendering folate supplementation ineffective unless B₁₂ is also addressed^[28]. The role of vitamin D is even more profound. Studies have found that it can directly enhance intestinal folate absorption and folate transport across the blood-brain barrier by upregulating the expression of the proton-coupled folate transporter (PCFT) and the reduced folate carrier (RFC)^[29]. An animal study further demonstrated that combined supplementation of vitamin D, folate, and B₁₂ could synergistically reverse learning and memory impairment induced by vitamin D deficiency by modulating 27-hydroxycholesterol and S-adenosylmethionine levels^[30]. Collectively, this evidence indicates that one-carbon metabolism is a networked system requiring precise coordination and dynamic balance among multiple nutrients, where a deficiency in any component can impact overall health outcomes.

The role of vitamins extends far beyond the prevention of classical deficiency diseases, with modern epidemiology revealing complex, non-linear, and condition-specific associations with chronic disease risk.

Zinc demonstrates complex, non-linear relationships with health outcomes that vary by both the specific endpoint and the exposure matrix (diet vs. plasma).

• J-shaped association with incident cardiometabolic diseases: Nationwide cohort studies consistently report J-shaped associations between dietary zinc intake and the risk of new-onset hypertension^[4] and type 2 diabetes^[5], indicating an optimal range where both deficiency and excess confer risk.

• Inverse association with disease progression and complications: In contrast, among individuals with established hypertension, higher plasma zinc concentration is associated with better outcomes, showing an inverse linear association with the development of proteinuria (a marker of renal damage)^[54] and the risk of first hemorrhagic stroke^[55]. This suggests that once disease is present, adequate zinc status may be protective against progression and severe complications.

• Potential linear benefit for neurological health: For cognitive aging, higher dietary zinc intake is associated with a lower risk of cognitive decline in older adults^[56], pointing to a potential linear, beneficial role in maintaining neurological function.

Copper intake demonstrates some of the most consistent non-linear patterns across diverse endpoints:

• Kidney and metabolic health: Dietary copper shows a U-shaped association with new-onset CKD^[57] and new-onset hypertension^[58].

• Mortality and cognitive function: The relationship with all-cause mortality follows a J-shaped curve^[59], while its association with cognitive decline in the elderly is non-linear and inverse, with an inflection point around 1.3 mg/day^[60].

• Stroke risk: In hypertensive patients, higher plasma copper concentration is positively associated with the risk of first stroke, particularly among individuals with higher BMI^[61].

Dietary phosphorus intake exhibits typical U-shaped associations with the risk of new-onset hypertension^[2] and type 2 diabetes^[3], establishing an optimal intake range. Furthermore, higher serum phosphate is an independent risk factor for new-onset hyperuricemia^[62], suggesting a pathophysiological link between phosphorus metabolism and purine metabolism disorders.

As a key antioxidant component of glutathione peroxidases, higher plasma selenium levels are consistently associated with a lower risk of first ischemic stroke^[63,64]. Paradoxically, in the context of diabetes, higher plasma selenium has been associated with an increased risk of new-onset diabetes among hypertensive patients^[65], highlighting a potential dual role that may depend on metabolic status.

The health associations of manganese are highly context-dependent, often mediated by obesity and inflammatory status:

• Hypertension: In the general population, higher dietary manganese intake is positively associated with new-onset hypertension^[66].

• CKD and thrombosis in diabetes: Among individuals with poorly controlled diabetes, higher manganese intake is associated with a lower risk of new-onset CKD, an effect largely mediated by reduced obesity^[67]. Similarly, higher manganese intake is inversely associated with incident venous thromboembolism (VTE), primarily mediated through lower obesity and inflammatory biomarkers^[68].

Magnesium demonstrates a U-shaped association with cancer risk, where both low and high plasma concentrations are linked to an increased incident risk^[69]. Conversely, higher plasma magnesium is associated with a decreased risk of new-onset hyperuricemia in hypertensive adults^[70].

