Paper reading (三十一)：Personalized Nutrition by Prediction of Glycemic Responses（overview）

论文题目：Personalized Nutrition by Prediction of Glycemic Responses

scholar 引用：768

页数：17

发表时间：2015.11

发表刊物：Cell

作者：David Zeevi, Tal Korem, Niv Zmora, ..., Zamir Halpern, Eran Elinav, Eran Segal

Department of Computer Science and Applied Mathematics, Weizmann Institute of Science, Rehovot 7610001, Israel

Highlights：

High interpersonal variability(人与人之间的差异) in post-meal glucose(餐后血糖) observed in an 800-person cohort(群)
Using personal and microbiome(微生物组) features enables accurate glucose response prediction
Prediction is accurate and superior to common practice(常规做法) in an independent cohort
Short-term personalized dietary interventions(个性化饮食干预) successfully lower post-meal glucose

摘要：

Elevated(升高) postprandial(餐后的) blood glucose levels constitute a global epidemic and a major risk factor for prediabetes(前驱糖尿病) and type Ⅱ diabetes, but existing dietary methods for controlling them have limited efficacy(功效，效力). Here, we continuously monitored week-long(为期一周的) glucose levels in an 800-person cohort, measured repsonses to 46,898 meals, and found high variability in the response to identical meals, suggesting that universal dietary recommendations may have limited utility. We devised a machine-learning algorithm that integrates blood parameters, dietary habits, anthropometrics(人体测量学), physical activity, and gut microbiota measured in this cohort and showed that it accurately predicts personlized postprandial glycemic response to real-life meals. We validated these predictions in an independent 100-person cohort. Finally, a blinded randomized controlled dietary interventions based this algorithm resulted in significantly lower postgrandial responses and consistent alterations to gut microbiota configuration. Together, out results suggest that personalized diets may successfully modified elevated postprandial blood glucose and its metobolic consequences.

餐后血糖水平升高是糖尿病的主因，目前已有的控制餐后血糖水平的营养学方法收效甚微。
我们对800个人进行了为期一周的血糖水平监测，总计测量了46,898顿饭的餐后血糖指标（粗略估计46898÷800÷7=8.37，所以人均一天吃8顿饭？？？有点多了吧。。。），我们发现每天吃同样的食物，不同人的血糖反应差异较大，因此，给不同的人推荐同样的营养学方法可能作用有限。（问题1：这个监测是如何进行的？得出这个结论的过程是否严谨？实验中哪些因素可能影响结果，从而影响结论？）
我们设计了一种机器学习算法，该算法集成了血相参数、饮食习惯、人体测量学、体育锻炼和肠道菌群的数据，实验结果表明，该算法可以准确预测日常饮食后的个体餐后血糖指标。（问题:2：算法的数据处理、结构、输入、输出等，重点关注）
我们在100个人身上进行了预测测试。基于我们的算法的盲法随机对照实验对饮食干预后，餐后血糖水平显著下降，肠道菌群也发生了一致性的变化。（问题3：实验方法具体是如何的？重点关注）
总之，我们的实验结构表明，个性化饮食可以有效改善餐后血糖水平升高现象以及代谢产物。

Video Abstract:

When we eat, the carbohydrates from the food are broken down to simple sugars, which are then absorbed from out intestine into blood stream. Blood sugar levels are a key factor that affects the pathogenesis of diseases such as diabetes and obesity. Diets aimed at controlling blood glucose levels are often similar even for different people. But what if we told you that these common diets aimed at maintaining stable blood sugar levels, may in some people achieve the exact opposite?

How is this possible?

People are different in many ways. For example, in their genetic makeup, in their lifestyle, and also, in their microbiomes. The microbiome is a huge ecosystem of trillions of bacteria living inside our body with more than 100 times the numbers of genes contained in the human genome. The microbiome is influenced by what we eat, and in turn, affects our response to food, and as the microbiome differs greatly from one person to another, it can also affect the blood sugar response to food. For the past few years, scientists at the Weizmann Institute have studied the factors underlying variation in post-meal blood sugar responses. They collected health and lifestyle data from 800 voluteers who were connected to a device that monitored their blood sugar level every five minutes for an entire week. The participants also used a mobile app to ducument what and when they ate, exercised, slept and ... so on. Stool(粪便) samples were collected in order to analyze the composition and activity of their microbiomes. The scientists discovered that when different people ate identical foods, they often reacted in a very different way. For example, the blood sugar level of some people rose more significantly after eating sushi(寿司) than after eating ice-cream. The scientists were able to integrate all of the data they collected into an algorithm that successfully predicted the blood sugar response to the meals of the 800 participants. The same algorithm achieved similar accuracy when predicting the sugar reponses of 100 new participiants. the scientists also showed how the algotithm could be used to prescribe(规定) personalized diets. A good diet that lowers post-meal sugar response. And a bad diet that raises sugar responses. Interestingly, some foods that appeared on the good diet of one person appeared on the bad diet of another. Let's hear what the researchers themselves have to say.

