Developing a sustainable dietary pattern for Korean adults using quadratic programming

Abstract

BACKGROUND/OBJECTIVES

Dietary patterns play a critical role in mitigating environmental impacts while ensuring nutritional adequacy. Although several global models of sustainable diets have been proposed, evidence-based research reflecting Korea’s unique dietary culture remains limited. This study aimed to develop a sustainable dietary pattern for Korean adults that balances cultural acceptability, nutritional adequacy, and environmental sustainability.

SUBJECTS/METHODS

Quadratic programming was applied to derive an optimized dietary pattern minimizing the deviations from the observed dietary intakes while satisfying cultural acceptability, nutritional adequacy, and environmental sustainability constraints. The model parameters were derived from nationally representative 24-h dietary recall data from 3,983 Korean adults aged 19–64 yrs in the 2023 Korea National Health and Nutrition Examination Survey. The cultural acceptability constraints were defined within the intake-distribution bounds at the food group and food subgroup levels. The nutritional adequacy constraints were based on the 2025 Korean Dietary Reference Intakes. The environmental sustainability constraints were implemented by progressively tightening the allowable level of diet-related greenhouse gas (GHG) emissions. Elasticity analysis identified the final optimized pattern, hereafter termed the sustainable dietary pattern.

RESULTS

A sustainable dietary pattern was identified at a diet-related GHG emissions level of 2.19 kg CO₂eq/day, representing a 49% reduction from the current level (4.29 kg CO₂eq/day). Plant-based foods accounted for 68.5% of total intake by weight, animal-based foods for 18.3%, and sugars, seasonings, and fats for 13.2%. Vegetables (26.9%) and grains (23.5%) together comprised approximately half of the total food intake by weight, followed by fruits (12.9%) and animal-based protein foods (12.0%). Red meat intake decreased from 81.1 to 49.1 g/day, whereas combined fruit and vegetable intake increased from 395.9 to 500.0 g/day.

CONCLUSION

A sustainable dietary pattern satisfying the cultural acceptability, nutritional adequacy, and environmental sustainability constraints was identified. This pattern provides a scientific basis for developing Korean-style sustainable dietary guidelines and may inform future nutrition policies and food service practices.

INTRODUCTION

Climate change and escalating environmental pressures demand a sustainable transition across all aspects of the food system. The food system is recognized as a key sector influencing global greenhouse gas (GHG) emissions, land use, and biodiversity loss. In 2023, agrifood systems accounted for approximately 32% of global GHG emissions [1], suggesting that dietary choices play a crucial role in environmental sustainability.

Against this backdrop, in 2019, the EAT–Lancet Commission proposed the concept of the Planetary Health Diet (PHD), a dietary framework designed to promote human health while remaining within planetary boundaries [2]. Subsequently, in 2025, the Commission further expanded this framework through a follow-up report that explicitly incorporated equity alongside health and environmental sustainability [3]. Meanwhile, the Food and Agriculture Organization of the United Nations (FAO) and the World Health Organization (WHO) jointly articulated fundamental principles for Sustainable Healthy Diets (SHD) that holistically consider nutritional, environmental, and sociocultural factors, recommending their incorporation into national dietary policies and guidelines [4].

The World Wide Fund for Nature (WWF) has undertaken a series of initiatives to promote nutritionally adequate and environmentally sustainable diets [5,6,7]. The Livewell diet, originally developed using UK food consumption data, integrates nutritional adequacy and environmental considerations [5] and has subsequently been adapted to France, Spain, and Sweden to reflect country-specific dietary contexts [8].

Within the Korean context, a previous study examined the relationship between food consumption and environmental impacts by estimating diet-related GHG emissions among Koreans using the dietary data from the Korea National Health and Nutrition Examination Survey (KNHANES). That study identified meat as the largest contributor to diet-related emissions and highlighted the substantial contribution of animal-based foods to diet-related GHG emissions in Korean diets. The study also developed a GHG emission factor database tailored to the Korean food system [9].

Meanwhile, dietary patterns among Korean adults have continued to change. Chun et al. [10] analyzed the KNHANES data from Korean adults aged ≥ 19 yrs to examine changes in dietary intake from 2010 to 2020 and reported that mean daily energy intake decreased from 2,102 to 1,902 kcal/day, while the percentage of energy intake from carbohydrates decreased from 65.0% to 59.7%. In contrast, the percentages of energy intake from protein and fat increased from 15.2% to 15.9% and from 19.7% to 24.3%, respectively, suggesting a shift toward higher-protein and higher-fat dietary patterns with reduced carbohydrate intake. These dietary shifts may have implications for public health and environmental sustainability.

Recent Korean studies have highlighted the multidimensional nature of sustainable diets and the gaps between current dietary patterns and PHD recommendations. A scoping review proposed the development of a multidimensional sustainable diet index tailored to the Korean context, incorporating nutritional, environmental, economic, and sociocultural dimensions [11]. An empirical study of Korean adults evaluated adherence to the PHD in relation to Korean Healthy Eating Index scores and found that intakes of whole grains, legumes, and nuts were substantially below the recommended levels in both men and women. In contrast, red meat intake exceeded the recommended range in both groups [12]. Taken together with previous work on nutrient intake patterns, dietary change trends, and diet-related GHG emissions [9,10], these studies indicate that sustainable diets in Korea involve multiple dimensions and that meaningful gaps remain between observed Korean diets and PHD food-group recommendations. Nevertheless, few studies in Korea have applied optimization approaches to derive a single dietary pattern that simultaneously satisfies constraints for cultural acceptability, nutritional adequacy, and environmental sustainability, rather than assessing these dimensions separately.

This study aimed to develop a sustainable dietary pattern that balances nutritional adequacy and environmental sustainability while maintaining cultural acceptability. The findings provide scientific evidence to support the development of feasible dietary guidance that respects Korean dietary traditions and may inform public health policies, community nutrition programs, and food service settings.

SUBJECTS AND METHODS

Data preparation for quadratic programming

Dietary intake data

Dietary intake data were obtained from the second year of the ninth KNHANES, conducted in 2023. KNHANES is a nationally representative survey of the Korean population that uses a stratified, multistage probability sampling design [13].