The association between iron intake and hypertension varies by its dietary form. Total and non-heme iron intake exhibit a U-shaped relationship, whereas heme iron (primarily from meat) shows an L-shaped (inverse) association with new-onset hypertension^[71]. This underscores the importance of distinguishing between iron sources in nutritional epidemiology.

The Guidelines elevate “dramatic reduction in highly processed foods” to the level of national health strategy and detail types of additives to avoid. This directly supports findings in Section 2.1.3: high intake of ultra-processed foods (UPFs) is associated with increased risk of CKD, with a more pronounced effect in individuals with diabetes^[15]. Together, they indicate that reducing exposure to industrial refined ingredients and additives is a foundational public health measure for chronic disease prevention.

The DGA recommends prioritizing a variety of high-quality protein foods and paying attention to cooking methods. This provides a direct and actionable translation for the findings in Section 2.2: “greater variety of protein sources is associated with lower risk of new-onset hypertension and diabetes” and “specific circulating amino acid profiles are closely associated with CKD.” Precision nutrition must not only determine total protein intake but also optimize its source composition.

The Guidelines recommend choosing oils rich in essential fatty acids (e.g., olive oil) and acknowledge the value of full-fat dairy (without added sugars) for specific groups (e.g., children). This aligns with evolving scientific understanding that fat quality is more critical than simple classification, and resonates with trends in Chinese research re-evaluating the role of full-fat dairy.

The DGA sets strict limits on added sugars (e.g., ≤ 10 grams per meal) and extensively lists their alternative names. This provides strong policy support and public education tools for the key finding in Section 2.1.2: “unsweetened coffee/tea is associated with protective effects against hypertension, AKI, and VTE, which disappear when sweeteners are added.” Health effects are highly dependent on the final consumed form of food.

The Guidelines explicitly state that older adults should prioritize nutrient-dense foods to address the paradox of declining calorie needs but stable or increased nutrient requirements. This is precisely the core issue addressed by the extensive U/J-shaped curve research in this review: how to precisely determine the optimal intake ranges for various trace elements (e.g., Zn, Se, Cu) and vitamins (e.g., vitamin D, B₁₂) in older adults, avoiding the dual risks of “deficiency” and “excess.” For example, the DGA’s emphasis on protein, vitamin D, and calcium for older adults, combined with this review’s elucidation of vitamin D’s role in dementia and CKD and its gene interactions, as well as the association between protein intake and health outcomes in aging, forms a complete evidence chain from policy recommendation to molecular mechanism.

While the Guidelines do not directly mention advanced technologies, their deep stratification based on individual differences, life stages, and disease status inherently demands more sophisticated and dynamic assessment and intervention tools. This creates an urgent application scenario and policy interface for the core mechanisms proposed in Chapter 3 (e.g., one-carbon metabolism, redox stress network) and the technological framework envisioned in this review [multi-omics phenotyping, artificial intelligence (AI)-driven dynamic optimization]. For instance, combining rapid, accessible metabolomic or trace element testing to assess levels of vitamin D active metabolites, zinc/selenium status, and MTHFR genotype in older adults can inform personalized supplementation strategies to prevent cognitive decline, sarcopenia, and osteoporosis, achieving the leap from “population advice” to “individual prescription.”

Provide a clear, authoritative, life-stage-stratified, and actionable dietary practice framework for the public and all healthcare providers. They focus on establishing a healthy dietary pattern as a universal foundation through policy guidance and food system reform, emphasizing the inclusive principle of “eating real food.”

Based on evidence from one of the world’s largest population cohorts, it provides a deep explanation of scientific principles (e.g., the one-carbon metabolism network, the Nrf2 pathway in oxidative stress, non-linear nutrient dynamics) and a forward-looking technological pathway for personalization (multi-omics + AI + IoT closed-loop systems). It reveals why personalization is needed and how it can be achieved through technology.

