Computationally Modeling the Physiologic Impact of the Ratio of Fats to Carbohydrates in the Diet on Intake Among Metabolically Healthy Adults

Abstract

Background

When it comes to how effectively a diet can help reduce or maintain body weight, a key question is how that diet affects a person’s hunger, satiety, and subsequent eating.

Objectives

This study aimed to analyze modeling, from a physiologic perspective, how varying the ratio of fats to carbohydrates in a diet impacts hunger, satiety, and subsequent eating among metabolically healthy adults.

Methods

We developed a model representing an adult, their dietary intake, gastrointestinal tract, hunger/satiety levels, and meal consumption. We simulated agents eating fixed ratios of macronutrients and measured their subsequent eating over 24 h driven by physiologic responses.

Results

When increasing the proportion of energy from fats relative to carbohydrates, daily calories decrease by on mean 149 and 110 calories per 10% increase in fats for males and females, respectively. Additionally, a simulated diet with a relative ratio of energy from fats:carbohydrates of 20%:80% results in individuals snacking after 21:00 for ∼93% of days in both sexes, whereas a relative fat:carbohydrate ratio of 80%:20% results in late-night snacking ∼55% and ∼60% of days for males and females, respectively. Agents consuming at least a 40%:60% relative ratio of energy from fat:carbohydrate ratio can achieve the largest reductions in total calories consumed and late-night snacking compared with consuming higher relative proportions of carbohydrates.

Conclusions

Eating a diet with ≥40% of its energy from fats relative to carbohydrates can achieve the largest reductions in total calories consumed and late-night snacking each day than consuming higher proportions of carbohydrates, with even further reductions as more fat is added to the diet, when considering the physiologic responses to dietary intake alone. Future research should layer in other strong contributing factors to eating such as stress, social context, palatability, physical activity, and types of macronutrients, and also represent other metabolic profiles and ages.

Introduction

A number of nutrition influencers have been asserting that diets higher in fat and lower in carbohydrate content will help better manage dietary intake and thus weight, which begs the question, can such diet compositions alone decrease hunger and change how much a person consumes? There are also claims that high-carbohydrate diets rich in whole fruits, nonstarchy vegetables, legumes, and whole grains can help individuals lose weight and body fat without adding any exercise [[1], [2], [3], [4], [5], [6]]. Both extreme diets of high-carbohydrate and high-fat compositions have been advertised to metabolically healthy young adults for these benefits of reducing overall body weight [1,2,[7], [8], [9]].

One of the criticisms of these claims is that they discount the other factors that may influence hunger and satiety, such as stress eating, ingrained daily eating habits, such as eating 3 large meals a day regardless of hunger levels; the social context of eating (i.e., eating behavior being influenced by the people one eats with); the palatability of foods, where one might be driven to eat by food/drink they find appetizing (i.e., aromatic and flavorful) despite not being physically hungry; and compensatory eating from physical activity [[10], [11], [12], [13], [14], [15], [16], [17], [18]]. However, the question remains, what is the relationship between the ratio of dietary fats and carbohydrates on hunger/satiety and subsequent caloric intake among metabolically healthy adults? Therefore, we developed a computational simulation model to start to explore this by assessing the physiologic impact of varying the relative ratio of energy from fats to carbohydrates in the diet on hunger and satiety levels and subsequent eating behavior alone, holding all other factors that impact eating behavior equal, to better understand the physiologic underpinnings that may predict eating behavior responses among metabolically healthy young adults [19].

Methods

Model overview

We developed an agent-based computational simulation model in Python version 3.9 representing a metabolically healthy adult (e.g., normal range for cholesterol, triglycerides, insulin, and glucose), aged 25–40 y, their dietary intake, gastrointestinal (GI) tract, hunger/satiety levels, and meals (Figure 1). We represented this age range because adults over 40 y may take medications that affect GI processes, metabolism, and hunger. Each agent consumes a diet with a fixed ratio of energy from fats and carbohydrates and a fixed percentage of energy from protein, assuming 100% diet adherence. Table 1 [[20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36]] shows model input parameters, values, and sources. Supplemental Figure 1 describes the procedure used to identify values for each of the model parameters. Supplemental Methods also describes the model validation and calibration. The model proceeds in 1-min time steps for a 24-h duration and includes the submodels as described further.

