METHODS FOR PROVIDING PERSONALIZED DIET AND ACTIVITY RECOMMENDATIONS THAT ADAPT TO THE METABOLISM OF EACH DIETER INDIVIDUALLY USING FREQUENT MEASUREMENTS OF ACTIVITY AND TRIGLYCERIDE LEVELS
FIELD OF THE INVENTION
The present subject matter relates to weight management models and methods. More specifically, the present subject matter relates to computer models, algorithms, and methods that provide a personalized adaptiveweight management solution.
BACKGROUND
For many reasons, including health, appearance, and body image, many people are concerned with achieving and maintaining a desired body weight, most of themwant to reduce theirs weight. Weight loss plans typically rely upon gross metrics in order to suit a diet plan for the individual, these often include gender, age, food preferences and weight-loss goals.
It should be noted that most weight loss regiments do not take into consideration the individual's unique physiology and reaction to differing dietary intake and activity levels and are not dynamically adaptable over time. It has been shown that people respond very differently to the food they eat and that their fitness shape significantly affects how they respond to food intake. Most current weight loss and healthy diet systems provide generic guidance to people, but virtually none rely upon biological feedback in order to tailor the diet to the user's unique metabolism at any given point in their life.
83% of dieters gain back more weight than they had lost within 2 years according to a wide research conducted in UCLA. The weight comes off, but the weight loss isn't sustainable because it's too hard to maintain the extreme habits that one-size-fits-all diets require over time.
Previous research has suggested that personalization may be useful for weight management programs. While promising, these previous efforts used tools such as progress charts and journals which may be difficult for many individuals to maintain and use consistently. Thus, there remains a need for an easy to use personalized weight loss program. SUMMARY
The present subject matter relates to models and methods for providing personalized diet and activity recommendations that adapt to the metabolism of each dieter individually using frequent measurements of activity and triglyceride levels. In general, the subject matter provides the means for generating weekly plans and daily diet and activity guidance using repeated, frequent and ongoing tracking of: body weight, physical activity, triglyceride levels, and possible use of various other physiological and behavioral parameters such as age and food preferences.
In one embodiment the present subject matter provides a method for generating a personal metabolism reference model for an individual that can be used to predict how the person's triglyceride levels will change in response to food consumption and physical activity. The method for generating the metabolism reference model groups food by amount, density, type and quality of fats, carbohydrates, and protein that are typical for food in the group. The method groups physical activities by duration, type, frequency and intensity. For each of the groups of food and activity a triglyceride metabolic tolerance score is defined as a function of the delta in triglyceride levels in certain points-of-time after eating the food or performing the activity in certain points of timeduring the day. The personal metabolism reference model for an individual is determined during a several days of pre-defined diet and activity program - during this period the user follows a subscribed diet plan and the system frequently measures how his triglyceride levels are affected by certain food intake and certain activities.
In another embodiment the present subject matter provides a method for generating a personal diet reference model for an individual using detailed, frequent and ongoing records of his activity, triglyceride levels and weight over time. The model can be used to determine the range of required triglyceride levels in every point of time during the day for an individual to reach his weight management goals. The method uses cohorts of people with similar physical properties and similar metabolism reference models, the levels of triglyceride are measured for people in the cohort over several days for a predefined set of diet and activity plans and the triglyceride threshold levels range for various weight management goals is determined as a function of these measurements. The personal diet reference model for the user is determined in a several days of diet and activity program in which the user is associated with one or more similarity cohorts.
In another embodiment the present subject matter provides a method for generating diet and activity recommendations for each individual to help him reach his weight management goals using his personal metabolism reference model, his personal diet reference model, his weight management goals, his recent records of activity, triglyceride levels and body weight, his profile, his food preferences and his feedback. The system generates weekly diet and activity plan that provides a framework for the user's diet. The user's personal diet reference model, his weight management goals, his food and activity preferences, his recent tolerance levels and his recent weight changes are used to define a diet framework for the user's weekly plan. The diet framework determines the size, composite, and time of meals and activities in the plan for the individual. The user's weekly diet framework is used to provide food recommendations for the weekly plan. The specific food recommendations is determined using the user's metabolism reference model and food and activity preferences, using a knowledge base that maps food plans to different metabolism models. The system generates daily diet and activity guidance using repeated, frequent and ongoing tracking of physical activity and triglyceride levels - in case the triglyceride levels and physical activity levels are not within the thresholds defined in the user's weekly diet framework, the system generates diet and activity guidancethat is aimed to keep the user's diet balanced and get his triglyceride levels to levels that are expected to help him reach his weight management goals.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 - illustrates 2 measures of triglyceride levels after eating a food from group A;
Figure 2 - illustrates a personal diet reference model for a user with a specific weight management goal;
Figure 3 - illustrates the measurements of triglyceride levels for a cohort of similar people on a given diet during a week;
Figure 4 - illustrates the steps for building a weekly diet plan for a user;
Figure 5 - illustrates the process of providing weekly food recommendations from the diet framework;
Figure 6 - illustrates the process of providing daily food and activity recommendations in the case of triglyceride imbalance.
