EP3844767A1 - Utilisation de mesures à domicile pour prédire un état clinique et améliorer la précision de mesures/prédictions à domicile en utilisant des données associées au rythme circadien et aux horaires des repas - Google Patents

Utilisation de mesures à domicile pour prédire un état clinique et améliorer la précision de mesures/prédictions à domicile en utilisant des données associées au rythme circadien et aux horaires des repas

Info

Publication number
EP3844767A1
EP3844767A1 EP19759372.6A EP19759372A EP3844767A1 EP 3844767 A1 EP3844767 A1 EP 3844767A1 EP 19759372 A EP19759372 A EP 19759372A EP 3844767 A1 EP3844767 A1 EP 3844767A1
Authority
EP
European Patent Office
Prior art keywords
data
individual
food
meal
home
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
EP19759372.6A
Other languages
German (de)
English (en)
Inventor
Jonathan Thomas WOLF
George HADJIGEORGIOU
Richard James Davies
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zoe Ltd
Original Assignee
Zoe Global Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US16/120,039 external-priority patent/US20200066181A1/en
Priority claimed from US16/272,865 external-priority patent/US11915151B2/en
Priority claimed from US16/434,135 external-priority patent/US11295860B2/en
Application filed by Zoe Global Ltd filed Critical Zoe Global Ltd
Publication of EP3844767A1 publication Critical patent/EP3844767A1/fr
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/60—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to nutrition control, e.g. diets
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
    • G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
    • G01N33/92—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing involving lipids, e.g. cholesterol, lipoproteins, or their receptors