FIGURE 1.

Model of fat and carbohydrate consumption’s impact on hunger, satiety, and subsequent intake. CCK, cholecystokinin; GI, gastrointestinal; GLP, glucagon-like peptide; PYY, peptide tyrosine tyrosine.

TABLE 1.

Table of inputs.

Parameter			Distribution, value		Source
Density of fats (g/mL)			Uniform, 0.7–0.96		[20]
Density of carbohydrates (g/mL)			Uniform, 0.117–1.4		[20]
Mouth and upper GI submodel
Eating rate: meal (kcal/min)
Males			Uniform, 28.7–32.6		[21], assuming meals are 30-min long, using 95% CIs
Females			Uniform, 20.0–22.3		[21], assuming meals are 30-min long, using 95% CIs
Eating rate: snack (kcal/min)
Males			Uniform, 3.3–6.4		[21], assuming meals are 30-min long, using 95% Cis
Females			Uniform, 2.13–3.63		[21], assuming meals are 30-min long, using 95% Cis
Percent protein
Males			0.16 (0.003)2		[21]
Females			0.15 (0.002)2		[21]
Stomach submodel
Initial gastric distention (at baseline) (mL)			296		[22]
Gastric distention constant			0.709		[22]
Gastric emptying (min)
Half-life of fats in stomach			193		[23]
Half-life of carbohydrates in stomach			43		[23]
Breakdown of total calories consumed by nutrient type and ratio of fats to carbs
Ratio of Fats to Carbs	Males			Females
	Protein (%)	Fats (%)	Carbohydrates (%)	Protein (%)	Fats (%)	Carbohydrates (%)
10:90	16.0	8.4	75.6	15.0	8.5	76.5
20:80	16.0	16.8	67.2	15.0	17.0	68.0
30:70	16.0	25.2	58.8	15.0	25.5	59.5
40:60	16.0	33.6	50.4	15.0	34.0	51.0
50:50	16.0	42.0	42.0	15.0	42.5	42.5
60:40	16.0	50.4	33.6	15.0	51.0	34.0
70:30	16.0	58.8	25.2	15.0	59.5	25.5
80:20	16.0	67.2	16.8	15.0	68.0	17.0
90:10	16.0	75.6	8.4	15.0	76.5	8.5
Small and large intestine submodel
Small intestine emptying (h)
Half-life of fats in upper small intestine			6		[24]
Half-life of carbohydrates in upper small intestine			3		[24]
Half-life of fats in lower small intestine			6		[24]
Half-life of carbohydrates in lower small intestine			3		[24]
Absorption rate in small intestine (during transit from upper to lower small intestine) (%)
Fats			30		Calibrated,1 expert opinion
Carbohydrates			80		Calibrated, expert opinion
Absorption rate in small intestine (in lower small intestine) (%)
Fats			50		Calibrated,1 expert opinion
Carbohydrates			100		Expert opinion
Rate of leaving the large intestine (mL/min)
Fats			0.05		Calibrated1
Carbohydrates			0		Expert opinion
GLP-1
Effect of eating on GLP-1 production (aGLP1) (pM/min)			0.75	[[25], [26], [27], [28]]
Effect of lower small intestine fats on GLP-1 production (bGLP1) (pM/min/mL)			0.2	[[25], [26], [27], [28]]
Effect of lower small intestine carbohydrates on GLP-1 production (dGLP1) (pM/min/mL)			0.2	[[25], [26], [27], [28]]
Effect of large intestine fats on GLP-1 production (cGLP1) (pM/min/mL)			0.2	[[25], [26], [27], [28]]
GLP-1 decay rate (eGLP1) (per min)			0.06	[[25], [26], [27], [28]]
CCK
Effect of upper small intestine fats on CCK production (aCCK) (pM/min/mL)			0.01	[[29], [30], [31]]
Effect of upper small intestine carbohydrates on CCK production (bCCK) (pM/min/mL)			0.005	[[29], [30], [31]]
CCK decay rate (cCCK) (per min)			0.03	[[29], [30], [31]]
PYY
Effect of lower small intestine fats on PYY production (aPYY) (pM/min/mL)			2	[32]
Effect of lower small intestine carbohydrates on PYY production (cPYY) (pM/min/mL)			0.8	[32]
Effect of large intestine fats on PYY production (bPYY) (pM/min/mL)			1.5	[32]
PYY decay rate (dPYY) (per min)			0.075	[32]
Ghrelin
Effect of upper small intestine fats on ghrelin production (bGhrelin) (pM/min/mL)			−0.005	[28,33,34]
Effect of upper small intestine carbohydrates on ghrelin production (dGhrelin) (pM/min/mL)			−0.005	[28,33,34]
Effect of stomach fats on ghrelin production (aGhrelin) (pM/min/mL)			−0.01	[28,33,34]
Effect of stomach carbohydrates on ghrelin production (cGhrelin) (pM/min/mL)			−0.01	[28,33,34]
Ghrelin decay rate (eGhrelin) (per min)			0.04	[28,33,34]
Ghrelin base concentration (pM)			110	[35]
Brain submodel
Impact of gastric distention on satiety (aSatiety) (per mL)
Gastric distention <296 mL			0	[22,36]
Gastric distention >296 to <500 mL			0.0025	[22,36], Calibrated1
Gastric distention >500 mL			0.0035	[22,36], Calibrated1
Impact of PYY on satiety (cSatiety) (per pM)			0.08	Calibrated1
Impact of GLP-1 on satiety (dSatiety) (per pM)			0.2	Calibrated1
Impact of Ghrelin on satiety (eSatiety) (per pM)			0.02	Calibrated1
Impact of CCK on satiety (bSatiety) (per pM)			1.2	Calibrated1