DETAILED INVENTION
The present invention will now be described in detail with reference to several embodiments thereof as illustrated in the accompanying figures. In the following description, numerous specific details are set forth in order to provide a thorough understanding of embodiments of the present invention. It will be apparent, however, to one skilled in the art, that embodiments may be practiced without some or all of these specific details. In other instances, well known process steps and/or structures have not been described in detail in order to not unnecessarily obscure the present invention. The features and advantages of embodiments may be better understood with reference to the drawings and discussions that follow.
The following description of some embodiments will be provided in relation to numerous subsections. The use of subsections, with headings, is intended to provide greater clarity and structure to the present invention. In no way are the subsections intended to limit or constrain the disclosure contained therein. Thus, disclosures in any one section are intended to apply to all other sections, as is applicable. METHOD FOR GENERATING A PERSONAL METABOLISM REFERENCE MODEL
The present subject matter provides a method for generating a personal metabolism reference model for an individual that can be used to predict how the person's triglyceride levels will change in response to certain food consumption and physical activity. In general terms, the reference model is a map of food and activity groupsto metabolic tolerance scores that can be used to predict the person's triglyceride levels in a given situation. The personal metabolism reference model for the individual is his unique tolerance scores as were measured during a diet and activity plan that is aimed at evaluating his metabolism. Two individuals, consuming the same diet, and generally having the same level of activity, may have very different responses to the same food intake. Some individuals, through their unique physiology are more tolerant to carbohydrate rich foods for example, whereas others may be far more sensitive/intolerant. As such, simply going on a "low carb" diet may be unnecessarily strict for some, and insufficient for others to effectuate weight loss or other desirable outcomes. The tolerance scores are measured for different situations and points of times during the day. For example, two individuals, consuming the same diet, and generally having the same level of activity, may have very different responses to high-carb food during the morning hours or on an empty stomach while their carbohydrates tolerance score may be similar if the carbs are consumed during a balanced lunch.
An important input for the method for generating the metabolism reference model is the type and quality of food consumed. For practical reasons the method first generates groups of food by amount, density, type and quality of fats, carbohydrates, and protein that are typical for food in the group. The general usage of the food groups is that they are later mapped to metabolic tolerance scores for each individual. In general the groups are determined so they cover the most common foods and so that they represent the various nutritional components that are known to have an effect on triglyceride levels. The food groups also cover various serving sizes as the tolerance to different serving size is not always linear. The food groups may include among others:
Small, medium and big servings of food with high density of simple sugars.
Small, medium and big servings of food with high density of complex carbs.
Small, medium and big servings of food with high density of proteins.
Small, medium and big servings of food with high density of saturated or transfats.
Small, medium and big servings of food with high density of saturated or transfats and simple sugars.
Small, medium and big servings of food with high density of saturated or transfats and proteins. Small, medium, and big servings of food with high density of good fats (Omega 3/6/9).
Small, medium, and big servings of food with high density of good fats (Omega 3/6/9) and complex carbs.
The algorithm also uses groups of physical activities by duration, type, frequency and intensity. The general usage of the activity groups is that they are later mapped to metabolic tolerance scores for each individual. In general the groups are determined so they cover the most common activities and so that they represent the various parameters that are known to have an effect on triglyceride levels. The activity groups may include among others:
10/30/60/180 min of high-med-low intensity of aerobic physical activity.
10/30/60/180 min of high-med-low intensity of strength physical activity.
Active day with around 10,000 or more steps or an equivalent aerobic activity.
Very low activity day with 1000 or less steps or an equivalent aerobic activity.