Definitions

  • At home measurements may be used to predict clinical state and the accuracy of the at home measurements/predictions may be improved using data associated with the timing of meals.
  • standardized meals and digital devices may in various embodiments be utilized outside a clinical setting to assist in determining/predicting one or more clinical states associated with one or more individuals, or predicting risk of a disease risk/health outcome (e.g. % risk of cardiovascular disease over the next decade).
  • This second meal may provide discrimination to better determine an individual’s postprandial responses, such as the response of many different types of lipids to a meal.
  • An individual’s postprandial responses may not be well differentiated from average responses when obtained with a single meal eaten outside of the clinic which is an environment with lower levels of accuracy.
  • various embodiments may advantageously improve the accuracy of measured biomarkers and the at home measurements/predictions by using data associated with circadian rhythms and/or meal timing data.
  • the invention relates to a method comprising: receiving data associated with performance of one or more tests in a non-clinical setting, wherein the one or more tests are associated with an identification of one or more nutritional responses.
  • the method further comprises causing performance of one or more of the following actions: confirming the data; adjusting at least a portion of the data; adjusting a weighting associated with a machine learning mechanism; or generating one or more of a prediction of a clinical state or a health outcome.
  • receiving the data comprises receiving food data comprising first food data and second food data associated with the individual in an at-home setting, wherein the first food data indicates one or more first foods consumed by the individual to evoke a first nutritional response and the second food data indicates one or more second foods consumed by the individual to evoke a second nutritional response, wherein the one or more second foods are consumed by the individual at a predetermined amount of time from a time the one or more first foods were consumed by the individual.
  • the method further comprises determining a first postprandial response associated with the first food data and determining a second postprandial response associated with the second food data, wherein the first postprandial response and the second postprandial response are associated with lipid responses.
  • the predetermined time is at least four hours.
  • the one or more second foods comprise a higher percentage of carbohydrates compared to the one or more first foods.
  • the method further comprises determining a first postprandial response associated with the first food data and determining a second postprandial response associated with the second food data, wherein the first postprandial response and the second postprandial response are associated with glucose responses.
  • generating the one or more of the prediction of the clinical state or the health outcome comprises predicting remnant lipoproteins (RLP).
  • the method further comprises generating one or more of a quality score or a classification, comprising performing one or more of: generating a food data score that provides an indication of a quality of the food data received; generating a biome data score that provides an indication of a quality of biome data associated with the individual; or generating a device test data score that provides an indication of a quality of device test data captured by an electronic device.
  • the method further comprises generating one or more of a recommendation or a prediction using at least a portion of the data.
  • the invention relates to a system, comprising a data ingestion service, including one or more processors, configured to receive test data associated with performance of one or more tests in a non-clinical setting, wherein the one or more tests is associated with an identification of one or more nutritional responses.
  • the data ingestion service is further configured to cause, based, at least in part, on the accuracy of the test data, one or more actions to be performed, the actions including to: confirm the test data; generate an indication of quality for the test data; adjust at least a portion of the test data; or adjust a machine learning mechanism.
  • receiving the test data comprises receiving food data and biome date, wherein the food data comprises first food data and second food data associated with an individual in an at-home setting, wherein the first food data indicates one or more first foods consumed by the individual to evoke a first nutritional response and the second food data indicates one or more second foods consumed by the individual to evoke a second nutritional response, wherein the one or more second foods are consumed by the individual at least four hours from a time the one or more first foods were consumed by the individual.
  • the one or more first foods are different from the one or more second foods.
  • the method further comprises determining a first postprandial response associated with the first food data and determining a second postprandial response associated with the second food data, wherein the first postprandial response and the second postprandial response are associated with one or more of lipid responses or glucose responses.
  • the actions to be performed includes generating one or more of a prediction of a clinical state or a health outcome, comprising predicting remnant lipoproteins (RLP).
  • the actions to be performed includes generating one or more of a quality score or a classification, comprising performing one or more of: generating a food data score that provides an indication of a quality of the food data received; generating a biome data score that provides an indication of a quality of the biome data; or generating a device test data score that provides an indication of a quality of device test data captured by an electronic device.
  • system is arranged to carry out any one of the embodiments of the method described above, or combinations thereof.
  • FIG. 1 shows the rise of postprandial triglycerides after a single high fat meal measured at home.
  • FIG. 2 shows the rise of postprandial triglycerides measured at several time points.
  • FIG. 3 shows the rise of postprandial triglycerides from a two meal protocol conducted at home
  • FIG. 4 shows an accuracy of a two meal protocol conducted at home vs in clinic.
  • FIG. 5 shows a repeatability of a one meal protocol conducted at home on different days.
  • FIGs. 6A, 6B, 6C, and 6D show predicting remnant Lipoprotein levels with and without postprandial information.
  • FIGs. 7A and 7B show predicting Glyca (GP) levels with and without postprandial information.
  • FIGs. 8A, 8B, and 8C show significant variation in how individuals respond to the same meal at breakfast and lunch in at home conditions.
  • FIGs. 9A and 9B illustrate example results of two machine learning models: a regressor and a classifier approach in predicting the triglycerides levels at six hours.
  • FIGs. 10A and 10B illustrate example results of two machine learning models: a regressor and a classifier approach in predicting the glucose baseline levels.
  • FIG. 11 is a block diagram depicting an illustrative operating environment in which at home measurements are used to predict clinical state and the accuracy of the at home measurements/predictions can be improved using data associated with circadian rhythms.
  • FIG. 12 is a flow diagram showing a routine illustrating aspects of a mechanism disclosed herein for generating predictions of clinical state/health outcomes using data obtained in a non-clinical setting.
  • FIG. 13 is a flow diagram showing a routine illustrating aspects of a mechanism disclosed herein for utilizing data associated with circadian rhythms for improving the accuracy of at home measurements/predictions.
  • FIG. 14 is a block diagram depicting an illustrative operating environment in which a data ingestion service receives, and processes data associated with at home measurements of responses.
  • FIG. 15 is a computer architecture diagram showing one illustrative computer hardware architecture for implementing a computing device that might be utilized to implement aspects of the various examples presented herein.
  • standardized meals and digital devices can be utilized outside a clinical setting to assist in determining/predicting one or more clinical states associated with one or more individuals.
  • a clinical state can include but is not limited to a state such as a diabetic state, a prediabetic state or one or more metabolic syndrome states, and the like.
  • the standardized meals and digital devices can also be utilized to predict risk of a disease risk/health outcome (e.g. % risk of cardiovascular disease over the next decade). As will be appreciated, these are just a few of the examples.
  • nutritional response measurements are obtained for two fat meals instead of a single fat meal.
  • the second fat meal can be eaten at some period of time (e.g., a few hours) after the first fat meal.
  • the use of two meals results in a higher variation of response compared to a single fat meal ln some instances, the second meal may not have fat in it.
  • This second meal provides discrimination to better determine an individual’s postprandial responses, such as the response of many different types of lipids to a meal.
  • An individual’s postprandial responses may not be well differentiated from average responses when obtained with a single meal eaten outside of the clinic which is an environment with lower levels of accuracy. This discrimination outside the clinic was uncertain until more than one fat meal was utilized ln some examples, using a single meal at home and blood collection methods such as dried blood spots the results between individuals cannot be significantly differentiated.
  • F1G. 1 shows that when a postprandial triglycerides measurement is done in non-clinical conditions with a single fat meal the human body does not respond as was anticipated, and the responses are highly clustered with a small average response. As a result, this sort of single meal protocol may be less useful than anticipated to measure postprandial responses for lipids or similar biomarkers, especially as at home measurement are likely to be less accurate than in the clinic. According to some examples, in order to obtain a high variation of response at home, the individual consumes a second meal.
  • FIG. 1 shows the rise of postprandial triglycerides, measured at lh and 4h after a breakfast consisting of a standardized muffin containing 40.5g of carbs and 34.8 of fat consumed at home. At 4 hours the average rise is 0.41 with a standard deviation of 0.70. This measurement was performed on 259 individuals of which only 213 (-82%) had a triglycerides at 4h higher than at fasting.
  • FIG. 2 shows a two meal protocol measured in the clinic. This demonstrates that a second meal generates a high variation of lipid response in a clinic. Clinical tests have previously shown that a second meal may help to rapidly push up lipid levels in the blood if there was previously a fat meal, possibly by pushing lipids out of the lining of the gut and into the blood. It was unknown if this would lead to a high variation in individual responses outside the clinic to an extent large enough to be measurable using at home technologies.
  • FIG. 2 shows the rise of postprandial triglycerides measured at several time points. Breakfast was consumed at Oh and at 4h lunch was consumed. At 4h the average rise is 0.49 with a standard deviation of 0.80 while at 5h the average rise is 1.09 with a standard deviation of 1.04. At 5h the rise is approximately double that at 4h and with a larger variation in the population. This measurement was performed on 598 individuals of which 549 (-91%) had a triglycerides at 5h higher than at fasting
  • FIG. 3 demonstrates that the use of a second meal can generate a high variation in individual lipid responses outside the clinic to an extent large enough to be measurable using at home technologies. It was also unknown if these measurements would be repeatable or chance:
  • FIG. 3 shows the rise of postprandial triglycerides measured at 5h and 6h after a breakfast and lunch were consumed at Oh and 4h respectively at home.
  • the mean rise from fasting levels is 0.98 and 0.57 respectively, with standard deviations of 0.93 and 0.88.
  • the meals consisted of breakfast with 28.2g of carbs and 39.3g of fat and a lunch with 7l .2g carbs and 22.2g of fat. This protocol was carried out 615 times (often in duplicate per individual). On 597 occasions (-97%) the triglycerides at 5h were higher than at fasting.