Of the 6,929 participants in the 2023 KNHANES, individuals younger than 19 yrs or older than 64 yrs and those who did not complete the nutrition survey, including the single 24-h dietary recall, were excluded. As a result, 3,983 adults aged 19–64 yrs were included in the final analysis. The 2023 KNHANES was conducted with approval from the Institutional Review Board (IRB) of the Korea Disease Control and Prevention Agency (IRB No. 2022-11-16-R-A). Additional IRB approval was waived because this study involved the secondary analysis of publicly available, anonymized data.

Food classification system

A hierarchical food classification system reflecting Korean dietary characteristics was established as the basis for developing an optimized dietary pattern. This classification system was informed by international dietary frameworks [2,3,5,6,7,14] and national references, including the Korean Food Balance Wheel [15] and a previous Korean dietary optimization study [16].

The classification system comprised 3 levels: 3 major food groups (plant-based foods, animal-based foods, and sugars, seasonings, and fats), 9 food groups, and 24 food subgroups (Table 1).

Table 1. Classification of major food groups, food groups, and food subgroups.

Major food group	Food group	Food subgroup	Number of representative foods included	Examples of representative foods
Plant-based foods	Grains	Whole grains & mixed grains	9	Oats, brown rice
		Refined grains	10	Ramen, wheat flour
		Potatoes & starches	7	Potatoes, sweet potatoes
	Vegetables	Fresh vegetables	22	Carrots, cabbage
		Mushrooms	6	Oyster mushrooms, pine mushrooms
		Seaweeds	8	Kelp, seaweed
		Fermented & pickled vegetables	13	Kkakdugi, cabbage kimchi
	Fruits	Fresh fruits	15	Pears, apples
		Canned fruits	1	Canned fruits
	Plant-based protein foods	Legumes	5	Soybeans, tofu
		Nuts	13	Peanuts, almonds
Animal-based foods	Animal-based protein foods	Red meat	5	Pork, beef
		Poultry meat	3	Chicken, duck
		Processed meat	3	Processed meat, processed pork products
		Unprocessed seafood	29	Horse mackerel, mackerel
		Processed seafood	5	Fish cakes, fish sausage
		Eggs	2	Eggs, quail eggs
	Dairy products	Dairy products	4	Milk, yogurt
Sugars, seasonings, and fats	Sugars	Sweetened dairy products	5	Flavored milk, ice cream
		Sweetened beverages	6	Other sweetened beverages, carbonated drinks
		Sugar products	12	Honey, candies
	Seasonings	Seasonings	25	Soy sauce, red pepper paste
	Fats	Plant-based fats	8	Sesame oil, soybean oil
		Animal-based fats	3	Lard, butter

Specific food items were further categorized into subgroups based on their main ingredients, nutritional characteristics, and processing status, using the KNHANES food codes and detailed food descriptions. Grain products were classified as whole and mixed grains or refined grains according to the grain type and processing information recorded in the food descriptions. Vegetables were categorized as fresh vegetables, mushrooms, seaweeds, and fermented or pickled vegetables (e.g., kimchi and jangajji). Fermented or pickled vegetables were treated as a distinct subgroup because of their prominent role in the Korean diet and contribution to sodium intake. Fruits identified as canned were assigned to the canned fruits subgroup to account for the differences in sugar content between canned and fresh fruits.

Within the animal-based protein food group, meat items were classified into red meat, poultry meat, and processed meat to reflect the differences in GHG emissions across meat types. Seafood items were categorized into unprocessed and processed subgroups. Dairy items, such as milk and yogurt, were included in the dairy products group, whereas flavored milk and ice cream were assigned to the sweetened dairy products subgroup within the sugars group based on their sugar content.

Alcoholic beverages and unsweetened beverages were excluded from the optimization model. Alcoholic beverages were excluded because they are not considered essential for nutritionally adequate dietary patterns and are not included in the Korean Food Balance Wheel [15]. Unsweetened beverages were excluded because they contribute negligible amounts of energy and nutrients.

Representative foods were selected for each of the 24 food subgroups among the 2,068 foods reported in the 2023 KNHANES. Within each food subgroup, foods contributing ≥ 1% of the total daily average intake by weight of that subgroup were designated as representative foods. Intake amounts for items contributing less than 1% were proportionally redistributed among the representative foods within the same subgroup so that the total observed intake of each subgroup was preserved. A total of 219 representative foods were identified through this procedure.

Current dietary intake estimates

The weighted mean intakes of the representative foods, the 10th and 90th percentiles of food group intakes, and the 90th percentiles of food subgroup intakes were estimated using IBM SPSS Statistics (version 31.0; IBM Corp., Armonk, NY, USA). The complex sampling design of the KNHANES, including the sampling weights, strata, and clusters, was incorporated to obtain nationally representative estimates. The weighted mean intakes were used as the observed intake values in the objective function, whereas the percentile estimates were used to define the cultural acceptability constraints.

Nutrient profiles and GHG emission factors

The nutrient profiles were derived for each of the 219 representative foods identified above using the nutrient composition data included in the integrated KNHANES dataset. Each nutrient profile comprised energy content and values for 17 nutrients—including carbohydrates, protein, total fat, saturated fatty acids, total sugars, dietary fiber, sodium, calcium, phosphorus, potassium, iron, vitamin A, vitamin C, thiamine, riboflavin, niacin, and cholesterol—all standardized per 100 g. For use in the optimization model, these values were converted to a per-gram basis and applied at the representative-food level.

The GHG emission data for the 219 representative foods were compiled using the life cycle assessment (LCA)-based GHG emission factors. The data sources were applied sequentially to maximize coverage while prioritizing Korea-specific or directly matched factors. When direct matches were unavailable, GHG emission factors from nutritionally and taxonomically similar foods within the same food subgroup were substituted according to the same hierarchical order.

The GHG emission factors were first obtained from the database developed by Tan and Shin [17], which integrated data from South Korean environmental reports and international databases. This source was applied to 150 representative foods (68.5%). For foods not covered by this primary source, the median values from the global meta-analysis by Poore and Nemecek [18] were applied to 25 foods (11.4%), followed by the median values from Clune et al. [19] for 30 foods (13.7%).

For 14 representative foods (6.4%), all belonging to the seasoning subgroup, no direct or comparable emission factors were available from the above sources. Therefore, the median GHG emission factors from the Korea-specific database developed by Hong and Kim [9] were applied.

Optimization using quadratic programming

Decision variables and objective function

Dietary optimization was performed using quadratic programming (QP) implemented in Python with the CVXPY package. This mathematical modeling technique was used to derive an optimized dietary pattern that minimizes the deviation from the current intake pattern to support cultural acceptability while satisfying predefined nutritional adequacy and environmental sustainability constraints. Table 2 lists the final quadratic programming model established in this study.