Abbreviations: CCK, cholecystokinin; GLP, glucagon-like peptide; PYY, peptide tyrosine tyrosine.

¹Calibrated based on approximate hormone ranges and the relative effects of fats and carbohydrates on the release of each hormone.

²Values are mean (SD).

Representing dietary intake (intake submodel)

Each day, agents in the model have assigned eating windows where they can start and end meals based on their hunger/satiety level, which are driven purely by physiologic responses to previously consumed food intake.

When an agent initiates an eating event (driven by their hunger level), they consume a volume (milliliters) of carbohydrates, fats, and protein every minute based on the fixed ratio of fats to carbohydrates in their diet. We assumed a certain percentage of calories came from protein [21] which remained constant with the remaining percentage of calories broken down by the ratio of fats:carbohydrates (Table 1). Carbohydrates and fats have associated densities (drawn from a distribution, representing a combination of different types of fats and carbohydrates), with fats generally being more energy-dense than carbohydrates.

Representing gastric distention and emptying (stomach submodel)

Food then enters the stomach submodel, which represents gastric distention and emptying. Gastric distention is influenced by the volume (concentration in milliliters) of consumed food entering the stomach, and the rate at which food moves from the stomach into the small intestine (gastric emptying). The gastric emptying rate is determined by the volume of the food consumed and the type of macronutrients present because fats and carbohydrates have different rates of gastric emptying. We represent gastric emptying using exponential decay [37].

The following equations represent gastric emptying for fats and carbohydrates:

RateGastricEmptying_fats = Volume_{FatsInStomach} × [−log(0.5)/Half-life_{FatsInStomach}]

RateGastricEmptying_{Carbohydrates} = Volume_{CarbohydratesInStomach} × [−log(0.5)/Half-life_{CarbohydratesInStomach}]

Given a lack of studies quantifying gastric emptying rates with different types and quality of fats and carbohydrates, we assumed half-lives are constant for all types of fats and constant for all types of nonfiber carbohydrates [23]. The speed of emptying is fastest right after a meal and decreases sharply over time.