  • FIG. 4 demonstrates that there is a good correlation between at home measurement of variation in lipid response and that seen for exactly the same individual eating a high fat meal followed by a second meal in the controlled clinic setting.
  • FIG. 4 shows the venous measurement of triglycerides in the hospital environment vs an at home blood collection measurement (dried blood spots in this particular example) of triglycerides at home for the same individual.
  • the measurements are taken on separate days. Both measurements were taken at 5 hours after the participants had started the protocol consisting of a fat breakfast at 0 hours and then lunch at 4 hours.
  • This figure demonstrates that there is a good correlation between these two measures, and therefore that a two meal protocol at home gives repeatable data about triglycerides responses and is not overwhelmed by noise.
  • FIG. 5 demonstrates that the repeatability of a one meal protocol conducted at home on different days is much weaker than if a two meal protocol is used. This information allows us to use the data collected at home after eating a meal or series of meals to understand individual variation in lipid responses and therefore improve the accuracy of models that are used to predict clinical states and/or to predict individual responses to food such that guidance can be given on what is the right food to minimize health risks or improve an existing condition such as diabetes or cardiovascular disease.
  • FIG. 5 shows the triglycerides rise measured with at home blood collection measurement (dried blood spots in this particular example) 4 hours after breakfast on 2 separate days. A very weak correlation between the 2 measurements can be seen. This is contrast with FIG. 4 where a strong correlation was found between the clinic measurements and the home measurements of the triglycerides when a 2 meal protocol is followed.
  • FIGs. 6A-6D also demonstrates that by having a person undertake a high fat test meal at home and measuring triglycerides levels post-prandially a very high quality prediction of the levels of RLP can be made. By taking at-home post-prandial measurements it is possible to improve on a model built on fasting bloods and therefore improve the guidance given to individuals.
  • measuring post-prandial measures like triglycerides at home will be cheaper than measuring RLP directly.
  • the person can undertake a high fat meal test at home and one can measure RLP levels directly for example by collecting liquid blood using an at-home collection device and having it assayed using NMR techniques. Given the role of RLP in inflammation and its impact on long-term health risks, by adding this ability to predict or directly measure postprandial RLP responses it is possible to improve prediction of current health states and future health risks.
  • the Lipid responses can be used to predict food responses, but also to measure risks of CVD, fatty liver disease, and a whole range of metabolic disease including diabetes.
  • Measuring post-prandial lipids is just one example of an at home measurement. There are many other examples of at home measurements such as using a CGM with one or a series of different test meals, or measuring the response to an OGTT (Oral Glucose Tolerance Test) in addition to the lipid test described above. In other examples, one would combine blood taken at fasting (sometimes called a baseline blood) as is often done for medical tests alongside some sort of measurement of a post-prandial response, which could be a glucose response, a lipid response, or something else.
  • OGTT Oral Glucose Tolerance Test
  • FIGs. 6A-6D compare the prediction performance of two models that predict the remnant lipoproteins (RLP) in blood 6 hours after a fat meal (protocol as in FIG. 2). By adding postprandial responses into the model the accuracy of the predictions of RLP are significantly improved. Similar results were also observed with predicting RLP at 4 hours after a meal. This data is on 486 individuals undergoing the same meal as in FIG. 2. RLP was measured in all these individuals.
  • the first model predicts the remnant lipoproteins using an extensive number of predictors including fasting blood measurements and clinic measurements illustrated in the table of FIG. 6A.
  • the second model adds a small number of postpandrial triglycerides responses shown in the table of FIG. 6A.
  • linear regression with log-scaled standardized inputs to predict each of the remnant lipoproteins at 6 hours was utilized. Predictions were compared to the measured values in terms of coefficient of determination (Rsquared).
  • FIG. 6B show the differences in prediction performance between these two models.
  • scatter plots display the predicted vs measured of the remnant lipoproteins for the first model - without postpandrial lipemic responses.
  • Plots on the second row show the corresponding predictions for the second model - with postpandrial lipemic responses. From these scatter plots and their associated R-squared values, it can be concluded that the model with postpandrial response has a significantly stronger association with the remnant lipoproteins than the models without.
  • FIG. 6C illustrates which fasting blood measurements have the highest contribution to the prediction of remnant lipoproteins at 6 hours.
  • fasting triglycerides contributes more than 50% to explaining the concentration of extremely large VLDL and cholesterol carried by these lipoproteins and more than 60% in very large VLDL.
  • Cholesterol measures explain less than 25% of the RDL. It can be seen that the coefficient of determination (Rsquared) for each single fasting blood measurement to predict the remnant lipoproteins.
  • Rsquared coefficient of determination
  • GlycA protein Glycosylation
  • FIGs. 7A and 7B show that by having a person undertake a high fat meal test at home and measuring triglycerides levels post-prandially it may be possible to generate a significantly better prediction of the levels of GlycA compared to using fasting levels alone. Given the role of GlycA in inflammation and its impact on long-term health risks, by adding this ability to predict or measure postprandial GlycA responses it is possible to improve prediction of current health states and future health risks.
  • FIGs. 7A and 7B follow the same approach as described with reference to FIGs. 6A-6D to predict the level of glycoprotein acetylation, an inflammation marker that has been shown to be predictive of cardiovascular disease (CVD), heart attacks and type 2 diabetes. This protein has also been measured at 6 hours after the first meal.
  • CVD cardiovascular disease
  • FIG. 7B the lower chart 720 illustrated in FIG. 7B with the addition of post-prandial triglycerides responses shows a significantly better association than the higher chart 710 in FIG. 7A which has all the baseline measures but no post-prandial responses
  • the diagnostic can then combine data from multiple measurements and use machine learning to correlate these with clinical outcomes to improve prediction of one or more clinical states or to predict risk of a disease risk/health outcome (e.g. % risk of cardiovascular disease over the next decade).
  • a disease risk/health outcome e.g. % risk of cardiovascular disease over the next decade.
  • a very simple example is getting a traditional baseline blood assay from blood collected at home which might give readings for say 10+ different biomarkers (triglycerides, glucose, cholesterol, etc.) plus one dynamic measure (e.g., a post-prandial glucose reading after a set meal).
  • This data can then be combined with readings of a large number of other people and provided to a machine learning mechanism, or utilizing some other technique, to generate a prediction based on the aggregated data rather than just using a single biomarker like cholesterol or HbAlc as is commonly used today as a diagnostic.
  • the data includes other information about the individual, such as racial background, sex, weight, current medication, etc. then given enough people in the dataset it would be possible to generate a diagnostic that is more accurate for people with particular backgrounds, such as those with African-American heritage, or using particular drugs. In some cases, this diagnostic would predict a particular clinical state is likely, while in others it will be a prediction of long-term health risk for one or a series of health outcomes.
  • the test dataset for the model includes participants with that disease state, or who go on to be identified after the event as having that disease state.
  • participants with that disease state or who go on to be identified after the event as having that disease state.
  • the model By training the model with those individuals, who have undergone some or all of the tests that are carried out on the individual, it increases the accuracy of the model in identifying those disease states compared to traditional approaches that often use a single biomarker to identify a population at risk. In some configurations this allows a model to predict a current disease state based on at home tests without needing to use the same tests as are carried out in the clinic.
  • the test dataset for the model includes individuals with known disease risk scores calculated from currently validated algorithms. In some of these configurations, which may rely on data that cannot be collected on all individuals. In some configurations in order to predict disease risk there are individuals who are seen to develop the disease at a later time from the period where their measurements are taken. By training the model with those individuals, who have undergone some or all of the tests that are carried out on the individual, it increases the accuracy of the model in identifying those disease states compared to traditional approaches that often use a single biomarker to identify a population at risk.
  • Lipid response data, and glucose response data obtained in some experiments provide evidence of this use of the data in this uncertain science area.
  • the at home diagnostic uses two meals to diagnose a health outcome. This has been found to be useful for lipids of which there are very many - triglycerides is the aggregate name for a number of lipids.
  • the data from at home users can be collected over time thereby building a large data set that can be used in the generation of predictions and determination of one or more clinical states, or health risk outcomes.
  • a machine learning mechanism can utilize a plurality of different biomarkers to generate the predictions of clinical state/health risk outcomes.
  • the data from one user can be compared to data obtained from other users.
  • data associated with a circadian rhythm e.g., sleep times, awake times
  • the measured biomarkers can include many different types of health data such as microbiome data which may be referred to herein as“biome data”, blood data, glucose data, lipid data, nutrition data, wearable data, genetic data, biometric data, questionnaire data, psychological data (e.g., hunger, sleep quality, mood, ...), objective health data (e.g., age, sex, height, weight, medical history, ...), as well as other types of data.
  • “health data” can refer to any psychological, subjective and/or objective data that relates to and is associated with one or more individuals. The health data might be obtained through testing, self-reporting, and the like.
  • Some biomarkers change in response to eating food, such as blood glucose, insulin, c-peptides and triglycerides and their lipoprotein components.
  • biomarkers caused by eating food such as responses to one or more standardized meals (“post-prandial responses”) may be measured.
  • post-prandial responses responses to one or more standardized meals
  • systems can be utilized to analyze the data, to adjust the data.
  • the systems can include computing devices (remote and/or local) and/or individuals authorized to analyze and adjust the data.
  • the user can be given at least two set meals.
  • the first meal is a breakfast after fasting and the second meal is several hours later.
  • the first of these meals contain fat, and the second may also contain fat.
  • the at-home measurement process can involve one measurement at fasting before eating the breakfast and at least one measurement after the second meal. It is possible to take multiple measurements (e.g. one hour and two hours after the lunch) to better characterize the response curve of the individual’s lipids to this“challenge meal”.
  • Dried blood spots are one mechanism that can be used to measure these lipids at home, but any blood collection method (or other method to reliably measure lipids) could be used.