Table 2. Quadratic model applied in the study.

Decision variables			xi (i = 1, 2, …, 219)
Objective function
Constraints
	Food intake constraints
		Food group b
		Food subgroup j
	Energy constraint
			Etarget−100 ≤ E ≤ Etarget+100
	Energy/nutrient constraints
		Carbohydrate
		Protein
		Fat
		Saturated fatty acids
		Total sugars
	Other nutrient constraints
	Food-based constraint
	Environmental sustainability constraint
			GHG ≤ ε
	Non-negativity constraint		xi ≥ 0 (∀ i)

i = 1, 2, ..., 219: Representative food.

b = 1, 2, ..., 9: Food group.

j = 1, 2, ..., 24: Food subgroup.

k = 1, 2, ..., 12: Other nutrient.

x_i: Optimal daily intake (g/day) for representative food i.

B_b: Set of foods included in food group b.

S_j: Set of foods included in food subgroup j.

V: Set of representative foods belonging to the vegetable category.

F: Set of representative foods belonging to the fruit category.

: Average daily intake (g/day) of representative food i.

E: Total energy intake (kcal/day).

E_i: Energy content per gram of representative food i (kcal/g).

E^target: Target energy intake (kcal/day).

CHO_i, PRO_i, FAT_i, SFA_i, SUGAR_i: Carbohydrate, protein, fat, saturated fatty acids, and total sugar content per gram of representative food item i.

N_ik: Nutrient k content per gram of representative food i (g/g, mg/g or μg/g per nutrient).

L_k, U_k: Lower and upper bounds for other nutrient k.

P10_b, P90_b: 10th and 90th percentiles of intake for food group b.

P90_j: 90th percentile intake of food subgroup j.

GHG: Total daily greenhouse gas emissions (kg CO₂eq/day).

GHG_i: Greenhouse gas emission factor per gram of food i (kg CO₂eq/g).

ε: Stepwise-decreasing upper bound for total daily diet-related GHG emissions (kg CO₂eq/day).

The daily optimized intake x_i (g/day) of representative food i was defined as the decision variable to derive a food intake pattern that satisfies the nutritional adequacy constraints while minimizing divergence from the current dietary pattern. The observed weighted mean intake was used as the reference intake for each food item (i = 1, 2, …, 219). The objective function was defined as minimizing the sum of the squared differences between the optimal intake x_i for each food item and the mean intake . The objective function Y is expressed using the following formula.

$$\text{Minimize Y}=\sum _{i=1}^{219}{(x_i-x_i^{obs})}^2$$

The objective function minimizes the squared absolute deviations in intake (g/day), rather than the relative deviations from observed intake. This approach is intended to prevent the optimized dietary pattern from deviating excessively from current dietary habits and to encourage realistic dietary transitions that consider cultural acceptability.

Cultural acceptability constraints

Cultural acceptability constraints were established to ensure the practical feasibility of the optimized dietary pattern. Following the approach of previous studies [16,20], these constraints were based on actual intake distributions to prevent the optimized dietary pattern from deviating excessively from current practices. The constraints were applied at 2 levels: the food group and the food subgroup. At the food group level, the 10th and 90th percentiles of the current intake were designated as the lower and upper bounds, respectively, ensuring that the total intake for each group remained within the range of the observed intake distribution. At the food subgroup level, no lower bound was set to account for the possibility of non-consumption of certain food subgroups. Instead, only an upper bound of the 90th percentile was applied. Table 3 lists the specific constraints for the food groups and food subgroups.

Table 3. Cultural acceptability constraints: food group and food subgroup intake bounds derived from the observed intake distribution.

Food group	Food subgroup	10th percentile intake (g)	90th percentile intake (g)
Grains		116.0	499.3
	Whole grains & mixed grains	-	64.4
	Refined grains	-	416.6
	Potatoes & starches	-	100.8
Vegetables		67.5	525.3
	Fresh vegetables	-	354.9
	Mushrooms	-	14.9
	Seaweeds	-	8.3
	Fermented & pickled vegetables	-	233.6
Fruits		0.0	372.0
	Fresh fruits	-	371.4
	Canned fruits	-	0.0
Plant-based protein foods		0.0	167.1
	Legumes	-	158.7
	Nuts	-	11.2
Animal-based protein foods		47.7	415.5
	Red meat	-	213.8
	Poultry meat	-	112.2
	Processed meat	-	43.2
	Unprocessed seafood	-	104.6
	Processed seafood	-	26.8
	Eggs	-	108.3
Dairy products		0.0	274.6
	Dairy products	-	274.6
Sugars		0.0	386.3
	Sweetened beverages	-	371.7
	Sweetened dairy products	-	32.7
	Sugar products	-	24.8
Seasonings		7.4	87.9
	Seasonings	-	87.9
Fats		0.2	18.4
	Plant-based fats	-	16.9
	Animal-based fats	-	0.5

Values are 10th (P10) and 90th (P90) percentiles of daily intake estimated using the KNHANES complex sampling design. At the food-group level, P10 and P90 were applied as the lower and upper bounds, respectively; at the food-subgroup level, only P90 was applied, and “–” indicates no lower bound. Percentiles were estimated separately at each level; therefore, subgroup P90 values need not sum to the corresponding food-group P90.

KNHANES, Korea National Health and Nutrition Examination Survey.

Nutritional adequacy constraints

The target energy intake was set to the weighted mean energy intake of the study population (1,914.7 kcal/day), with a tolerance range of ± 100 kcal applied to maintain an approximately isocaloric condition. This ensured that changes in diet composition resulted from alterations in food selection rather than changes in total energy intake. Within this framework, the energy contribution percentages (%E) from carbohydrates, protein, and fat were each set within the recommended ranges specified by the 2025 Korean Dietary Reference Intakes (KDRIs) (carbohydrates: 50–65%, protein: 10–20%, fat: 15–30%) [21]. The intake of saturated fatty acids was constrained to less than 7% of total energy intake, while the total sugar intake was limited to no more than 20% of total energy intake. The constraints were designed to ensure qualitative dietary improvements while maintaining energy balance.

For nutrients with an established Recommended Nutrient Intake (RNI) or Adequate Intake (AI), the corresponding RNI or AI was applied as the lower-bound constraint. Where a Tolerable Upper Intake Level (UL) was available, the UL was applied as the upper-bound constraint. When the RNI or AI differed by sex and age group, the median value across the relevant sex- and age-specific reference values was used as the lower bound to establish a single population-level constraint for Korean adults aged 19–64 yrs.