Prof Eran Segal: If I highlight the key contributions of our work, I would say that the first is in highlighting the need for personlized nutrition which we demostrate by showing that the blood sugar response of different people to identical meals can be hugely different. And as soon as we saw this data, we realized that general dietary recommendations given to the entire population may have limited efficacy. The second is in then measuring, for every individual in our nearly 1000 people cohort a very comprehensive profile that includes their medical backgroud, questionnaires, physical activity, blood tests and gut microbiome function and composition and then integrating this data into a computational algorithm that could successful predict the personlized blood glucose response of people to arbitrary meals. And then, finally, in showing that applying this algorithm to design personally tailored dietary interventions in individuals could significantly lower their blood sugar response to food and that was accompanied by consistent alterarions to the gut microbiome.

Dr. Eran Elinav: We know that nutrition is a very important risk factor for human metabolic disease and especially, to the obesity and diabetes epidemics that are affecting the lives of close to half of the world's population. In this work, we link nutrition, in a personalized manner to human risk to develop elevated blood sugar levels and their many complications. As scientists, we often deal with very basic quesions. but in this work, we are very happy to also introduce a potential that, if further developed, would allow to benefit the health of millions across the world.

This research marks an important step towards personalized nutrition by predicting post-meal blood glucose(glycemic) responses. The scientists hopes that this approach will help to achieve a healthier lifestyle and prevent metabolic disease worldwide.

Graphical Abstract:

Discussion：

PPGR: post-meal glycemic responses
Such continuous assessment of PPGRs is complementary to other equally important clinical parameters such as BMI and HbA1c%, for which changes typically occur over longer timescales and are thus difficult to correlate to nutritional responses in real time.
In line with few small-scale studies that previously examined individual PPGRs (Vega-López et al., 2007, Vrolix and Mensink, 2010)，所以这是他们早期的一些研究，如果摘要部分的问题1我在本篇paper中没有找到想要的答案的话，可以看看这些paper。
In some cases, such as for Actinobacteria(放线菌), Proteobacteria(变形门菌) , Enterobacteriaceae(肠杆菌科), the direction of our associations are consistent with previous associations reported between these taxa(分类群) and higher-level phenotypes such as dietary habits, obesity and overall glycemic control, rising testable hypotheses about how these taxa may mediate these host metabolic effects.
Our study further attempts to analyze real-life meals that are consumed in complex food combinations, at different times of the day, and in varying proximity to previous meals, physical activity, and sleep.
While clearly of higher translational relevance, the use of “real-life” nutritional input also introduces noise into the meal composition data. 这个noise指什么？就是一些不可控因素吧？
algorithm's input: a comprehensive clinical and microbiome profile
a data-driven unbiased approach 这是什么方法？
In many such cases, microbiome factors found to be beneficial with respect to PPGRs are also negatively associated with risk factors such as HbA1c% and cholesterol levels.
It will be interesting to evaluate the utility of such personalized intervention over prolonged periods of several months and even years. 后续准备研究长期个性化干预饮食的影响，这是2015年的文章，那么4年过去，是否他们已经有了新的结果？
our individualized nutritional study protocols may be applicable to address other clinically relevant issues
More broadly, accurate personalized predictions of nutritional effects in these scenarios may be of great practical value, as they will integrate nutritional modifications more extensively into the clinical decision-making scheme. 所以说给每一个病人都进行一次个性化饮食推荐可能比较耗时，假想一下，有那么多种疾病，每个人的情况也是不一样的，但是治疗规范基本上也是国际统一的，所以这种规范化的流程，就是充分考虑个体差异性，尽量提出覆盖率能最广的统一的流程吧？所以比如说糖尿病人的营养学方法推荐，假想一下，可以根据每个人的生活习性，再加肠道菌群的检验，然后合并更多的变量，再来做出推荐的饮食习惯，而这些可以让算法来做。比如说孕期的营养门诊，还要让孕妇手填表格，感觉太不智能了吧。