  • a device such as a smartphone, can be used to remind the user to eat the lunch at precisely the right time after the breakfast and to remind them to measure their lipids at the appropriate times.
  • users can consume multiple set meals to measure one or more other biomarkers which is relatively slow moving (i.e. does not return to baseline within 2-3 hours) and where the second meal accentuates the response underway from the first meal by increasing the overall stress on the person’s biological systems.
  • biological systems are designed to maintain homeostasis, it is attempted to try and disturb these systems in order to measure how rapidly the biological systems can return to their starting equilibrium and how far away the biological systems move from this equilibrium under a given pressure. It turns out that a second meal appears to be very effective at differentiating people who easily return to equilibrium and those whose systems are much more disturbed before this happens.
  • Inflammation markers such as IL-6 are an example, where responses are even slower to return to baseline than triglycerides.
  • the meals are set meals with known levels of nutrients so that responses can be accurately compared between individuals because they are eating the same meals with the same timings between the different meals.
  • Meals are eaten rapidly (e.g. within 10 or 15 minutes) to help ensure nutrients arrive at approximately the same times across individuals.
  • the blood spots can be taken at fasting and 1 and 2 hours after the second meal (or some other timing). If only two measurements are taken, then fasting and a timepoint somewhere between 1 and 2 hours after the second meal that is the same for all users.
  • circadian rhythm refers to the 24-hour cycle that tells our bodies when to sleep, rise, and eat.
  • circadian data is measured/monitored in a non-clinical setting.
  • the circadian data can include but is not limited to sleep data (e.g., sleep times, awake times), Heart Rate (“HR”) data, and the like.
  • the circadian data can be measured by at home electronic data collection devices, and/or determined using other mechanisms. Today, these electronic data collection devices can be used to accurately measure sleep/wake times, HR, and the like for an individual.
  • Chrono nutrition generally refers to the timing of meals in view of the differing responses to the same meal at different times, which may be influenced by the circadian rhythm associated with an individual, the amount of sleep they have had, when they woke up, how long since they last ate, etc.
  • a disruption in circadian rhythm has been associated with metabolic syndrome, obesity, cardiovascular disease, and cancer. Alterations in timing of meals may also increase the risk of metabolic diseases.
  • FIG. 8 demonstrates that there is a significant variation in how individuals respond to the same meal at breakfast and lunch, with most individuals having a much larger glucose response to the meal at lunch after having the same meal at breakfast. Furthermore, the ratio between these two meals is not constant, demonstrating individual variation between these relationships. This variation can be predicted in order to provide individual guidance on the impact of meal timing and circadian rhythm on biological responses to food.
  • FIGs. 8A, 8B, and 8C illustrate data associated with 460 participants that ate a standardized muffin meal for breakfast and the same meal again at lunchtime. Each meal contained 71 g Carbohydrate, 22 of Fat, and lOg of Protein. The meals were then repeated on another day. Their glucose response (incremental Area Under the curve) varied as illustrated. Chart 810 of F1G. 8 A shows a meal response for breakfast and lunch. F1G 8B shows a table 815 for breakfast and lunch. Chart 820 shows a breakfast/lunch response.
  • a data service can utilize the circadian data (e.g., sleep data) and/or meal timing data about a user and/or many different users in determining the impact of the body clock, sleep duration and quality and the timing of previous meals on how a user will respond to food. For instance, when a user is determined to have disrupted their sleep cycle, a data service can perform an action (e.g., adjust the data) to adjust for the disruption in the circadian rhythm ln other examples, the data service can adjust the data for one or more measured biomarkers based on the circadian data (e.g., the time of day, how much sleep the user had, ).
  • the circadian data e.g., sleep data
  • meal timing data e.g., meal timing data about a user and/or many different users in determining the impact of the body clock, sleep duration and quality and the timing of previous meals on how a user will respond to food. For instance, when a user is determined to have disrupted their sleep cycle, a data service can perform an action (
  • the data service can also utilize a data analyst, a machine learning, and/or some other mechanism to adjust the as well as providing the data to one or more other services, for predictions, recommendations, and the like.
  • circadian data and/or meal timing data is provided to one or more machine learning mechanisms. As a result, the predictions of biomarker responses to a meal can respond to the time of day, the user’s normal circadian rhythm and previous meals so as to better predict the actual response that will be seen to a meal.
  • a data analyst viewing the received data may utilize different tools (electronic and/or non electronic) to interact with the timing data.
  • UIs such as graphical user interfaces (“GUIs”)
  • GUIs graphical user interfaces
  • stool collection and microbiome analysis can be utilized to predict post-prandial blood fat levels.
  • stool collection, metagenomics analysis, measurement of post-prandial responses can be utilized in an at-home and/or clinic setting (as covered herein and in the referenced applications) and machine learning that can take a stool collected at-home and use this to predict the responses of blood fat levels to different meals.
  • machine learning that can take a stool collected at-home and use this to predict the responses of blood fat levels to different meals.
  • gut microbiome which is not part of the human body but is a changing ecosystem of bacteria influenced by and influencing the host body
  • FIGs. 9A and B demonstrates that this data can be used to predict post-prandial blood fat levels with significant accuracy as demonstrated by the correlation coefficients, whereas FlGs. 10A and 10B indicates that this data does not work well for measuring glucose responses as demonstrated by the correlation coefficients despite some prior claims to the contrary.
  • the microbiome can be defined as the collection of microorganisms (bacteria, fungi, viruses, and micro-eukaryotes) that live in and on the human body.
  • microorganisms bacteria, fungi, viruses, and micro-eukaryotes
  • One method by which the human gut microbiome, in particular, can be studied is by collecting stool samples from which all the DNA material is extracted and sequenced using a high-throughput platform to get the genomes of the microorganisms. This technique is called metagenomics and allows one to assess which microbes are present in the sample, without cultivation intensive efforts, such as isolation of every single species present in a microbiome. Metagenomics, moreover, solves the problem of surveying of those microbes that cannot be cultivated and isolated in a laboratory.
  • the data from a metagenomics high-throughput sequencing is a large set of millions of stretches of short DNA fragments (reads) that all together are called a metagenome and are a representation of a microbiome sample.
  • a primary analysis of a metagenome is the taxonomic and abundance profiling, which aims to indicate which microorganisms are present and at what relative abundance.
  • taxonomic profiling There are several approaches available to perform taxonomic profiling, but the one minimizing the false positive ratio is a marker based one.
  • a marker-based profiler is highly accurate in determining which species are present in a metagenome because it exploits species-specific markers that can be pre-computed exploiting the large availability of genomes in public databases.
  • One approach is to use the taxonomic profile to predict human biomedical measures, and this, in turn, is showing that the microbiome can be used as a non-invasive way to estimate human health. This can be used to predict post-prandial responses of various metabolites in the blood such as triglycerides and its components (HDL, LDL, remnant lipoproteins, etc.).
  • a machine learning model is built using this data and/or other data described herein and in the referenced applications.
  • One approach is that the machine learning regressor has the aim of predicting the triglyceride level values of the people using the relative abundance of the microbiome species profiled, which can be done by ensuring these individuals are given a standardized meal that can differentiate their fat responses sufficiently for this to be clearly measured, as described herein and in the referenced applications.
  • the data was converted into two classes by dividing the dataset into low-triglyceride (the lower quartile of the distribution) and high- triglyceride (higher quartile of the distribution) classes. Then a machine learning classifier was employed to classify people based on their microbiome composition into the two defined classes. More classes may be utilized. Similarly, other scoring mechanisms can be utilized.
  • microbiome can be utilized to estimate also other biomedical relevant post-prandial measures for humans.
  • FIG. 10 The example of FIG. 10. demonstrates that microbiome cannot predict well all post-prandial responses, and that prior to carrying out this biological study it was not possible to know whether microbiome could be used to predict post-prandial lipid responses.
  • FIG. 9. illustrates example results of the two machine learning models: a regressor and a classifier approach in predicting the triglycerides levels at six hours.
  • the two panels in FIG. 9 report the results of the two machine learning models in predicting the triglycerides levels at six hours after breakfast. The breakfast meal is followed by lunch after two hours.
  • FIG. 10 illustrates example results of the two machine learning models: a regressor and a classifier approach in predicting the glucose baseline levels.
  • the two panels in FIG. 10 report the results of the two machine learning models in predicting the glucose baseline levels measured before breakfast.
  • A Result of the machine learning regressor model, with a median Pearson correlation coefficient of -0.009.
  • B Result of the machine learning classifier showing a median Area Under the Curve of the Receiver Operating Characteristic curve of 0.47 with a Matthews correlation coefficient of -0.063.
  • FIG. 11 is a block diagram depicting an illustrative operating environment 1100 in which at home measurements are used to predict clinical state and/or disease risks and the accuracy of the at home measurements/predictions are improved using data associated with circadian rhythms.
  • An individual may communicate with the environment 106 using a computing device 102 and possibly other computing devices, such as electronic data collection devices 103.
  • one or more electronic data collection devices 103 can be used to collect some of the data 108.
  • the electronic collection devices 103 could include devices configured to collect circadian data (data associated with the circadian rhythm of a user such as sleep data, HR data, activity data, ...), a CGM device, an activity monitor, a smart watch, and the like.
  • circadian data data associated with the circadian rhythm of a user such as sleep data, HR data, activity data, ...)
  • CGM device an activity monitor
  • a smart watch and the like.
  • These electronic data collection devices 103 may be worn on the body in which case they generate “wearable” data.
  • an individual may wear a fitness device, such as an activity-monitoring device, that monitors sleep, motion, heart rate, blood pressure and the like and can be used to determine how much an individual has slept, the number of calories burned, activities performed, body temperature, and the like.
  • the individual may also wear a CGM that monitors blood glucose levels often by measuring levels of glucose in interstitial fluid.
  • CGM that monitors blood glucose levels often by measuring levels of glucose in interstitial fluid.