For iron, however, the median of the age-specific RNIs for women aged 19–64 yrs (12 mg/day) was applied because iron requirements differ substantially by sex. Inadequate iron intake is an important nutritional concern among women. Furthermore, the upper limits for sodium and cholesterol were set at 2,300 and 300 mg, respectively, with the sodium limit based on the Chronic Disease Risk Reduction intake (CDRR) and the cholesterol limit based on the recommended intake level. At the food group level, the lower bound for combined fruits and vegetables was set at 500 g, reflecting the Health Plan 2030 (HP2030) recommendation for preventing and managing chronic diseases [22]. Table 4 lists the specific levels for the nutritional adequacy constraints.

Table 4. Nutritional adequacy constraints applied in the optimization model.

Nutritional indicator	Lower bound	Upper bound	Basis for constraint
Total energy (kcal)	Etarget − 100	Etarget + 100	Weighted mean energy intake of the study population (1,914.7 kcal/day)
Carbohydrate (%E)	≥ 50	≤ 65	AMDR: 50–65%
Protein (%E)	≥ 10	≤ 20	AMDR: 10–20%
Total fat (%E)	≥ 15	≤ 30	AMDR: 15–30%
Saturated fatty acids (%E)	-	< 7	Recommended intake: < 7% of energy
Total sugars (%E)	-	≤ 20	Recommended intake: ≤ 20% of energy
Dietary fiber (g)	≥ 25	-	AI: men 30 g, women 20–25 g
Sodium (mg)	-	≤ 2,300	CDRR: 2,300 mg
Calcium (mg)	≥ 725	≤ 2,000	RNI: men 800 mg, women 650–750 mg; UL: 2,000 mg
Phosphorus (mg)	≥ 650	≤ 3,500	RNI: 650 mg; UL: 3,500 mg
Potassium (mg)	≥ 3,500	-	AI: 3,500 mg
Iron (mg)	≥ 12	≤ 45	RNI: men 8 mg, women 7–12 mg; UL: 45 mg
Vitamin A (µg RAE)	≥ 700	≤ 3,000	RNI: men 800 µg RAE, women 600 µg RAE; UL: 3,000 µg RAE
Vitamin C (mg)	≥ 100	≤ 2,000	RNI: 100 mg; UL: 2,000 mg
Thiamine (mg)	≥ 1.15	-	RNI: men 1.2 mg, women 1.1 mg
Riboflavin (mg)	≥ 1.35	-	RNI: men 1.5 mg, women 1.2 mg
Niacin (mg NE)	≥ 13.5	≤ 35	RNI: men 14 mg NE, women 13 mg NE; UL: 35 mg NE
Cholesterol (mg)	-	≤ 300	Recommended intake: < 300 mg
Fruits & vegetables (g)	≥ 500	-	Health Plan 2030 target for chronic disease prevention

Unless otherwise specified, the constraints were based on the 2025 Korean Dietary Reference Intakes [21]. The combined fruit and vegetable intake constraint was based on the Health Plan 2030 target [22].

AMDR, acceptable macronutrient distribution range; AI, Adequate Intake; CDRR, Chronic Disease Risk Reduction intake; RNI, Recommended Nutrient Intake; UL, Tolerable Upper Intake Level; RAE, retinol activity equivalents; NE, niacin equivalents.

Environmental sustainability constraint

An ε-constraint approach was employed to systematically reduce diet-related GHG emissions. In this approach, the upper limit for total daily diet-related GHG emissions was treated as a variable bound (ε), whereas the nutritional adequacy and cultural acceptability constraints remained fixed. First, the minimum possible GHG emissions achievable under these fixed constraints were calculated to establish the feasible range for emission reductions. The current diet-related GHG emission level was then used as the initial ε value, and the emission cap was progressively reduced by 0.1 kg CO₂eq/day at each step within the feasible range. At each step, the model minimized the objective function while satisfying the corresponding GHG emission cap and all fixed nutritional and cultural constraints. This stepwise procedure generated a series of feasible dietary patterns across progressively lower GHG emission targets, allowing changes in dietary deviation to be evaluated as the emission cap was tightened. This stepwise approach was structured by referencing the methodology used in the WWF Livewell model [6,7].

Selection and evaluation of the optimized dietary pattern

Following the stepwise approach described in the Sustainable Diets for Central and Eastern Europe model [23], penalty and elasticity indicators were calculated to evaluate the trade-off between dietary deviation and additional reductions in diet-related GHG emissions. The penalty represents the extent to which the optimal solution deviates from the current dietary pattern. Elasticity (E) is defined as the ratio of the relative increase in the penalty to the relative reduction in diet-related GHG emissions between consecutive optimization stages, as follows:

$$E_t=\frac{\frac{Penalt{y}_t-Penalt{y}_{t-1}}{Penalt{y}_{t-1}}}{\frac{GH{G}_{t-1}-GH{G}_t}{GH{G}_{t-1}}}$$

where Penalty_t and GHG_t represent the penalty value and diet-related GHG emissions at optimization stage t, respectively.

An elasticity value of 1.0 indicates that the relative increase in the penalty is equal to the relative reduction in diet-related GHG emissions between consecutive optimization stages. Following the approach of the Sustainable Diets for Central and Eastern Europe model [23], this point was considered to represent a balanced trade-off between further reductions in diet-related GHG emissions and increasing deviation from the observed dietary pattern. Accordingly, the emission cap was progressively reduced until the elasticity first exceeded 1.0, and the feasible solution from the immediately preceding stage was selected as the optimized dietary pattern.

The total energy and nutrient contents of the optimized dietary pattern were calculated using the procedure established by Asano et al. [16]. Specifically, the optimal daily intake of each representative food was multiplied by its corresponding energy and nutrient values per 100 g and divided by 100. The diet-related GHG emissions were calculated using the same approach based on the GHG emission factors compiled in this study. All values were summed to determine the final daily nutritional and environmental impacts of the optimized dietary pattern.

RESULTS

Optimization process and selection of the sustainable dietary pattern

With the nutritional adequacy and cultural acceptability constraints fixed, the diet-related GHG emission cap was progressively lowered from the current level of 4.29 kg CO₂eq/day in increments of 0.1 kg CO₂eq/day. Across the evaluated feasible solutions, the penalty increased monotonically, with larger successive increases as the emission cap was lowered.