Introduction：

Blood glucose levels 跟Ⅱ型糖尿病、肥胖症、高血压、非酒精性脂肪肝、高甘油三酯血症、心血管疾病等有关。
maintaining normal blood glucose levels is considered critical for preventing and controlling the metabolic syndrome
Despite their importance, no method exists for predicting PPGRs to food.
之前的方法有：use the meal carbohydrate content； glycemic index，quantifies PPGR to consumption of a single tested food type, and the derived glycemic load，比如说糖耐测试？喝一大杯葡萄糖，然后过几个小时测一下血糖这种喽?
the few small-scale (n = 23–40) studies that examined interpersonal differences in PPGRs found high variability in the response of different people to the same food 这个研究已经有人做过了，只是没有研究显著性差异结果背后的原因。
可能影响PPGRs的因子：基因，生活方式，胰岛素敏感度，胰腺外分泌，葡萄糖转运蛋白活性水平以及肠道菌群。
有很多研究工作是关于PPGRs的各种因子，但是很少是关于肠道菌群和PPGRs之间联系的。
Here, we set out to quantitatively measure individualized PPGRs, characterize their variability across people, and identify factors associated with this variability.
总共六段，第一二段讲述了Blood glucose levels和PPGR的重要性；第三段描述了研究PPGR与食物联系的一些方法，但是这些方法都存在局限；第四段讲述不同的人摄入相同的食物后PPGR有较大差异，而这个样本量还只是小样本23~40；第五段介绍了可能影响PPGRs的一些因子，以及已有的相关研究工作，指出关于肠道菌群与PPGRs联系的研究工作还很少；第六段介绍了本文的主要研究工作内容和成果。

正文组织架构：

1. Introduction

2. Results

2.1 Measurements of postprandial responses, clinical data, and gut microbiome

2.2 Postprandial glycemic responses associate with multiple risk factors

2.3 High interpersonal variability in the postprandial response to identical meals

2.4 Postprandial variability is associated with clinical and microbiome profiles

2.5 Prediction of personalized postprandial glycemic responses

2.6 validation of personalized postprandial glycemic responses predictions on an independent cohort

2.7 Factors underlying personalized predictions

2.8 Personally tailored dietary interventions improve postprandial responses

2.9 Alterations in gut microbiota following personally tailored dietary interventions

3. Discussion

4. Experimental Procedures

4.1 Human cohorts

4.2 Study design

4.3 Standardized meals

4.4 Stool sample collection

4.5 Genomic DNA extraction and filtering

4.6 Microbial analysis

4.7 Associating PPGRs with risk factors and microbiome profile

4.8 FDR correction

4.9 Meal preprocessing

4.10 PPGR predictor

4.11 Microbiome changes during dietary intervention

Paper reading (三十一)：Personalized Nutrition by Prediction of Glycemic Responses（overview）相关推荐

Paper reading (三十二)：Personalized Nutrition by Prediction of Glycemic Responses（Results）
正文部分内容摘录: 2. Results 2.1 Measurements of postprandial responses, clinical data, and gut microbiome h ...
Paper reading (二十一)：Human gut microbiome: hopes, threats and promises
论文题目:Human gut microbiome: hopes, threats and promises scholar 引用:120 页数:10 发表时间:2018.04 发表刊物:Gut 作者 ...
Paper reading (九十一)：Bifidobacterium breve Bif195 Protects Against Small-Intestinal Damage
论文题目:Bifidobacterium breve Bif195 Protects Against Small-Intestinal Damage Caused by Acetylsalicylic ...
Paper Reading：3DIoUMatch：Leveraging IoU Prediction for Semi-Supervised 3D Object Detection
Background 3D对象检测是一项要求很高的任务,它可以以具有语义意义的3D形式的边界框提供给下游任务.并且严重的依赖于3D对象的数据标注.虽然传感器技术的快速进步促进了3D场景数据手机的进步, ...
#Paper reading#DeepInf: Social Influence Prediction with Deep Learning
#Paper reading# DeepInf: Social Influence Prediction with Deep Learning 设计了一个端到端的框架DeepInf,研究用户层面的社会 ...
[paper reading] YOLO v1
[paper reading] YOLO v1 GitHub:Notes of Classic Detection Papers 本来想放到GitHub的,结果GitHub不支持公式. 没办法只能放到 ...
[paper reading] FCOS
[paper reading] FCOS GitHub:Notes of Classic Detection Papers 2020.11.09更新:更新了Use Yourself,即对于本文的理解和 ...
[paper reading] CenterNet (Object as Points)
[paper reading] CenterNet (Object as Points) GitHub:Notes of Classic Detection Papers 2020.11.09更新:更 ...
[paper reading] CenterNet (Triplets)
[paper reading] CenterNet (Triplets) GitHub:Notes of Classic Detection Papers 2020.11.09更新:更新了Use Yo ...
[paper reading] CornerNet
[paper reading] CornerNet GitHub:Notes of Classic Detection Papers 本来想放到GitHub的,结果GitHub不支持公式. 没办法只能 ...

Paper reading (三十一)：Personalized Nutrition by Prediction of Glycemic Responses（overview）

Paper reading (三十一)：Personalized Nutrition by Prediction of Glycemic Responses（overview）相关推荐

最新文章

热门文章