  • New devices are constantly becoming available to individuals. For example, devices for monitoring/measuring hydration status, metabolism, physical and psychological stress, sleep, and the like are becoming more readily available. Similarly, devices for measuring biomarkers are being developed for use by an individual.
  • an electronic data collection device 103 can include existing devices as well as devices that have yet to be developed.
  • an individual may generate and provide data 108 using a variety of at home biological collection devices 105, which collect a biological sample which requires a biological assay 109 to be performed to generate electronic data 108.
  • These devices may include but are not limited to“At Home Blood Tests” which use blood extraction devices such as finger pricks which in some examples are used with dried blood spot cards, button operated blood collection devices using small needles and vacuum to collect liquid capillary blood and the like.
  • there may be home biological collection devices such as a stool test which is then assayed to produce biomarker data such as gut microbiome data.
  • biomarker data such as blood glucose results collected by the CGM.
  • Some of this data may be non-biomarker data such as photos and time stamps.
  • An individual may also provide data using computing device 102A.
  • an individual can input data 108 into one or more software applications 130A.
  • an individual may enter the food they consumed, a value indicated by a measurement device, their waist measurement and the like.
  • data generated by other measurements can be used to assist in determining when a food was eaten, and/or a test was performed.
  • a CGM can be used to confirm the start point of a meal.
  • data recorded by an individual about when they started to eat can be verified by confirming that there is a rise in glucose detected by the CGM, provided there was sufficient carbohydrate in the meal.
  • the meal, or a different meal may be consumed one or more additional times to improve the accuracy of the at home testing.
  • the repeating of the eating of the same food can be used to obtain more accurate measurements of nutritional responses as compared to eating a single meal of the food.
  • different meals may be consumed to increase the accuracy.
  • the increased accuracy can also be used to generate better predictions.
  • the values of the nutritional responses can be adjusted based on the circadian data.
  • two or more measurements of the same biomarker can be combined to increase the accuracy of the measurement of nutritional responses.
  • measuring fasting bloods on more than one occasion and combining this data can be used to more accurately calculate baseline levels of biomarkers, and therefore improve calculations of nutritional responses.
  • the operating environment 1100 includes one or more computing devices 102, such as computing devices 102A in communication with an environment 106.
  • the environment 106 may be associated with and/or implemented by resources provided by a service provider network such as provided by a cloud computing company.
  • the environment 106 includes a data ingestion service 110, a data service 120, and a data store 140.
  • a service 130 can be utilized to generate clinical state predictions/recommendations.
  • the personalized nutritional recommendations can be generated using techniques described in U.S. Patent Application No. 15/894,798, filed on February 12, 2018, which is incorporated by reference herein in its entirety.
  • the environment 106 may include a collection of computing resources (e.g., computing devices such as servers).
  • the computing resources may include a number of computing, networking and storage devices in communication with one another.
  • the computing resources may correspond to physical computing devices and/or virtual computing devices implemented by one or more physical computing devices.
  • the data ingestion service 110 facilitates submission of data utilized by the data service 120 and, in some configurations, the nutritional service 130. Accordingly, utilizing a computing device 102, an electronic collection device 103, an at home biological collection device 105 or via in clinic biological collection 107, an individual may submit data 108 to the nutritional environment 106 via the data ingestion service 110. Some of the data 108 may be biomarker data, and some of the data 108 may be non-biomarker data such as photos, barcode scans, timing data, and the like. Data may also be obtained by the data ingestion service 110 from other data sources, such as data source(s) 150. For example, the data source(s) 150 can include, but are not limited to nutritional data (e.g., nutrition of particular foods, nutrition associated with the individual, and the like), health data records associated with the individual and/or other individuals, and the like.
  • nutritional data e.g., nutrition of particular foods, nutrition associated with the individual, and the like
  • health data records associated with the individual and/or other individuals and the
  • the data such as data 108, or data obtained from one or more data sources 150, may then be processed by the data manager 112 and/or the data service manager 122 and included in a memory, such as the data store 140.
  • the data store 140 can be configured to store user data 140A, other users’ data 140A2, other data 140B, nutritional data 140C, lab/clinic data 140E, circadian data 140F, and the like.
  • the user data 140A and other users’ data 140A2 includes data and health data that can include psychological data, subjective health data and objective health data.
  • the data 108 is associated with at home measurements of nutritional responses to food.
  • data sources 150 may include training data that can be obtained from a number of individuals (e.g., > 100, 500, 1000, ).
  • This training data may be the results of at home and clinical measurements of nutritional responses using the same or different devices as used for the data 108.
  • This training data may also include circadian data associated with a large number of users.
  • This training data may include the results generated by using the data service 120 on the data from these other individuals, such as adjusting certain data.
  • This training data can be provided to the data service 120 which may utilize a machine learning mechanism or other automated program to analyze the data 108 associated with an at home measurement of a nutritional response.
  • the data service 120 utilizing the data service manager 122 can analyze the data 108 associated with an at home measurement of a nutritional response, and then adjust the data when determined. For example, the data service manager may adjust the data 108 based on the timing of one or more tests, the circadian data, and the like.
  • the at home measurements may be associated with at least two different at home sources of data (e.g., combining at home CGM and at home blood measurements, or combining time recorded on a computing device with blood measurements).
  • the data service 120 utilizes both data associated with the user providing the data and data from other users performing similar tests.
  • the data utilized is associated only with the user.
  • the data can include data obtained from a clinical setting, which is typically more accurate than at home measurements.
  • the data service 120 is configured to determine the level of accuracy of data for the biomarkers associated with insulin, glucose, c-peptide, ketone bodies, triglycerides, IL-6 inflammation, microbiome, hunger, fullness, mood, and the like for an individual.
  • the data service manager 122 may utilize one or more machine learning mechanisms. For example, the data service manager 122 can use a classifier to classify the data within a classification category ln other examples, the data service manager 122 may generate a score that provides an indication of the likely level of accuracy of the data for one or more biomarkers.
  • the data ingestion service 110 and/or the data service 120 can generate one or more user interfaces, such as a user interface 104A and/or user interface 104B, through which an individual, utilizing the computing device 102, or some other computing device, may provide/receive data from the environment 106.
  • the data ingestion service 110 may provide a user interface 104A that allows an individual of the computing device 102A to submit data to the environment 106.
  • the data service 120 may provide a user interface 104A that provides adjustments and/or instructions to the user for performing at home tests for nutritional responses.
  • the test/meal may be repeated more than once to measure the one or more biomarkers ln other examples, two different meals may be consumed.
  • the data service 120 can improve the level of accuracy of the calculated nutritional response.
  • the data service 120 can utilize the circadian data to increase the accuracy of the measured biomarker ln some examples, the data service 120 receives at home measurements for one or more of insulin, c-peptide, glucose, ketone bodies, hunger, energy levels, 1L-6 inflammation and triglycerides lnsulin, glucose and triglycerides are metabolites that can affect body weight.
  • the data service 120 can also be utilized with other at home measurements not specifically discussed.
  • data ingestion service 110 the data service 120, the service 130 are illustrated separately, all or a portion of these services may be located in other locations or together with other components.
  • the data ingestion service 110 may be located within the data service 120.
  • the data service manager 122 may be part of a different service, and the like.
  • the measurements are taken by electronic data collection devices 103, such as the Continuous Glucose Monitor (CGM), that have a limited operating duration.
  • CGM Continuous Glucose Monitor
  • data 108 can also come from other sources, such as at home biological collection devices 105, and/or in-clinic biological collection.
  • food item(s) selected for a particular test are standardized across different individuals. These“standardized meals” are carefully measured meals that can be packaged and eaten by the individual at home lnstructions are provided to the users to eat the meal at predefined times (e.g., after waking up without eating anything else.) ln some examples, more than one meal is utilized. As will be discussed in more detail below, meals that include different nutritional ratios or composition may be included.
  • the at home tests performed can include one or more blood tests that are taken at different points in time relative to when the meal is consumed, depending on biomarkers being measured.
  • the components of carbohydrate, fat, protein (“macronutrients”) and fiber are chosen to help ensure that on average there is a significant post prandial change in the individual’s target biomarkers.
  • the target is to measure blood glucose and blood lipids
  • the meal will have significant carbohydrate levels (e.g. above 30 grams) and fat levels (e.g. above 20 grams.) lf there are a series of standardized meals, the standardized meals can be chosen to expose the individual to a variety of meals so as to measure the interpersonal variability of biomarker responses.
  • One example series of meal might explore a range of levels of macronutrients : (1) metabolic challenge (50g fat, 85g Carbohydrate); (2) high fat breakfast (35g fat, 35g Carbohydrate); (3) medium fat breakfast (22g fat, 71 g carbohydrate), (4) low fat / high carbohydrate breakfast (9g fat, 95g Carbohydrate); (5) Oral Glucose Tolerance Test. (Og fat, 75g Carbohydrate).
  • Other series of meals can be utilized.
  • another series of meals might explore different sorts of macronutrient composition such as varying type of carbohydrates with differing levels of processing (e.g. a bread using highly refined flour vs unrefined rye bread).
  • Not all meals provide the same amount of incremental information about a meal. For example, once the results of an Oral Glucose Tolerance Test (“OGTT”) have been measured then the incremental information from a high carbohydrate breakfast made of processed carbohydrates may be small.
  • the test series of meals may therefore be optimized to choose a set of meals that provide additional incremental information about personalized responses to food based on measuring the incremental information gained from different meals eaten by many individuals who have already had their responses measured ln some cases, the data service can choose a combination of standardized meals that is optimized for an individual that provides insight into the individual’s responses within the limitation of a certain number of standardized meals.