Representing the release of GI hormones (GI hormone submodel)

Once chyme enters the small intestine, specialized enteroendocrine cells detect macronutrients and/or their digestive products and trigger the release of different GI hormones. We represent the secretion of 4 key GI/hunger-satiety–related hormones: cholecystokinin, glucagon-like peptide (GLP)-1, peptide tyrosine tyrosine (PYY), and ghrelin. The concentrations of each hormone are represented as quantities that increase in response to fats and carbohydrates in the GI tract, at hormone-specific rates, and decrease (decay) over time if not actively secreted (Supplemental Material).

Representing hunger and satiety and subsequent caloric intake (brain submodel)

We defined hunger as the process, including physical sensations in the body, which produces the drive to eat. Inhibiting this process contributes to satiation during a meal, signaling the person to stop eating, and postprandial satiety between meals, prolonging the time in between eating events. We measure satiety on a score from 1 to 10, corresponding to standard visual analog scoring methods used to measure hunger/satiety [38]. Satiety increases and decreases throughout the day in response to gastric distention and the concentration of each hormone, where a_Satiety–e_Satiety are constants (Table 1), as follows:

SatietyScore = a_Satiety × GastricDistention + [b_Satiety × (cholecystokinin)] + [c_Satiety × (PYY) × d_Satiety × (GLP)-1] + [e_Satiety × (FastingGhrelinLevel − PostPrandialGhrelin)]

In an eating window, an agent evaluates their satiety level every hour to determine whether they will start eating. Each satiety level is linearly associated with a consumption probability (e.g., a score of 5 translates to a 50% probability of starting to eat). After 15 minutes of eating, an agent has a probability each minute to stop eating based on their satiety score. Hunger and satiety in this model are driven purely by physiologic responses to food intake, and not driven by daily habits, physical activity, social influences, or any other nonphysiologic factors [39,40].

Model validation

The model was validated through running multiple trials varying parameters and comparing the output against known data points from other studies for alignment. Our model aimed for a daily total caloric intake of between 1700 and 2700 for a metabolically healthy, normal-weight adult. Additional details on specific values can be found in the Supplemental Material.

Experimental scenarios

Experimental scenarios simulated an agent following a hypothetical diet with a fixed ratio of relative energy from fats to carbohydrates for 24 h, set at 100% adherence. The first set of scenarios simulated males, whereas the second simulated females. Different scenarios varied the fat:carbohydrate ratio from 10:90 to 90:10, holding all nonphysiologic factors constant. Sensitivity analyses varied the eating probability during an eating event window based on their satiety score, narrowing the range from 0%–100% to 10%–90% and then to 20%–80% (for satiety scores of 0–10), to represent that people may or may not eat, independent of physiology-driven hunger/satiety levels. Each experimental scenario consisted of running 1000 trials.

Results

Impact of fat:carbohydrate ratio on the mean and largest meal size consumed when evaluating physiologic responses to diet alone

Agents who eat a diet with a higher proportion of energy from fats to carbohydrates on mean eat larger meals (i.e., higher calorie/energy) per 24-h period as shown in Figure 2A. The calories in the largest meals consumed daily (e.g., meal with the most energy/calories) increases approximately linearly as the relative proportion of fats in the diet increases, by ∼2.6 calories and 0.36 calories per 10% relative increase in fats for males and females, respectively. There were no discernable trends in the timing of when the largest meal of the day occurred.

FIGURE 2.

Impact of varying ratio of fats to carbohydrates in diet on mean calories per largest eating event (A), number of eating events per day (B), time between meals (C), percentage of time late-night snacking (D), and total daily calories (E).

Additionally, the mean caloric size of each meal increases by ∼0.99 calories per 10% relative increase in fats for males and decreases by ∼0.25 calories per 10% relative increase in fats for females. For example, individuals consuming a diet with a relative 20%:80% fat:carbohydrate ratio consume an mean of 499 (95% CI: 495, 504) and 349 (95% CI: 346, 353) calories per meal for males and females, respectively. In contrast, individuals eating meals with 80% fats and 20% carbohydrates consume an mean of 506 (95% CI: 500, 513) and 348 (95% CI: 343, 352) calories per meal for males and females, respectively.