The elasticity remained below 1.0 through the solution at 2.19 kg CO₂eq/day (E = 0.90). When the emission cap was lowered to 2.09 kg CO₂eq/day, the elasticity exceeded 1.0 for the first time (E = 1.03), indicating that the proportional increase in the penalty exceeded the proportional reduction in GHG emissions. Accordingly, the solution at 2.19 kg CO₂eq/day was selected as the optimized dietary pattern because it was the last feasible solution before elasticity exceeded 1.0. This final pattern is hereafter referred to as the sustainable dietary pattern.

Food composition of the sustainable dietary pattern

The sustainable dietary pattern was characterized by a predominantly plant-based dietary composition while preserving the overall structure of the Korean diet. Overall, plant-based foods accounted for 68.5% of the total intake, whereas animal-based foods and sugars, seasonings, and fats accounted for 18.3% and 13.2%, respectively. Fig. 1 summarizes the food composition of the sustainable dietary pattern across food groups and selected food subgroups. Vegetables (26.9%) and grains (23.5%) together accounted for approximately half of the total food intake, followed by fruits (12.9%), animal-based protein foods (12.0%), sugars (9.5%), dairy products (6.4%), and plant-based protein foods (5.2%).

Fig. 1. Distribution of food groups and selected food subgroups in the sustainable dietary pattern for Korean adults.

The values indicate the percentage contribution to the total daily food intake by weight; the expanded segments show selected food subgroups within grains, vegetables, and animal-based protein foods.

Table 5 presents detailed changes in food intake between the current and sustainable dietary patterns across major food groups, food groups, and food subgroups. Compared with the current dietary pattern, the proportion of plant-based foods increased from 61.2% (734.8 g) to 68.5% (860.6 g), whereas animal-based foods decreased from 24.3% (291.7 g) to 18.3% (230.6 g). The share of sugars, seasonings, and fats also declined slightly from 14.5% (173.9 g) to 13.2% (166.2 g).

Table 5. Comparison of food intake between the current and sustainable dietary patterns among Korean adults (aged 19–64 yrs, per capita per day).

Major food group	Food group	Food subgroup	Food intake, g (% of total)		Contribution within food group (%)		Contribution within major food group (%)		Major food group intake, g (% of total)
			Current	Sustainable	Current	Sustainable	Current	Sustainable	Current	Sustainable
Plant-based foods	Grains	Whole grains & mixed grains	24.8 (2.1)	36.2 (2.9)	8.5	12.2	39.7	34.3	734.8 (61.2)	860.6 (68.5)
		Refined grains	231.2 (19.3)	225.4 (17.9)	79.3	76.3
		Potatoes & starches	35.5 (2.9)	34.0 (2.7)	12.2	11.5
		Subtotal	291.5 (24.3)	295.6 (23.5)	100.0	100.0
	Vegetables	Fresh vegetables	168.2 (14.0)	210.4 (16.7)	60.4	62.3	37.9	39.3
		Mushrooms	6.4 (0.5)	3.7 (0.3)	2.3	1.1
		Seaweeds	3.7 (0.3)	8.3 (0.7)	1.3	2.4
		Fermented & pickled vegetables	100.3 (8.4)	115.5 (9.2)	36.0	34.2
		Subtotal	278.6 (23.2)	337.9 (26.9)	100.0	100.0
	Fruits	Fresh fruits	116.6 (9.7)	162.1 (12.9)	99.4	100.0	16.0	18.8
		Canned fruits	0.7 (0.1)	0.0 (0.0)	0.6	0.0
		Subtotal	117.3 (9.8)	162.1 (12.9)	100.0	100.0
	Plant-based protein foods	Legumes	42.8 (3.5)	53.8 (4.3)	90.3	82.8	6.4	7.6
		Nuts	4.6 (0.4)	11.2 (0.9)	9.7	17.2
		Subtotal	47.4 (3.9)	65.0 (5.2)	100.0	100.0
Animal-based foods	Animal-based protein foods	Red meat	81.1 (6.7)	49.1 (3.9)	38.3	32.6	72.5	65.3	291.7 (24.3)	230.6 (18.3)
		Poultry meat	32.7 (2.7)	24.6 (2.0)	15.5	16.3
		Processed meat	13.9 (1.2)	7.3 (0.6)	6.6	4.8
		Unprocessed seafood	34.5 (2.9)	29.9 (2.4)	16.3	19.9
		Processed seafood	10.2 (0.8)	4.2 (0.3)	4.8	2.8
		Eggs	39.2 (3.3)	35.5 (2.8)	18.5	23.6
		Subtotal	211.6 (17.6)	150.6 (12.0)	100.0	100.0
	Dairy products	Dairy products	80.1 (6.7)	80.0 (6.4)	100.0	100.0	27.5	34.7
		Subtotal	80.1 (6.7)	80.0 (6.4)	100.0	100.0
Sugars, seasonings, and fats	Sugars	Sweetened beverages	96.4 (8.0)	96.1 (7.6)	78.2	80.6	70.9	71.7	173.9 (14.5)	166.2 (13.2)
		Sweetened dairy products	17.2 (1.5)	11.9 (0.9)	13.9	10.0
		Sugar products	9.7 (0.8)	11.2 (0.9)	7.9	9.4
		Subtotal	123.3 (10.3)	119.2 (9.5)	100.0	100.0
	Seasonings	Seasonings	42.8 (3.5)	35.4 (2.8)	100.0	100.0	24.6	21.3
		Subtotal	42.8 (3.5)	35.4 (2.8)	100.0	100.0
	Fats	Plant-based fats	7.3 (0.6)	11.4 (0.9)	93.7	98.3	4.5	7.0
		Animal-based fats	0.5 (0.04)	0.2 (0.02)	6.3	1.7
		Subtotal	7.8 (0.7)	11.6 (0.9)	100.0	100.0

Values are mean daily intakes (g/day). Percentages in parentheses indicate the share of total food intake. Totals may not sum exactly due to rounding.

Within the plant-based foods group, the share of grains remained relatively stable, changing slightly from 24.3% to 23.5% of the total intake. However, its composition shifted, with refined grains decreasing from 231.2 g (19.3%) to 225.4 g (17.9%) and whole and mixed grains increasing from 24.8 g (2.1%) to 36.2 g (2.9%). Intake of vegetables increased from 278.6 g (23.2%) to 337.9 g (26.9%), and intake of fruits increased from 117.3 g (9.8%) to 162.1 g (12.9%), entirely through fresh fruits. In addition, plant-based protein foods, including both legumes and nuts, increased from 47.4 g (3.9%) to 65.0 g (5.2%).