  • insulin post-prandial responses can be calculated by modelling insulin levels via c-peptide levels from at home blood measurements. C-peptide is released into the blood as a byproduct of the formation of insulin by the pancreas. A c-peptide test measures the amount of c- peptide in a blood or urine sample. Post-prandial c-peptide response can be used to provide an estimate of the insulin response. According to some examples, the accuracy of the at home measurements can be enhanced by lab/clinical measurements on the same user or other users using some combination of the same at home testing method and/or higher accuracy clinical assays.
  • Nutritional responses to these meals can be measured using one or more tests ln some examples, At Home Blood Tests can be utilized to not only identify the post-prandial response to these individual meals but may also be utilized to predict responses to meals that were not measured by for example building a model linking biomarker responses to the characteristics of the meal. This model can be implemented within a nutritional service 130 as described herein.
  • the standardized meals may be repeated. Given the high degree of noise inherent in measuring nutritional responses at home, this increases accuracy by having multiple measurements of biomarker responses to the same meals which may be statistically combined to generate a more accurate measure of the individual’s response to a particular meal. This may be done using the data service 120.
  • very high fat meals are utilized to allow measurement of triglycerides responses as well as carbohydrates.
  • this first meal may be followed by a second meal roughly one hour - four hours after the first meal.
  • the meals can be different ln some cases, this high fat meal may consist of two high fat muffins and a NESQU1K milkshake, or a similar drink. Nutrient profile: 869 kcal, 82g carb, 55g fat, 15g protein, 2g fiber. When there is a second meal it may consist of a muffin which might contain 600 kcal (75g Carb, 25g Fat).
  • At Home Blood Test results with other measurements such as blood glucose via a CGM may be utilized.
  • Post-prandial measurements can be timed to coincide with the peaks for the target biomarkers, or to measure the most relevant part of the response for understanding personalized responses. For example, measuring around one hour after eating a meal is optimized for peak c-peptide response, and around 4 hours after the meal is optimized for peak triglycerides response ln another example with triglycerides the most relevant measure may not be the peak value, but the level of triglycerides recorded beyond four hours when the lipoprotein components of the aggregate triglycerides that is measured will have remodeled leading to remnant triglyceride rich lipoproteins and an increased number of atherogenic lipoproteins.
  • the timing of blood tests at home is determined based on measurement of the particular post-prandial response to that meal in a clinical setting to identify the average person’s peaks for the target biomarker ln some examples an area under curve is desired to be modeled which uses at least two measures beyond the fasting measure.
  • biomarkers which peak rapidly after meals e.g. c-peptides
  • those that peak slowly e.g. fats such as triglycerides
  • this uses at least three time points for blood collection on a single meal. For example, one before the meal (fasting), one measurement at around one hour after the meal, and another measurement at around four hours after the meal. This timing allows peak levels to be captured for a range of biomarkers and then compared between individuals.
  • the timing of tests performed at home affects how accurate are the results of the test, as many biomarkers change significantly and rapidly post-prandially.
  • the clock used for recording the timing of the tests can be synchronized with the clock used by the CGM, and the clock used for the online food logging before the at home collection begins. This will allow the data to be accurately combined which may be done using the data service 120. If timing is inaccurate the values measured can be far from their true values.
  • the application 130 operating on the computing device 102 can be programmed to adjust the clocks on the applications running on the device and/or data service 120 can determine the time differences between each of the clocks such that the data is correlated. As also discussed above, the data service 120 can adjust the results based on the circadian data.
  • a Continuous Glucose Monitor may be attached to the individual for some or all of the period they are being measured at home. This CGM may be attached by the individual themselves at home.
  • the CGM’s clock can be synchronized with the clock on the computing device 102A, usually by synchronizing both to Internet time.
  • the CGM can also be used to check the timing and content of the meals. If a glucose spike is not triggered shortly after a meal that is known to lead to such spikes (for example most meals that have carbohydrates in them) then this can be used by the data service 120 to reinterpret the data from an At Home Blood Test, and either reject it or adjust for the correct starting point of the meal.
  • the size of the spike can also help to identify the accuracy of the food logged, so for example if the glucose spike is higher than expected then this could suggest the quantity of food is more than was logged.
  • c-peptide may be measured using At Home Blood Tests, and then the insulin level may be calculated from the value of the c-peptide. This works because c-peptide can be relatively stable over many days, especially if using dried blood tests or other blood collection methods which may be put in the fridge at home before sending them to be analyzed. To calculate insulin values from c-peptide, it is helpful to have a large benchmark of data from in clinic biological collections 107 comparing the calculated c-peptide levels using the At Home Blood Test process at the clinical visit with the known values of insulin measured very accurately using venous blood and clinical biochemistry at the same time as the At Home Blood Test process.
  • the data service 120 can utilize data received by many different users in determining whether or not a measurement is accurate.
  • the data service 120 can utilize a data analyst, a machine learning, and/or some other mechanism to generate a quality score for the data.
  • the activity level of individuals can be monitored using electronic data collection devices 105 that include devices like accelerometers and heart rate measurement. These can be used to calculate exercise and sleep amongst other things. These can be used as another check by the data service 120, as for example it isn’t possible to be running and doing blood tests at the same time, and it isn’t a good baseline blood measurement if the individual has been awake for twelve hours or did not sleep the night before.
  • electronic data collection devices 105 that include devices like accelerometers and heart rate measurement. These can be used to calculate exercise and sleep amongst other things. These can be used as another check by the data service 120, as for example it isn’t possible to be running and doing blood tests at the same time, and it isn’t a good baseline blood measurement if the individual has been awake for twelve hours or did not sleep the night before.
  • the sleep and wake times of individuals are measured at home.
  • electronic data collection devices 103 can be used to accurately measure sleep and wake times for an individual.
  • the sleep and wake times for an individual can then be used to improve the accuracy of predictions and to predict responses at certain times of day or given a certain amount of sleep.
  • the data service 120 can utilize data (e.g., sleep data) received by many different users.
  • the data service 120 can then utilize a data analyst, a machine learning, and/or some other mechanism to adjust this data 108B as well as providing the raw data to the nutritional service.
  • cross-checking can be performed by the data service 120, a support team including one or more data analysts, or some other computing device. For example, through cross-checking of various combinations of two or more of: questionnaires, photos taken by the mobile phone, food logged, CGM, At Home Blood Spot recording, activity level monitoring or other electronic data collection devices, and data recorded by the one or more data analysts a determination can be made as to whether the data is accurate.
  • the cross-checking can be performed by the data analyst, and/or the data service. As discussed above, circadian data can be used to improve the accuracy of the measured responses.
  • FIGs. 12 and 13 are flow diagrams showing routines 1200, 1300, in accordance with examples described herein. It should be appreciated that at least some of the logical operations described herein with respect to FIGs. 12 and 13, and the other FIGS., may be implemented (1) as a sequence of computer implemented acts or program modules running on a computing system and/or (2) as interconnected machine logic circuits or circuit modules within the computing system. [0132] The implementation of the various components described herein is a matter of choice dependent on the performance and other requirements of the computing system. Accordingly, the logical operations described herein are referred to variously as operations, structural devices, acts, or modules. These operations, structural devices, acts, and modules may be implemented in software, in firmware, in special purpose digital logic and any combination thereof. It should also be appreciated that more or fewer operations may be performed than shown in the FIGS and described herein. These operations may also be performed in parallel, or in a different order than those described herein.
  • FIG. 12 is a flow diagram showing routine 1200 illustrating aspects of a mechanism disclosed herein for generating predictions of clinical state/health outcomes using data obtained in a non-clinical setting.
  • the routine 1200 may begin at 252, where food (e.g., a meal) is selected that is consumed to evoke a nutritional response.
  • food e.g., a meal
  • the food can be a series of standardized meals, a single meal, a single food, or some other selection of food.
  • food data is received.
  • An individual may log the time the food was consumed, or record the food consumed using other mechanisms. In some configurations, the timing may be based on when an image was taken of the food to consume. In other configurations, the user can be instructed to take one or more pictures of the food before, during and/or after consuming the food.
  • a user may also provide other data relating to the food, such as but not limited to the food state (e.g., cooked/raw, an age of the food, ...), how long it took to consume the food, and the like.
  • a computing device 102A associated with the user provides the food data 304 to a data service 120 and/or a data ingestion service.
  • At 256 data associated with at home tests performed to measure one or more nutritional responses are received.
  • the tests are performed at one or more points in time after eating a particular food, or foods of a meal.
  • the at home tests can include blood tests, and/or other tests that measure other biomarkers.
  • the individual may perform a blood test.
  • the individual may perform some other type of test.
  • tests are automatically carried out by electronic data collection devices. For instance, a measure of blood glucose can be taken, or the acceleration of a smart watch recorded.
  • the data ingestion service 110 can receive the data from a computing device 102 or electronic data collection device 103 associated with the individual. In other examples, the data is received from another source (e.g., the individual returns the collected biological sample via mail or some other courier service and a biological assay 109 is performed that outputs data to the data ingestion service 110). The data ingestion service 110 can also receive other data, such as circadian data.
  • the data can be adjusted when determined.
  • the data service 120 is configured to determine whether the data is accurate. For example, was the data obtained at a proper time as indicated by the test protocol for a particular test?, was the proper data collected? (e.g., is the bloodspot filled correctly?), was the proper food consumed for the test?, was the proper amount of food consumed?, was the food properly classified?, is the sleep pattern of the user following the normal pattern for the user?, and the like.
  • the data service 120 can utilize data received from the computing device 102 (e.g., timing data, image data) to assist in determining whether the test protocol was followed.