Impact of fat:carbohydrate ratio on the number of eating events per day when evaluating physiologic responses to diet alone

Individuals who consume diets with a higher proportion of fats to carbohydrates also consume fewer meals (>250 calories) and snacks (≤250 calories) per day (Figure 2B). When individuals consume a diet with a relative fat:carbohydrate ratio of 20%:80%, they eat on mean 8 (95% CI: 7.97, 8.13) and 9.2 times (95% CI: 9.1, 9.3) per day for males and females, respectively. In contrast, individuals eating meals with 80% fats and 20% carbohydrates, end up eating on mean 5.2 (95% CI: 5.09, 5.23) and 6.4 times (95% CI: 6.3, 6.44) per day for males and females, respectively. For every 10% increase of relative fats to carbohydrates in the diet, males and females consume ∼0.5 fewer times per day.

Impact of fat:carbohydrate ratio on the mean time between eating events when evaluating physiologic responses to diet alone

Changing the ratio of fats to carbohydrates in a diet also impacts the mean amount of time between meals, such that agents consuming diets with a higher percentage of fats to carbohydrates eat less frequently. Figure 2C shows that when individuals consume a diet with a relative 20% to 80% ratio of fats to carbohydrates, they eat a meal on mean every 122 minutes (95% CI: 118, 127) for males and every 102 minutes (95% CI: 98, 106) for females. The meane time between meals increases approximately linearly as the proportion of relative fats to carbohydrates in the diet increases, by ∼15 min for males and by ∼10 min for females per 10% increase of fats in the diet relative to carbohydrates.

Impact of fat:carbohydrate ratio on late-night snacking when evaluating physiologic responses to diet alone

In addition to eating less frequently throughout the day, individuals who consume diets with a higher proportion of fats to carbohydrates are less likely to eat late-night (e.g., 21:00 to 23:00) snacks. Figure 2D shows how varying the relative proportion of energy from fats to carbohydrates in a diet affects how late in the day individuals consume snacks. When individuals consume a diet with a relative fats-to-carbohydrates ratio of 20% to 80%, they eat a snack after 21:00 on 93% of days for both males and females, and this percentage decreases by ∼6.1% for males and 4.8% for females for every 10% increase in relative fats in the diet. Males who consume a diet with a relative fat:carbohydrate ratio of ∼75%:25% and females with a relative ratio of 90%:10% have <60% likelihood of snacking after 21:00. Figure 2D also shows that the decrease becomes slightly steeper after eating a meal of 20% fats and 80% carbohydrates (males) and 40% fats and 60% carbohydrates (females). For males, increasing the ratio of relative fats to carbohydrates from 10% to 20% decreased the snacking probability after 21:00 by ∼1.8%, and increasing the relative proportion of fats further decreased the snacking probability after 21:00 by ∼6.3% per 10% increase in fats relative to carbohydrates. Between 10% and 30% fats relative to carbohydrates, increasing the ratio of fats to carbohydrates by 10% decreases the snacking probability by ∼2.7% in females. Between 40% and 90% relative fats compared with carbohydrates, the decrease in snacking probability was even more profound at a 5.9% decrease. These results show that the greatest gains can be achieved when the dietary ratio of fats to carbohydrates is >20%-40% fats, with further reductions in late-night snacking as the ratio of fats in the diet increases.

Impact of fat:carbohydrate ratio on total calories consumed daily when evaluating physiologic responses to diet alone

Despite consuming larger meals, individuals consume fewer calories overall as the proportion of fats in their diet increases. Figure 2E shows how varying the relative proportion of fats to carbohydrates from a ratio of 10%:90% to 90%:10% in a diet affects a person’s daily caloric intake. For example, when individuals consume a relative diet ratio of energy from 20% fats to 80% carbohydrates, they consume an mean total of 2989 (95% CI: 2969, 3009) cal/d for males and 2429 (95% CI: 2412, 2446) cal/d for females, corresponding to an mean of 498 and 349 calories per eating event for males and females, respectively. For females, between 10% and 30% relative fats, each increase in the ratio of fats by 10% decreased total calories consumed by ∼158 calories. This decreased to ∼92 calories per 10% increase in the ratio of relative fats as the ratio increases past 40%. For males, between 10% and 30% relative fats, each increase in the ratio of fats by 10% decreased total calories consumed by ∼218 calories, compared with a decrease of ∼110 calories per 10% increase in the ratio of relative fats as the ratio increases from 40% to 90%.