For each of the groups of food and activity triglyceride metabolic tolerance scores_are defined as a function of the delta in triglyceride levels in certain points-of-time after eating the food or performing the activity in a given situation. The tolerance scores are used for used for predicting the response of the user's triglyceride levels to various food consumption and performed activities - if for example the algorithm identifies high level of triglyceride 1-2 hours after a meal, he will use the tolerance scores to evaluate the impact of the imbalance on triglyceride in 5 hours. In one possible variation of the algorithm the tolerance scores are average normalized values of the delta in triglyceride levels that were measured, 1 hour, 2 hours and 5 hours after eating/performing an activity. To facilitate the discussion, Figure 1 illustrates 2 measures of triglyceride levels after eating a food from group A, for example after eating a medium breakfast with high density of complex carbs and protein. For group A, the method will determine 3 tolerance scores: the short term change tolerance score is a normalized value of the delta in triglyceride levels between points (fla) and (flf) in fig 1. The gap is illustrated in point (fie) in the graph and can be normalized to the value 0.16 by dividing the average gap by 300 (fixed number used for normalization). The short term change tolerance score is used in later embodiments for estimating the immediate effect of certain foods in order to generate a diet plan that is less prone to peaks and falls in energy levels. The max tolerance score is a normalized value of the average delta between the levels at the time of eating (fla) and the max change in triglyceride levels as illustrated in point (fib) in fig 1. The gap is illustrated in point (flc) in the graph and can be normalized to the value 0.32 by dividing the average gap by 300 (this is one possible normalization). The triglyceride clearance tolerance score is a normalized value of the average delta between the max levels of triglyceride(flb) and the change in triglyceride levels after 5 hours as illustrated in point (fid) in fig 1. The gap is illustrated in point (fid) in the graph and can be normalized to the value 0.16 by dividing the average gap by 300. The triglyceride clearance tolerance score is used in later embodiments in order to provide a predictor to the levels of triglyceride in the case the algorithm identified imbalance in the levels. A weighted tolerance score can be determined by weighting these three tolerance scores among others, depending on the purpose and use of the reference model. In a similar way the triglyceride changes are measured and tolerance scores are defined for physical activities.
The personal metabolism reference model for an individual is determined during a several days of pre-defined diet and activity program - during this period the user follows a subscribed diet plan and the system frequently measures how his triglyceride levels are affected by certain food intake and performed activities. The plan is determined as part of a diet program in which the user 'learns' how his body responds to different food and activity types. The user is guided to consume foods in certain groups and his triglyceride levels are measured continuously and frequently every hour in order to generate his personal reference model. The plan is designed in relation to an earlier knowledge base of tolerance scores for similar people, so the user don't need to eat all types of food and performall modeled activities in all situations. For example, the plan may include the following guidance in order to evaluate the user's tolerance to high-density carbohydrates:
The user is guided to eat a breakfast with high density of carbs after a 12 hour fast.
The user is guided to follow a 3 days of very restricted carbohydrates diet.
The user is guided to eat high density of carbs immediately after a low-carbs lunch and breakfast. The user is guided to eat a high density of carbs at the evening after medium intensity physical activity.
These 4 measurements may be sufficient in order to evaluate the tolerance of the user for several groups of carb-rich foods in different situations, by associating the user to cohorts of similar users with similar measurements, and applying regression techniques in order to estimate the user's tolerance to groups of foods and activities for which there are missing measurements for him.
Timing of food consumption and activity is important for evaluating the personal metabolism reference model for an individual. Physical activity may impact how a person responds to different foods and how different foods are consumed by the metabolic system. For example, the effect of a high-carb serving 1-2 hour before a medium-intensity aerobic activity is different comparing with the same food consumed without the following physical activity. Similarly, the effect of a big serving of protein with high fat density may differ after an intensive physical activity session. In order for the model to cover these variations the system may provide the following guidelines and generaterespective tolerance scores:
Eat medium serving of medium-density simple carbohydrates 1-2 hours before an aerobic activity session of over 30 min.
Eat a protein/carbs rich meal 1-2 hours after an aerobic/strength activity session.
These measurements and possibly others are used for determining two tolerance scores that enable later methods to estimate the effect of physical activity on the metabolism of sugars and rich-proteins. METHOD FOR GENERATING A PERSONAL DIET REFERENCE MODEL
The present subject matter provides a method for generating a personal diet reference model for an individual using detailed, frequent and ongoing records of his activity, triglyceride levels and weight over time. The model can be used to determine the range of required triglyceride levels in every point of time during the day for an individual to reach his weight management goals. Common weight management goals for this purpose are:
Quick weight loss - 3-5 pounds a week.
Moderate weight loss - 1-2 pounds a week.
Sustaining weight - up to 1 pound change a year.
Moderate weight gain - 1-2 pounds a week. The graph in figure 2 illustrates a personal diet reference model for a user with the goal of moderate weight loss of 1-2 pounds a week. The target level (f2c) provides the expected median range of triglyceride levels during every category of time during the day, for example the target triglyceride levels for this person to reach his goals in category of time (f2f) is in the range of 175 and 200. The category of time of day is not fixed but is rather determined by the time of eating and sleeping for the individual as follows:
2-3 hours after breakfast (f2d).