  • the data service 120 can also utilize one or more data analysts.
  • a decision is made as to whether to repeat the at home test and/or have the user perform a different test.
  • the protocol may specify that the user consume two or more meals instead of one, and that the spacing of the meals be some time period.
  • the process returns to 252.
  • the process flows to 262.
  • the data is utilized ln some examples, the data is used by a service to generate predictions of clinical state/health outcomes. This data may have been adjusted by the data service 120. ln other configurations, this data may have been provided with a weight by the data service which is taken into account by the service utilizing the data (e.g., predict one or more health outcomes for an individual), which in some cases may involve machine learning mechanisms.
  • F1G. 13 is a flow diagram showing routine 1300 illustrating aspects of a mechanism disclosed herein for utilizing data associated with circadian rhythms for improving the accuracy of at home measurements/predictions.
  • the routine 1300 may begin at 302, where data is received.
  • the data can include circadian data for one or more users.
  • the circadian data can include sleep data, HR data, temperature data, and the like.
  • all or a portion of the data can be provided to one or more machine learning mechanisms.
  • the data can be provided to a data service 120 and/or the data can be accessed by the machine learning mechanism from a memory and/or some other data store.
  • the machine learning mechanism can be trained to determine whether or not the circadian rhythm of a user is following a“normal” pattern or not. For instance, when the machine learning mechanism, or some other technique, determines that the user’s circadian rhythm is off, the data can be adjusted ln some configurations, the machine learning mechanism is trained using thousands of samples from other users. More or fewer samples, however, can be used for training.
  • updated/adjusted data is received from the machine learning mechanism.
  • the machine learning mechanisms can determine if the data is accurate, adjust the data (e.g., using the circadian data), adjust one or more parameters, and the like.
  • FIG. 14 is a block diagram depicting an illustrative operating environment 1400 in which a data ingestion service 110 receives and processes data associated with data associated with at home measurements of nutritional responses.
  • the data manager 112 is configured to receive data such as, health data 202 that can include, but is not limited to microbiome data 206A, triglycerides data 206B, glucose data 206C, blood data 206D, wearable data 206E that can include circadian data, questionnaire data 206F, psychological data (e.g., hunger, sleep quality, mood, ...) 206G, objective health data (e.g., height, weight, medical history, ...) 206H, nutritional data 140B, and other data 140C.
  • health data 202 can include, but is not limited to microbiome data 206A, triglycerides data 206B, glucose data 206C, blood data 206D, wearable data 206E that can include circadian data, questionnaire data 206F, psychological data (e.g., hunger, sleep quality, mood, ...) 206G, objective health data (e.g., height, weight, medical history, ...) 206H, nutritional data 140B, and other data 140C.
  • the microbiome data 206A includes data about the gut microbiome of an individual.
  • the gut microbiome can host a large number of microbial species (e.g., > 1000) that together have millions of genes.
  • the triglycerides data 206B may include data about triglycerides for an individual ln some examples, the triglycerides data 206B can be determined from an At Home Blood Test which in some cases is a finger prick on to a dried blood spot card.
  • the glucose data 206C includes data about blood glucose.
  • the glucose data 206C may be determined from various testing mechanisms, including at home measurements, such as a continuous glucose meter.
  • the blood data 206D may include blood tests relating to a variety of different biomarkers. As discussed above, at least some blood tests can be performed at home ln some configurations, the blood data 206D is associated with measuring blood sugar, insulin, c-peptides, triglycerides, 1L-6 inflammation, ketone bodies, nutrient levels, allergy sensitivities, iron levels, blood count levels, HbAlc, and the like.
  • the wearable data 206E can include any data received from a computing device associated with an individual.
  • an individual may wear an electronic data collection device 103, such as an activity-monitoring device, that monitors motion, heart rate, heart rate variability, determines how much an individual has slept that may include the types of sleep, the times an individual is awake, , the number of calories burned, activities performed, blood pressure, body temperature, and the like.
  • the individual may also wear a continuous glucose meter that monitors blood glucose levels.
  • the questionnaire data 206F can include data received from one or more questionnaires, and / or surveys received from one or more individuals.
  • the psychological data 206G may include data received from the individual and/or a computing device that generates data or input based on a subjective determination (e.g., the individual states that they are still hungry after a meal, or a device estimates sleep quality based on the movement of the user at night perhaps combined with heart rate data).
  • the objective health data 206H includes data that can be objectively measured, such as but not limited to height, weight, medical history, and the like.
  • the nutritional data 140B can include data about food, which is referred to herein as“food data”.
  • the nutritional data can include nutritional information about different food(s) such as their macronutrients and micronutrients or the bioavailability of its nutrients under different conditions (raw vs cooked, or whole vs ground up).
  • the other data 142B can include other data associated with the individual.
  • the other data 142B can include data that can be received directly from a computer application that logs information for an individual (e.g., food eaten, sleep, ...) and/or from the user via a user interface.
  • computing devices 102 associated with different users provide application data 204 to the data manager 112 for ingestion by the data ingestion service 110.
  • computing device 102A provides app data 204A to the data manager 112
  • computing device 104B provides app data 204B to the data manager 112
  • computing device 104N provides app data 204N to the data manager 112.
  • the data manager 112 receives data from different data sources, processes the data when needed (e.g., cleans up the data for storage in a uniform manner), and stores the data within one or more data stores, such as the data store 140.
  • the data manager 112 can be configured to perform processing on the data before storing the data in the data store 140. For example, the data manager 112 may receive data for ketone bodies and then use that data to generate ketone body ratios. Similarly, the data manager 112 may process food eaten and generate meal calories, number of carbohydrates, fat to carbohydrate rations, how much fiber consumed during a time period, and the like.
  • the data stored in the data store 140, or some other location, can be utilized by the data service 120 to determine an accuracy of at home measurements of nutritional responses performed by users.
  • the data outputted by the data service 108B to the nutritional service may therefore contain different values than are stored in the data store 140, for example if a food quantity is adjusted.
  • F1G. 15 shows an example computer architecture for a computer 1500.
  • the computer architecture shown in F1G. 15 illustrates a conventional server computer, workstation, desktop computer, laptop, mobile device, tablet, network appliance, digital cellular phone, smart watch, or other computing device, and may be utilized to execute any of the software components presented herein.
  • the computer architecture shown in F1G. 15 may be utilized to execute software components for performing operations as described above.
  • the computer architecture shown in F1G. 15 might also be utilized to implement a computing device 102, or any other of the computing systems described herein.
  • the computer 1500 includes a baseboard 502, or“motherboard,” to which a multitude of components or devices may be connected by way of a system bus or other electrical communication paths ln one illustrative example, one or more central processing units (“CPUs”) 504 operate in conjunction with a chipset 506.
  • the CPUs 504 may be standard programmable processors that perform arithmetic and logical operations necessary for the operation of the computer 1500.
  • the chipset 506 provides an interface between the CPUs 504 and the remainder of the components and devices on the baseboard 502.
  • the chipset 506 may provide an interface to memory 506, such as RAM and/or read-only memory (“ROM”).
  • the memory may store software components utilized for the operation of the computer 1500 in accordance with the examples described herein.
  • the computer 1500 may operate in a networked environment using logical connections to remote computing devices and computer systems through a network, such as the network 520.
  • the chipset 506 may include functionality for providing network connectivity through a network interface controller, such as a cellular network adapter, WiFi network adapter, Ethernet adapter, and the like.
  • the NIC 512 is capable of connecting the computer 1500 to other computing devices over the network 520.
  • the computer 1500 may be connected to a mass storage device 516 that provides storage for the computer.
  • the mass storage device 516 may store system programs, application programs, other program modules and data, which have been described in greater detail herein.
  • the mass storage device 516 may be connected to the computer 1500 through a storage controller 514 connected to the chipset 506.
  • the mass storage device 516 may consist of one or more physical storage units.
  • the storage controller 514 may interface with the physical storage units through various type of interfaces.
  • the mass storage device 516 described above, the computer 1500 may have access to other computer-readable storage media to store and retrieve information, such as program modules, data structures, or other data.
  • computer-readable storage media may include volatile and non-volatile, removable and non-removable media implemented in any method or technology.
  • the mass storage device 516 may store an operating system 530 utilized to control the operation of the computer 1500.
  • the operating system may include, but is not limited to the UNIX operating system, the LINUX operating system, the WINDOWS® operating system from MICROSOFT Corporation, the iOS operating system from APPLE Corporation, the ANDROID operating system from GOOGLE, and the like. Other operating systems may also be utilized.
  • the mass storage device 516 may store other system or application programs and data utilized by the computer 1500, such as components that include the data manager 122, and/or any of the other software components and data described above. The mass storage device 516 might also store other programs and data not specifically identified herein.
  • the mass storage device 516 or other computer-readable storage media is encoded with computer-executable instructions that, when loaded into the computer 1500, implement the examples described herein.
  • the computer 1500 has access to computer-readable storage media storing computer-executable instructions which, when executed by the computer 1500, may be configured to perform the various routines described above.
  • the computer 1500 might also include computer-readable storage media for performing any of the other computer-implemented operations described herein.
  • the computer 1500 may also include one or more input/output controllers 516 for receiving and processing input from a number of input devices, such as an electronic data collection device, a keyboard, a mouse, a touchpad, a touch screen, an electronic stylus, or other type of input device. Similarly, the input/output controller 516 may provide output to one or more types of output devices (e.g., a display, a projector, a printer, ).
  • the computer 1500 may not include all of the components shown in FIG. 15, may include other components that are not explicitly shown in FIG. 15, or may utilize an architecture completely different than that shown in FIG. 15.