Effect of varying effect of hunger and satiety on caloric consumption when evaluating physiologic responses to diet alone

To represent that there are other factors influencing caloric consumption, we decreased the effect of hunger/satiety on the chance of eating, meaning that individuals have a chance of eating when not hungry and a chance of not eating when hungry. Varying the ratio of relative fats to carbohydrates in the diet continued to show similar trends, with an increased ratio of relative fats resulting in decreased calories consumed, fewer and slightly larger meals, and longer times between meals. For example, when decreasing the effects of hunger/satiety on an individuals’ chance of initiating or stopping an eating event, for males, every 10% increase in relative fats to carbohydrates in the diet results in 0.25 fewer meals, which translated to 141 fewer calories consumed, and time between meals increased by ∼14 minutes. For females, every 10% increase in relative fats to carbohydrates in the diet results in 0.25 fewer meals, which translated to 100 fewer calories consumed, and time between meals increased by ∼9 minutes.

Discussion

Our results showed that consuming a diet with higher relative fat-to-carbohydrate ratios is ideal for reducing hunger and eating less over the course of the day, but a threshold for achieving the greatest health benefits while still being realistic to follow (i.e., taste and cost) is a relative macronutrient ratio of ≥40% fats and 60% carbohydrates. The increase from a 30% to 40% energy from fat relative to carbohydrates leads to the greatest reductions in hunger, leading to fewer total calories consumed and less late-night snacking each day, relative to other fat proportion increases. Studies show there are health-promoting benefits to consuming a higher proportion of energy early in the day, as well as reduced meal frequency (e.g., 2–3 meals/d) [41]; thus, it may be important to consume a diet with ≥40% of its energy from fats relative to carbohydrates to achieve these benefits.

Our study confirms that when homing in on the physiologic response to dietary intake alone, fats are more satiating than carbohydrates, leading to fewer calories consumed in a day. This differs from some previous studies that found carbohydrates to be more satiating [[42], [43], [44]]. The previous experiments that supported the hypothesis that carbohydrates are more satiating than fats focused on the influence of a single meal on subsequent food intake at the next meal with fixed intervals between meals [42,[45], [46], [47]]. Our model found fats to be more satiating due to extending beyond a single eating event because we represented eating over 24 h, including ending a meal early, the time between meals, and daily calorie amounts. Because carbohydrates have lower energy density than fats, they contribute more to feelings of fullness, a sensation typically ascribed to gastric filling and distension, commonly associated with meal termination [48,49]. However, fats are digested and absorbed more slowly than carbohydrates, leading to a prolonged intestinal contact time that results in a sustained release of GLP-1 and PYY, which contributes to extended satiety, or time between meals. This is consistent with our findings showing longer intermeal intervals and prolonged satiety as the relative fat ratio increases. Our findings are also consistent with human studies demonstrating suppressed hunger on high-fat ketogenic-type diets [50,51], and animal studies focused on understanding gastric and intestinal phases of satiation/satiety [[52], [53], [54], [55], [56], [57], [58], [59]].

This model importantly brings together and mechanistically represents the physiologic responses related to dietary intake and its effect on hunger and satiety and subsequent eating over time, showing the impact various diets have on hunger and satiety and caloric consumption. Isolating this impact of the dietary macronutrient ratio on hunger/satiety and eating, before layering in other confounding factors that influence eating, is necessary in order to first answer the question of how a diet alone may help with maintaining or reducing weight. Several studies have modeled the effect of specific hormones on hunger [23,36,60,61], and various experimental studies have linked specific hormones with perceptions of hunger/satiety after single meals [25,62,63]. However, these studies have not investigated the subsequent effect on food intake over time. Additionally, although previous models have represented the effect of food intake on hormone secretion, gastric emptying, and macronutrient absorption [23] and how food intake ultimately affects metabolism and body mass [64,65], these models have not incorporated how these processes affect and are affected by hunger/satiety. Another study [66] modeled the initiation of eating events based on computationally derived plasma ghrelin and glucose concentrations and their impacts on hunger/satiety levels throughout the day but did not include other hormones and GI processes. Our work addresses the gaps in the existing research by explicitly representing the effects of macronutrients on hormones, food intake, and hunger/satiety while varying ratios of fats to carbohydrates. This approach provides a foundation for adding in other important factors that, when combined to represent the full system of factors, can help to provide insights into more practical and applicable dietary recommendations for the general population.