2-3 hours after morning snack (f2e).
- 1 hour after lunch (f2f).
2-3 hours after lunch (f2g).
2-3 hours after post-lunch snack (f2h).
2- 3 hours after dinner (f2i).
3- 5 hours after dinner (f2j).
5-10 hours after dinner and before breakfast (f2k).
1- 2 hours before med-high intensity physical activity (not in the figure).
1 hour after med-high intensity physical activity (not in the figure).
2- 3 hours after med-high intensity physical activity (not in the figure).
The dashed lines above the reference target graph (f2a) illustrate the upper thresholds for each category of time during the day. These are used in later embodiments to trigger alerts and diet and activity guidance to help the user balance his diet.
The dashed lines below the reference target graph (f2b) illustrate the lower thresholds for each category of time during the day. These are used in later embodiments to trigger alerts to the user and diet and activity recommendations that will make sure the user is eating enough to achieve his weight management goals.
The method uses cohorts of people with similar physical properties and similar metabolism reference models in order to generate the diet reference model baseline. This baseline will be used in a later embodiment to generate a personal diet reference model for the user. The levels of triglyceride are measured for people in the cohort over several days for a predefined set of diet and activity plans and the triglyceride threshold levels range for various weight management goals is determined as a function of these measurements. The cohorts are determined using a machine learning process that find clusters of similar people and use physical features, behavioral features and mainly the metabolism tolerance scores as defined in earlier embodiments. The clustering algorithm may use the following physical and behavioral features: age, gender, height, body fat mass, body lean mass, thermic effect of eating certain foods in certain times of the day, physical activity duration during the week, physical activity frequency during the week, physical activity level during days when the user don't work out, changes in the weight during the last: week/month/quarter/year, duration of current diet plan, number of meals a day, category of time for main meal of the day. The main features for the similarity function are the metabolism reference model tolerance scores as defined in section 1 of the subject matter's detailed description. The machine learning process starts with a manually predefined set of cohorts, it uses measurements from many people of triglyceride levels in every category of time during the day (for a given diet and weight management goals) and converge the cohorts so they best reflect groups of people that responds in a similar way to a similar diet (and have reached the weight management goals).
Figure 3 illustrates the measurements of triglyceride levels for a cohort of similar people on a given diet during a week. For each person, the range of measured triglyceride levels during each category of time during the weekis marked similar to (f3a). The machine learning clustering algorithm will group the people with similar measurements into the cohort and the target diet threshold line will be determined as an average of these measurements (f3b). The upper thresholds and lower thresholds are determined as max/min values for the cohort, or they are set as a percentage of the target line depending on the purpose of reference model.
The personal diet reference model for the user is determined in a several days of diet and activity program in which the user is associated to one or more of the similarity cohorts. The user follows a representative diet plan and his triglyceride levels are measured continuously and frequently during a period of several weeks. Following these several weeks the user main features, triglyceride levels, behavioral features and weight change measurements are compared with those of the different cohorts and the user is associated to the cohort of highest similarity. The personal diet reference model for the user is now defined to be the one for the cohort of similar people. The model may change and needs to be updated every several months due to various reasons such as: User does not reach his weight goals while following the recommended diet; User changes his behavior, body weight, fitness; user changes his weight management goals; user health requires change in recommended diet; user feedback on energy levels is negative. III. METHOD FOR GENERATING A PERSONAL DIET REFERENCE MODEL
The present subject matter provides a method for generating diet and activity recommendations for each individual to help him reach his weight management goals using his personal metabolism reference model, his personal diet reference model, his weight management goals, his recent records of activity, triglyceride levels and body weight, his profile, his food preferences and his feedback. The recommendations may be presented to the user using a software application, website or in any other form, but in general they require a system capable of providing alerts, getting feedback from the user and frequent interaction with the system that implements the embodied algorithm.