Within the animal-based foods group, intake of animal-based protein foods decreased from 211.6 g (17.6%) to 150.6 g (12.0%). Red meat intake declined from 81.1 g (6.7%) to 49.1 g (3.9%), and processed meat was reduced from 13.9 g (1.2%) to 7.3 g (0.6%). Although the absolute intake of unprocessed seafood and eggs showed slight decreases, their relative shares within the animal-based protein food group increased. Dairy product intake remained essentially unchanged, at 80 g, while its share of total intake decreased slightly from 6.7% to 6.4%.

Within the sugars, seasonings, and fats group, intake from the sugars food group decreased slightly from 123.3 g (10.3%) to 119.2 g (9.5%), driven mainly by a decrease in sweetened dairy products (17.2 to 11.9 g), while sweetened beverages remained largely unchanged (96.4 to 96.1 g). Plant-based fats accounted for 98.3% of intake within the fats food group in the sustainable dietary pattern.

Nutritional adequacy of the sustainable dietary pattern

The sustainable dietary pattern satisfied all predefined nutritional adequacy constraints and corrected several nutrient inadequacies or excesses observed in the current dietary pattern. Table 6 compares the nutritional profiles of the current and sustainable dietary patterns. Compared with the current dietary pattern, the proportion of energy derived from carbohydrates increased from 57.7% to 60.4%, whereas the proportions derived from protein and total fat decreased from 16.1% to 14.6% and from 26.2% to 25.4%, respectively. All 3 values remained within the recommended ranges of the 2025 KDRIs. Saturated fatty acid intake decreased from 7.3% to 7.0% of total energy. Total sugars contributed 12.9% and 13.0% of total energy in the current and sustainable dietary patterns, respectively, and both values remained below the predefined upper limit of 20%.

Table 6. Nutritional profiles of the current and sustainable dietary patterns.

Nutritional indicator	Current pattern	Sustainable pattern
Total energy (kcal)	1,914.7	1,958.9
Carbohydrate (%E)	57.7	60.4
Protein (%E)	16.1	14.6
Total fat (%E)	26.2	25.4
Saturated fatty acids (%E)	7.3	7.0
Total sugars (%E)	12.9	13.0
Dietary fiber (g)	23.6	34.6
Sodium (mg)	3,332.7	2,300.0
Calcium (mg)	505.6	750.0
Phosphorus (mg)	1,067.5	1,180.0
Potassium (mg)	2,687.5	3,500.0
Iron (mg)	9.6	12.0
Vitamin A (µg RAE)	399.2	700.0
Vitamin C (mg)	69.2	100.0
Thiamine (mg)	1.2	1.6
Riboflavin (mg)	1.7	2.3
Niacin (mg NE)	12.9	15.4
Cholesterol (mg)	283.5	207.9
Fruits & vegetables (g)	395.9	500.0

Values are rounded; the unrounded optimized value satisfied the predefined constraint. The percentages of energy from carbohydrate, protein, and fat were calculated using 4, 4, and 9 kcal/g, respectively. The 3 macronutrient percentages may not sum exactly to 100% because total energy was calculated from the food-specific energy values in the KNHANES nutrient composition data.

RAE, retinol activity equivalents; NE, niacin equivalents; KNHANES, Korea National Health and Nutrition Examination Survey.

Dietary fiber intake increased substantially from 23.6 g/day to 34.6 g/day. Intakes of several micronutrients also increased, including calcium (505.6 to 750.0 mg/day), potassium (2,687.5 to 3,500.0 mg/day), iron (9.6 to 12.0 mg/day), vitamin A (399.2 to 700.0 μg RAE/day), vitamin C (69.2 to 100.0 mg/day), thiamine (1.2 to 1.6 mg/day), riboflavin (1.7 to 2.3 mg/day), and niacin (12.9 to 15.4 mg NE/day). In contrast, sodium intake decreased from 3,332.7 mg/day to the CDRR level of 2,300.0 mg/day, and cholesterol intake decreased from 283.5 mg/day to 207.9 mg/day. The combined intake of fruits and vegetables increased from 395.9 g/day to 500.0 g/day, meeting the HP2030 target.

Environmental sustainability of the sustainable dietary pattern

The sustainable dietary pattern reduced total daily diet-related GHG emissions by approximately 49%, from 4.29 kg CO₂eq/day in the current dietary pattern to 2.19 kg CO₂eq/day. Table 7 compares diet-related GHG emissions between the current and sustainable dietary patterns across major food groups, food groups, and food subgroups. Emissions from animal-based foods decreased from 2.66 to 0.68 kg CO₂eq/day, and their contribution to total diet-related GHG emissions declined from 62.0% to 31.0%. In particular, emissions from red meat, which accounted for the largest share of current diet-related GHG emissions, decreased from 1.89 to 0.21 kg CO₂eq/day, corresponding to a decline in its contribution from 44.1% to 9.6%. Because the optimization was conducted at the representative-food level, the larger percentage reduction in GHG emissions from red meat (88.9%) than in red meat intake by weight (39.5%) reflects changes in the relative amounts of representative foods within the subgroup and variation in their food-specific GHG emission factors.

Table 7. Comparison of diet-related GHG emissions between the current and sustainable dietary patterns among Korean adults (aged 19–64 yrs, per capita per day).