The integration of these two approaches outlines a clear future pathway: using the stratified, high-quality dietary patterns exemplified by the U.S. Guidelines as a universal, public health foundation for entire populations. Building upon this foundation, the precision assessment and intervention technologies developed within the Chinese framework can be applied to deliver “targeted intensification” and “dynamic adjustment” for high-risk individuals (e.g., those with specific genotypes, metabolic abnormalities) and special physiological groups (e.g., older adults, chronic disease patients, pregnant women). This marks the official transition of chronic disease nutrition intervention from the era of “universal recommendations” to a new era of “universal foundation + precision upgrade.”

In the context of China’s accelerating population aging and growing chronic disease burden, actively learning from the advanced concepts of precision and life-course orientation in the U.S. Dietary Guidelines - a universal foundation that, to achieve true precision, requires the technological enablement we detail in Section 5 - while fully leveraging China’s unique advantages in large-scale cohorts, multi-omics research, and digital technology application, is imperative to accelerate the construction and implementation of a precision nutrition system tailored to the Chinese population. This is not only an inevitable trend in scientific development but also an urgent need to improve national health and achieve the strategic goal of “healthy aging.” It will transform nutrition from a public health discipline based on population averages into a core technology for dynamic, precise health management based on individual biological characteristics and life course.

For instance, the MTHFR C677T genotype is a classic genetic marker determining folate metabolism efficiency^[75-77]. Vitamin D receptor (VDR) gene polymorphisms have been shown to modify the association between serum 25-hydroxyvitamin D levels and the risk of diseases such as dementia and venous thromboembolism^[48,51].

Provides real-time, dynamic feedback on metabolic functional status.

Includes comorbidity patterns (e.g., H-type hypertension, diabetic nephropathy), key organ function (e.g., estimated glomerular filtration rate, eGFR), detailed dietary records, geographical characteristics, and behavioral habits.

Leveraging unsupervised machine learning algorithms (e.g., clustering analysis) to integrate and analyze these vast, multi-dimensional datasets enables the automatic identification of population subgroups with similar metabolic characteristics and risk patterns. Examples include “the H-type hypertension subgroup with the MTHFR 677TT genotype, low folate, and mildly decreased renal function” or “the prediabetes subgroup characterized by vitamin D deficiency, insulin resistance, and specific gut microbiota features.” Such fine stratification is a prerequisite for the precise allocation of public health resources and the development of targeted group intervention strategies.

For individuals carrying the MTHFR 677TT genotype, direct supplementation with its metabolic end-product, 5-methyltetrahydrofolate (5-MTHF), can bypass the metabolic bottleneck caused by reduced enzyme activity, leading to higher bioavailability. Research confirms that 5-MTHF itself also possesses independent antioxidant properties and improves endothelium-dependent vasodilation^[78-80].

For the one-carbon metabolism network, developing scientifically formulated complexes containing folate, vitamin B₁₂, and vitamin D can synergistically support the smooth operation of the entire methyl donor cycle. For elements with established U-shaped relationships, such as zinc and copper, supplementation plans must be strictly based on baseline nutritional status testing, aiming to precisely adjust their levels to the individual’s “optimal physiological window,” rather than simply “supplementing if deficient.”

Even within the same genetic subgroup, the optimal supplementation dose may vary. Research suggests that for individuals with the MTHFR TT genotype, 0.8 mg/day of folic acid reduces tHcy, but 1.2 mg/day may offer additional benefit^[75]. In the future, AI systems based on pharmacokinetic/pharmacodynamic models will be able to integrate multiple individual parameters, such as genotype, baseline concentration, body weight, and hepatic/renal function, to predict and recommend the optimal starting dose and adjustment strategies.

Covering the complete chain of biomarkers from genetic susceptibility (genotyping), functional metabolites (multi-targeted metabolomics), and key nutritional status (vitamin D, B₁₂, trace elements) to end-organ damage markers (e.g., urine albumin-to-creatinine ratio).

Vigorously promoting the research, development, and popularization of Point-of-Care Testing (POCT) technologies suitable for primary care and home use. This enables convenient and rapid monitoring of core metabolic indicators such as blood glucose, lipids, and Hcy, forming a continuous, real-world individual health data stream.

Establishing and promoting standardized protocols and quality control systems for multi-omics detection technologies. This ensures consistency and comparability of data across different laboratories and platforms, which is fundamental for large-scale data aggregation and AI analysis.

Applying machine learning (e.g., XGBoost, Random Forest) and deep learning algorithms to mine data from million-scale population cohorts integrated with genomic, metabolomic, electronic health record, and long-term follow-up outcome data. This trains high-precision, individualized chronic disease risk prediction models. For example, a model could integrate an individual’s MTHFR genotype, current Hcy and vitamin D levels, ambulatory blood pressure, eGFR, and dietary records to precisely calculate their absolute risk of ischemic stroke over the next five years, parse the contribution of each risk factor, and thereby identify the highest-priority intervention targets.

Based on predicted risk and identified core issues, the AI system automatically generates structured, personalized nutritional advice (including type, dose, timing, and dietary adjustments). Integrating Generative Artificial Intelligence (e.g., Large Language Models) can transform complex medical nutritional recommendations into personalized, easy-to-understand health education materials and follow-up reminders with behavioral guidance, significantly enhancing patient comprehension and long-term adherence.

Utilizing 5G and Internet of Things (IoT) technologies, physiological and biochemical data collected from wearable devices (continuously monitoring blood pressure, heart rate, activity) and home-based smart POCT devices can be transmitted in real-time and securely to a cloud-based health management platform. AI models deployed on the cloud analyze this continuous data stream in real-time. If biomarkers deviate from the expected optimization trajectory (e.g., a sharp increase in urinary zinc excretion after zinc supplementation suggesting potential overdose risk), the system can immediately alert both the patient and the physician and automatically recommend adjustment plans. This enables a fully automated, intelligent management closed loop of “assessment-intervention-monitoring-feedback-re-optimization.”

Utilizing blockchain technology to build a decentralized, tamper-proof infrastructure for secure health data storage and sharing. Under the full protection of personal privacy and data sovereignty, this enables the secure and trusted flow of medical and health data across institutions and regions with patient authorization. This provides a high-quality data source for training more powerful national-level AI models and supporting public health decision-making, while also protecting data providers’ rights through smart contracts.

By targeting interventions to genetically or metabolically defined high-risk subgroups, public health resources are used more efficiently. An economic evaluation in China suggested that an MTHFR genotype-guided folate supplementation strategy for stroke prevention is cost-effective^[81]. This avoids wasting resources on blanket supplementation for those who derive little or no benefit.

AI-assisted decision support systems integrated into primary care electronic health records can empower general practitioners and community health workers to deliver sophisticated, evidence-based nutritional guidance, bridging the gap between specialized clinics and population-wide care.

By identifying and addressing the specific nutritional vulnerabilities of individuals based on objective data, precision nutrition can, paradoxically, promote greater health equity than generic advice, which may fail subpopulations with unique needs.

The upfront cost of multi-omics testing, while decreasing, remains a barrier to widespread adoption. A staged approach, starting with targeted genotyping and basic metabolomic panels for high-risk individuals, may be necessary.

Most precision nutrition strategies are supported by epidemiological evidence and short-term biomarker studies. Large-scale, long-term randomized controlled trials with “hard” clinical endpoints (e.g., stroke, myocardial infarction, mortality) are urgently needed to definitively prove that this approach improves outcomes.

The collection and use of genomic and deep phenotypic data raise profound ethical questions. Clear regulations, informed consent processes, and robust cybersecurity measures are non-negotiable. Who owns the data? How is it used? How are algorithmic biases prevented? These questions must be answered through interdisciplinary dialogue involving ethicists, lawyers, scientists, and patient advocates.

Success requires breaking down silos between nutritionists, physicians, bioinformaticians, data scientists, software engineers, and behavioral psychologists. New training programs and collaborative platforms are needed.