The system generates weekly diet and activity plan that provides a framework for the user's diet. The user's personal diet reference model, his weight management goals, his food and activity preferences, his recent tolerance levels and his recent weight changes are used to define a diet framework for the user's weekly plan. The diet framework determines the size, composite, and time of meals and activities in the plan for the individual. Figure 4 illustrates the steps for building a weekly diet plan for a user. The algorithm uses the user's weight management goals (f4a) and recent weight changes (f4b) to determine a goals framework for the weekly plan (f4c). The goal framework will determine how aggressive is the diet framework and which triglyceride target thresholds should be chosen from the user's diet reference model - if the user's goal for example is rapid weight loss of 3-4 pounds a week a lower threshold for triglyceride levels will be selected comparing with a similar user with a moderate weight loss goal of 1-2 pounds a week. The recent weight changes may also determine the aggressiveness of the framework, so if a user is not reaching his goals in recent weeks a more aggressive diet framework will be selected. The goals framework contains two metrics: the level of aggressiveness and the goal scope, these are indexes for selection of the appropriate triglyceride thresholds from the user's diet reference model (f4e). First, a general triglyceride threshold will be selected, then the food preferences (f4d) will determine the mixture of food types and meals for the food framework (f4g). Recent tolerance changes (f4f) are now taken into account, and in general they boost or limit the amounts of carbs, fats and proteins in the food framework if the metabolic tolerance for these components increased in recent evaluations of the user's metabolism reference model. In the next stage the user's activity preferences and recent fitness shape as measured by the system and are used to a) determine the recommended physical activity framework and b) to increase or decrease the food consumption thresholds and match them to the level of expected physical activity.
Figure 5 illustrates the process of providing food recommendations from the weekly diet framework. The user's weekly diet framework (f5a in figure 5) is used to provide food recommendations for the weekly plan. The specific food recommendations is determined using the user's metabolism reference model and food and activity preferences, using a knowledge base that maps food plans to different metabolism models. The diet framework (f5a) provides for each meal the range of triglyceride levels that is expected to bring the user to his diet management goals and the thresholds of fats, carbs and proteins in the meal. In the second phase possible food alternatives are fetched from the nutritional knowledge base (f5b) and are filtered using the user's food preferences (f5b). In the next step of the algorithm the selected dishes are further filtered using the personal metabolism reference model, in this stage the serving size and dish components may be switched due to the user's metabolic tolerance to the food in the dish. These steps produce possible diet and activity menus (f5e) - these are presented to the user that can choose and switch dishes to suit his preferences.
The subject matter provides a method for generating daily diet and activity guidance using repeated, frequent and ongoing tracking of physical activity and triglyceride levels. The implementing system is expected to monitor the user's triglyceride levels and physical activity levels and frequently compare these with the thresholds defined in the user's weekly diet framework - those selected from his personal diet reference model and define upper threshold to trigger the system. In case the user's triglyceride levels or activity are not within the weekly plan's framework the system generates diet and activity guidance that are aimed to keep the user's diet balanced and get his triglyceride levels to a range expected to help him reach his weight management goals. The daily guidance may include the following guidelines:
The user over-consumed food on a certain day, and the algorithm suggests a variation of weekly plan that is designed to balance the triglyceride levels and balance the diet.
The user over-consumed simple sugars, and the algorithm suggests dishes and activity to balance the user's triglyceride levels.
The user skipped a meal, and the algorithm suggests supplements to the next plan meal to balance the triglyceride levels.
The algorithm uses frequent measurements of user's triglyceride levels, together with the user's personal diet reference model and his personal metabolism reference model to evaluate the imbalance level and to generate appropriate diet and activity guidance. Figure 6 illustrates the process of providing daily food and activity recommendations in the case of triglyceride imbalance. The algorithm identifies the user's triglyceride levels are out of the thresholds defined in the weekly diet framework (f6a), this process should be triggered around 1-2 hours after the misbalancingmeal/activity. The algorithm then uses the user's personal diet and metabolism reference models (f6b) to evaluate the imbalance and estimate the required change in the weekly plan in order to balance the user's triglyceride levels. The diet reference model provides a triglyceride prediction graph such as the one illustrated in Figure 2. The algorithm measures the unordinary levels of triglyceride levels 1 hour and 2 hours after the food consumption. This enables it to predict the effect of the imbalance using the clearance tolerance score from the metabolism reference model of the user. The target gap is defined within 2-3 hours from the food consumption by adding the predicted change in triglyceride levels to the planned diet and finding the delta to the weekly plan. Once the gap is determined the algorithm selects foods and activities to close it, this may result in recommendations to change the next several meals and add physical activity to the plan. In the second phase possible food alternatives (f6e) for balancing the diet are fetched from the nutritional knowledge base (f6d) and are filtered using the user's food preferences (f6c). In the next step of the algorithm the selected dishes are further filtered using the personal metabolism reference model (f6f), in this stage the serving size and dish components may be switched due to the user tolerance to the food in the dish. These steps produce possible diet and activity recommendations (f6g) - these are presented to the user that can choose and switch dishes and activities to suit his preferences.