Major food group	Food group	Food subgroup	GHG emissions, kg CO2eq (% of total)		Contribution within food group (%)		Contribution within major food group (%)		Major food group GHG emissions, kg CO2eq (% of total)
			Current	Sustainable	Current	Sustainable	Current	Sustainable	Current	Sustainable
Plant-based foods	Grains	Whole grains & mixed grains	0.09 (2.1)	0.04 (1.8)	10.1	5.0	67.6	63.2	1.31 (30.5)	1.25 (57.1)
		Refined grains	0.78 (18.2)	0.74 (33.8)	87.6	93.7
		Potatoes & starches	0.02 (0.4)	0.01 (0.5)	2.3	1.3
		Subtotal	0.89 (20.7)	0.79 (36.1)	100.0	100.0
	Vegetables	Fresh vegetables	0.16 (3.7)	0.16 (7.3)	48.5	48.5	25.1	26.4
		Mushrooms	0.03 (0.7)	0.02 (0.9)	9.1	6.0
		Seaweeds	0.01 (0.2)	< 0.001 (0.0)	3.0	0.0
		Fermented & pickled vegetables	0.13 (3.0)	0.15 (6.9)	39.4	45.5
		Subtotal	0.33 (7.7)	0.33 (15.1)	100.0	100.0
	Fruits	Fresh fruits	0.06 (1.4)	0.09 (4.1)	98.4	100.0	4.6	7.2
		Canned fruits	0.001 (0.0)	0.00 (0.0)	1.6	0.0
		Subtotal	0.06 (1.4)	0.09 (4.1)	100.0	100.0
	Plant-based protein foods	Legumes	0.03 (0.6)	0.03 (1.4)	85.7	75.0	2.7	3.2
		Nuts	0.005 (0.1)	0.01 (0.4)	14.3	25.0
		Subtotal	0.03 (0.7)	0.04 (1.8)	100.0	100.0
Animal-based foods	Animal-based protein foods	Red meat	1.89 (44.1)	0.21 (9.6)	74.4	37.5	95.5	82.4	2.66 (62.0)	0.68 (31.0)
		Poultry meat	0.25 (5.8)	0.18 (8.2)	9.9	32.1
		Processed meat	0.17 (4.0)	0.02 (0.9)	6.7	3.6
		Unprocessed seafood	0.13 (3.0)	0.07 (3.2)	5.1	12.5
		Processed seafood	0.02 (0.5)	0.01 (0.5)	0.8	1.8
		Eggs	0.08 (1.9)	0.07 (3.2)	3.1	12.5
		Subtotal	2.54 (59.2)	0.56 (25.6)	100.0	100.0
	Dairy products	Dairy products	0.12 (2.8)	0.12 (5.5)	100.0	100.0	4.5	17.6
		Subtotal	0.12 (2.8)	0.12 (5.5)	100.0	100.0
Sugars, seasonings, and fats	Sugars	Sweetened beverages	0.18 (4.2)	0.15 (6.8)	75.0	83.3	75.0	69.0	0.32 (7.5)	0.26 (11.9)
		Sweetened dairy products	0.03 (0.7)	0.01 (0.5)	12.5	5.6
		Sugar products	0.03 (0.7)	0.02 (0.9)	12.5	11.1
		Subtotal	0.24 (5.6)	0.18 (8.2)	100.0	100.0
	Seasonings	Seasonings	0.05 (1.2)	0.04 (1.8)	100.0	100.0	15.6	15.3
		Subtotal	0.05 (1.2)	0.04 (1.8)	100.0	100.0
	Fats	Plant-based fats	0.03 (0.7)	0.04 (1.8)	93.8	97.6	9.4	15.7
		Animal-based fats	0.002 (0.0)	0.001 (0.0)	6.2	2.4
		Subtotal	0.03 (0.7)	0.04 (1.8)	100.0	100.0

Values are mean daily diet-related GHG emissions (kg CO₂eq/day). GHG emission values are presented to two decimal places, with values below 0.01 kg CO₂eq/day shown to three decimal places; positive values below 0.001 kg CO₂eq/day are reported as < 0.001. Percentages in parentheses indicate contributions to total daily diet-related GHG emissions. Percentages under “Contribution within food group” and “Contribution within major food group” indicate the relative contribution of each subgroup or food group to the corresponding higher-level category. Totals may not sum exactly due to rounding.

GHG, greenhouse gas.

Although absolute emissions from plant-based foods decreased slightly from 1.31 to 1.25 kg CO₂eq/day, their relative contribution to total diet-related GHG emissions increased from 30.5% to 57.1% as emissions from animal-based foods declined. Within plant-based foods, grains remained the largest source of emissions, with their share of total diet-related GHG emissions increasing from 20.7% to 36.1%. The corresponding shares of vegetables, fruits, and plant-based protein foods increased from 7.7% to 15.1%, from 1.4% to 4.1%, and from 0.7% to 1.8%, respectively.

Emissions from the sugars, seasonings, and fats group decreased from 0.32 to 0.26 kg CO₂eq/day. Within this major food group, emissions from sweetened beverages decreased from 0.18 to 0.15 kg CO₂eq/day, although their contribution to total diet-related GHG emissions increased from 4.2% to 6.8% as total diet-related GHG emissions declined.

DISCUSSION

This study developed a sustainable dietary pattern for Korean adults by integrating cultural acceptability, nutritional adequacy, and environmental sustainability using nationally representative dietary intake data from the 2023 KNHANES [13]. The sustainable dietary pattern reduced diet-related GHG emissions from 4.29 to 2.19 kg CO₂eq/day while satisfying all predefined nutritional adequacy and cultural acceptability constraints. It was characterized by increased intakes of plant-based foods, including vegetables, fruits, whole and mixed grains, and plant-based protein foods, together with reduced intakes of red and processed meat, while total grain intake remained relatively stable.

Previous sustainable diet frameworks, such as the EAT-Lancet reference diet, have emphasized shifts toward more plant-based dietary patterns [2,3], while global modeling studies have shown that reductions in red meat consumption can contribute to lower environmental impacts and improved health outcomes [24,25]. Consistent with these frameworks, the sustainable dietary pattern in the present study included more plant-based foods and fewer high-emission animal-based foods. The increases in whole and mixed grains and plant-based protein foods, together with the decrease in red meat, were directionally consistent with the food-group gaps reported between Korean adult diets and PHD recommendations [12]. Rather than directly applying food-group targets from international reference diets, however, this study applied cultural acceptability constraints based on observed Korean intake distributions. Under these constraints, total grain intake remained close to the observed level, while the relative contribution of whole and mixed grains increased. The present study also used a detailed hierarchical food classification system comprising 9 food groups and 24 food subgroups, distinguished according to their nutritional characteristics and environmental relevance.

Within the grain group, the share of whole and mixed grains increased from 8.5% to 12.2%, while the intake of plant-based protein foods also increased. The vegetable and fruit intakes in the sustainable dietary pattern were 338 and 162 g/day, respectively, with their combined intake increasing from 396 to 500 g/day, thereby meeting the HP2030 target [22]. The increase in vegetable intake was consistent with a previous Korean optimization study [16], which reported that the optimal vegetable intake was higher than current intake levels across almost all demographic groups. Unlike that framework, which did not include dietary fiber as a nutritional constraint because the dietary fiber values were unavailable in the KNHANES data used at the time, the present study explicitly incorporated dietary fiber.

The estimated mean daily diet-related GHG emissions of the current dietary pattern among Korean adults aged 19–64 yrs (4.29 kg CO₂eq/day) fell within the range reported across country- and sex-specific adult groups in 5 European countries (3.40–5.79 kg CO₂eq/day) [26] and were lower than the 5.08 kg CO₂eq/day reported in an earlier Korean study [9]. These differences likely reflect methodological variations, including emission-factor databases, food-item matching procedures, survey years, and analytical design. Therefore, absolute estimates should be compared cautiously across studies.

Extending previous Korean research on diet-related GHG emissions and food-specific emission-factor databases [9], the present study demonstrated that a 49% reduction in diet-related GHG emissions was achievable under the specified nutritional adequacy and cultural acceptability constraints. This reduction was within the range projected by a global dietary scenario analysis [25] and below the theoretical maximum reductions reported in a multicountry European dietary optimization study [26]. Direct comparisons remain difficult because of differences in baseline diets, reference scenarios, system boundaries, emission-factor databases, and modeling assumptions. The potential for emission reductions varies by country [27], and the relatively large reduction observed in this study may partly reflect the composition of the Korean diet. Consistent with evidence that red meat has a substantially higher environmental footprint than plant-based protein sources [18], red meat intake was markedly reduced in the sustainable dietary pattern.

Beyond environmental outcomes, the sustainable dietary pattern may have public health relevance because it increased fruit and vegetable intake and reduced sodium intake, changes consistent with the objectives of HP2030 [22]. Although health outcomes were not directly evaluated, this interpretation is broadly consistent with the EAT-Lancet Commission’s conclusion that shifts toward healthier dietary patterns could reduce premature mortality [2].

At the food-subgroup level, the sustainable dietary pattern showed a substantial reduction in processed meat intake. Given that the International Agency for Research on Cancer has classified processed meat as carcinogenic to humans (Group 1) based on sufficient evidence for colorectal cancer [28], this reduction may have additional public health relevance, although disease outcomes were not assessed in this study. In contrast, sweetened beverage intake remained essentially unchanged, decreasing from 96.4 to 96.1 g/day. This limited change should be interpreted in the context of the nonbinding constraint on total sugars, with total sugars contributing 13.0% of energy compared with the 20% upper bound, and the objective of minimizing deviations from observed food intakes.

Dietary optimization provides a useful framework for balancing nutritional, environmental, and cultural acceptability considerations [29]. Methodologically, this study combined a hierarchical food classification system with observed intake-based cultural acceptability constraints and elasticity analysis to narrow the gap between theoretical optimality and compatibility with observed dietary practices. Compared with the earlier Korean optimization study, which used linear programming to minimize the sum of absolute relative deviations across 24 food subgroups [16], the present study applied quadratic programming at the representative-food level. This objective can help limit extreme deviations and favor smaller changes across a broader set of foods because squared deviations assign progressively greater penalties to larger item-level changes [29]. Nevertheless, the two studies also differed in the resolution of decision variables and the scaling of deviations. Therefore, differences between their solutions cannot be attributed solely to the use of linear versus quadratic programming. Overall, the present study extends the previous Korean optimization framework [16] by integrating representative-food-level quadratic programming with diet-related GHG emission constraints and elasticity analysis.

Several limitations should be acknowledged. Dietary intake was assessed using a single 24-h dietary recall from the KNHANES, which may not fully capture usual individual intake. Cultural acceptability was inferred from observed intake distributions and was not directly evaluated through consumer acceptance or intervention studies. Therefore, the cultural acceptability constraints used in this study should be interpreted as reflecting compatibility with observed dietary practices rather than demonstrated consumer acceptance or actual adoptability. Because these percentile-based bounds were derived from the distribution of individual intakes but applied to the modeled population-average dietary pattern, they should be interpreted as plausibility bounds for that pattern rather than as limits applicable to individual diets. Economic and practical considerations, such as food cost and accessibility, also warrant further investigation. In addition, as with all optimization models, the resulting dietary pattern depends on the specification of the objective function and the predefined nutritional adequacy, cultural acceptability, and environmental sustainability constraints. Because some food-subgroup intakes, including those of nuts and seaweeds, were at or near their reported upper bounds, future studies should assess the robustness of the solution under alternative acceptability bounds. Moreover, its applicability to other countries may be limited because baseline dietary patterns and food-system characteristics differ across countries [27].

The scope of nutritional and environmental assessments was also limited. The optimization model included energy and 17 nutrients but did not include vitamin B12, zinc, folate, or vitamin D. Therefore, the adequacy of the sustainable dietary pattern with respect to these nutrients was not evaluated and should be examined in future studies using appropriate food-composition data. Environmental impacts were limited to GHG emissions; future research should incorporate broader sustainability indicators, such as water use, land use, and biodiversity impacts. As demonstrated by Sandström et al. [30] in their analysis of EU diets, estimates of diet-related GHG emissions can also be influenced by international trade and the countries of origin of food products because production efficiencies and land-use histories vary across sourcing regions. In this study, GHG emission factors were compiled from multiple sources, including the Korea-specific food system GHG emission factor database developed by Hong and Kim [9], and substitute values were used when direct matches were unavailable, potentially introducing uncertainty into food-specific emission estimates. Future studies in Korea should therefore incorporate more detailed origin-specific and trade-adjusted emission factors and further refine and validate Korea-specific food-system GHG emission-factor databases.

Finally, because the optimization model was developed to derive a nationally representative dietary pattern for the general Korean adult population, sex- and age-specific dietary requirements and consumption patterns were not modeled separately. The resulting dietary pattern may therefore not fully reflect the nutritional needs or dietary characteristics of specific demographic subgroups. Future studies should develop demographic-specific optimization models by sex and age and evaluate the long-term health and environmental outcomes associated with adoption of sustainable dietary patterns in institutional and community settings.

In conclusion, this study identified a sustainable dietary pattern for Korean adults that reduced diet-related GHG emissions by 49% while satisfying predefined nutritional adequacy and cultural acceptability constraints. The sustainable dietary pattern was characterized by increased intakes of plant-based foods and reduced intakes of high-emission animal-based protein foods, while remaining relatively similar to the current Korean dietary pattern. These findings may contribute to the evidence base for developing sustainable dietary guidance tailored to the Korean context and inform future nutrition and environmental policies.