Landscapes

  • Health & Medical Sciences (AREA)
  • Nutrition Science (AREA)
  • Engineering & Computer Science (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Primary Health Care (AREA)
  • Public Health (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

L'invention concerne des techniques pour utiliser des repas normalisés et des dispositifs numériques qui peuvent être utilisés à l'extérieur d'un établissement clinique pour assister dans la détermination/prédiction d'un ou plusieurs états cliniques associés à un ou plusieurs individus. L'utilisation des technologies décrites ici permet d'utiliser différentes techniques utilisées pour appliquer des mesures de réponse nutritionnelle. Des mesures de réponse peuvent être obtenues, par exemple, pour deux repas gras au lieu d'un repas gras unique. Des données associées à un rythme circadien (par exemple des temps de sommeil, des temps d'éveil) peuvent également être utilisées pour améliorer la précision des biomarqueurs mesurés. Dans d'autres exemples, des données de biome et d'autres données peuvent être utilisées.
EP19759372.6A 2018-08-27 2019-08-27 Utilisation de mesures à domicile pour prédire un état clinique et améliorer la précision de mesures/prédictions à domicile en utilisant des données associées au rythme circadien et aux horaires des repas Pending EP3844767A1 (fr)

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
US201862723429P 2018-08-27 2018-08-27
US201862723424P 2018-08-27 2018-08-27
US16/120,039 US20200066181A1 (en) 2018-08-27 2018-08-31 Generating Personalized Food Recommendations from Different Food Sources
US16/272,865 US11915151B2 (en) 2018-08-27 2019-02-11 Accuracy of test data outside the clinic
US201962808180P 2019-02-20 2019-02-20
US201962821949P 2019-03-21 2019-03-21
US16/434,135 US11295860B2 (en) 2018-02-12 2019-06-06 Using at home measures to predict clinical state and improving the accuracy of at home measurements/predictions data associated with circadian rhythm and meal timing
PCT/EP2019/072806 WO2020043706A1 (fr) 2018-08-27 2019-08-27 Utilisation de mesures à domicile pour prédire un état clinique et améliorer la précision de mesures/prédictions à domicile en utilisant des données associées au rythme circadien et aux horaires des repas

Publications (1)

Publication Number Publication Date
EP3844767A1 true EP3844767A1 (fr) 2021-07-07

Family

ID=69642966

Family Applications (1)

Application Number Title Priority Date Filing Date
EP19759372.6A Pending EP3844767A1 (fr) 2018-08-27 2019-08-27 Utilisation de mesures à domicile pour prédire un état clinique et améliorer la précision de mesures/prédictions à domicile en utilisant des données associées au rythme circadien et aux horaires des repas

Country Status (2)

Country Link
EP (1) EP3844767A1 (fr)
WO (1) WO2020043706A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3896700A1 (fr) * 2020-04-17 2021-10-20 Perfood GmbH Méthode améliorée pour la détermination des réponses glycémiques
WO2023046761A1 (fr) 2021-09-22 2023-03-30 Zoe Limited Modification directionnelle de la composition du microbiome par l'intermédiaire de la modification du régime alimentaire

Also Published As

Publication number Publication date
WO2020043706A1 (fr) 2020-03-05

Similar Documents

Publication Publication Date Title
US11295860B2 (en) Using at home measures to predict clinical state and improving the accuracy of at home measurements/predictions data associated with circadian rhythm and meal timing
US20240152783A1 (en) Improving the accuracy of test data outside the clinic
US11183080B2 (en) Generating predicted values of biomarkers for scoring food
US11183291B2 (en) Generating personalized nutritional recommendations using predicted values of biomarkers
US20200066181A1 (en) Generating Personalized Food Recommendations from Different Food Sources
AU2022291499A1 (en) Analysis of glucose median, variability, and hypoglycemia risk for therapy guidance
Ståhl et al. Diabetes mellitus modeling and short-term prediction based on blood glucose measurements
US11817200B2 (en) Generating personalized food guidance using predicted food responses
CN102802508B (zh) 用于根据血糖的结构化单点测量结果来提供估计真实平均血糖值和估计糖化血红蛋白(HbA1c)值两者的方法、系统和计算机程序产品
US11348479B2 (en) Accuracy of measuring nutritional responses in a non-clinical setting
WO2020043706A1 (fr) Utilisation de mesures à domicile pour prédire un état clinique et améliorer la précision de mesures/prédictions à domicile en utilisant des données associées au rythme circadien et aux horaires des repas
WO2021087608A1 (fr) Système et procédé pour l'évaluation de l'homéostasie du glucose
US20210265029A1 (en) Generating microbiome fingerprints, dietary fingerprints, and microbiome ancestry
US20220130512A1 (en) Generating personalized food guidance using predicted hunger
US20220354392A1 (en) Personalized glucose ranges for making healthy choices
US20220367050A1 (en) Predicting gut microbiome diversity
Das et al. Predicting the macronutrient composition of mixed meals from dietary biomarkers in blood
WO2023046761A1 (fr) Modification directionnelle de la composition du microbiome par l'intermédiaire de la modification du régime alimentaire
WO2022232938A1 (fr) Procédés et systèmes pour la classification de sujets dans des phénotypes de l'homéostasie du glucose

Legal Events

Date Code Title Description
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: UNKNOWN

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE

PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20210224

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

DAV Request for validation of the european patent (deleted)
DAX Request for extension of the european patent (deleted)
RIN1 Information on inventor provided before grant (corrected)

Inventor name: DAVIES, RICHARD JAMES

Inventor name: HADJIGEORGIOU, GEORGE

Inventor name: WOLF, JONATHAN THOMAS

RAP3 Party data changed (applicant data changed or rights of an application transferred)

Owner name: ZOE LIMITED

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: EXAMINATION IS IN PROGRESS