One of the next steps is to add into the model the other factors that contribute to hunger/satiety and eating, such as unconscious eating, stress, the social context of eating events (e.g., the people one eats with), the palatability of foods, physical activity, water consumption, and the quality and types of fats and carbohydrates [[10], [11], [12], [13], [14], [15], [16], [17], [18]]. It will also be important to represent these effects for other age and BMI (in kg/m²) groups, as well so that the model can then translate into tools that can offer personalized dietary advice for individual diet plans according to their unique metabolic profile. Dietary guidelines derived from such insights may mitigate the struggle many individuals experience during weight loss and while improving their metabolic health by targeting an appropriate macronutrient ratio.

Limitations

Models, by definition, are simplifications of reality and cannot capture the full complexity of every factor involved in a system. We did not represent every factor impacting intake and appetite such as the taste, texture and flavor of foods, or carbohydrate and fat quality [10,12,13,[67], [68], [69], [70], [71], [72]]. We also did not account for all dietary components (e.g., indigestible fiber and water). To isolate and quantify the impact of the relative ratio of energy from fats to carbohydrates on hunger/satiety and eating behavior, our scenarios assume that all individuals have unlimited access to food, although existing studies have shown that people’s eating habits and subsequent hunger levels are influenced by a range of contextual factors including their social networks, economic status, home neighborhood, and their access to different types of food [73]. Additionally, although protein is represented, we did not explicitly represent its effects on hormone release and GI processes. We also did not explicitly represent the impact of different GI hormones on the gastric emptying rate or intestinal motility, which are accounted for in the equation constants. Our model represents the impact of different diets over the course of a day on metabolically healthy adults, and future iterations can incorporate longer-term processes such as the effects of leptin and how other, nonhealthy individuals may adapt to diets over time. Finally, we did not represent unique caloric needs for different individuals or hormone release and size differences between males and females. As updated empirical data from actual dietary intervention studies in the real world emerge, we can continuously update and further validate model parameters to reflect a variety of populations.

Conclusions

Our study shows that eating a diet with ≥ 40% of its energy from fats relative to carbohydrates can achieve the largest reductions in total calories consumed and late-night snacking each day than consuming higher proportions of carbohydrates, with even further reductions as more fat is added to the diet, when considering the physiologic responses to dietary intake alone. Higher proportions of fats in the diet also lead to fewer and slightly larger eating events with longer time intervals between them.

Author contributions

The authors’ responsibilities were as follows – MFM, JH, THM, BB-F, JL, SLB, SK, JMO, KdlH, SMB, BYL: designed the research; MFM, JH, CW, KV, TDS, KLC, SMB, AD, SAS, BLY: conducted the research; MFM, JH, CW, KV, TDS, KLC, SMB, BYL: analyzed data; BYL: had primary responsibility for final content; and all authors: wrote the article and read and approved the final manuscript.

Data availability

Data described in the manuscript, code book, and analytic code will be made available upon request pending approval.

Funding

This work was supported by the NIH Common Fund’s Nutrition for Precision Health, powered by the All of Us Research Program and the National Center for Advancing Translational Sciences of the NIH (U54TR004279); the Agency for Healthcare Research and Quality through (1R01HS028165-01); the National Institute of General Medical Sciences as part of the Models of Infectious Disease Agent Study network (R01GM127512 and 3R01GM127512-01A1S1); and the National Science Foundation (NSF; 2054858). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of or imply endorsement by NIH, Agency for Healthcare Research and Quality, or NSF.

Conflict of Interest

The authors report no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: