Paper reading (七十五)：I sense overeating: Motif-based machine learning framework to detect overeating

论文题目：I sense overeating: Motif-based machine learning framework to detect overeating using wrist-worn sensing

scholar 引用：12

页数：11

发表时间：2017.08

发表刊物：Information Fusion

作者：Shibo Zhanga, William Stogina, Nabil Alshurafa

摘要：

Obesity, caused primarily by overeating, is a preventable chronic disease yielding staggering healthcare costs. To detect overeating passively, a machine learning framework was designed to detect and accurately count the number of feeding gestures during an eating episode to characterize each eating episode with a feeding gesture count using a 6-axis inertial wrist-worn sensor. Moreover, detecting feeding gestures is useful to aid in end-of-day dietary recalls. It has been shown that feeding gesture count correlates with caloric intake; the more one eats, the more calories one is likely consuming. Recent research has shown promise in passively detecting feeding gestures, but this effort focuses on bridging detection of feeding gesture count and identifying overeating episodes. This paper presents results on three experiments: highly structured (participants pretending to eat), in-lab structured with confounding activities (participants eating while performing other scripted activities), and unstructured overeating (participants induced to overeat while watching television and eating their favorite foods). Our experiment successfully induced overeating in 50% of the participants, showing a correlation between feeding gesture count and caloric intake in unstructured eating (r=.79, p-value=.007). Results provide an approximate upper bound on feeding gesture classification using exact segmentation techniques, and show improvement when compared to prior sliding window techniques. Results also suggest the importance of stressing the challenge of accurate segmentation over identifying the accurate classification technique in detection of feeding gestures. Since participant-dependent models provide optimal results, a motif-based time-point fusion classification (MTFC) framework is proposed using spectral energy density, K-Spectral Centroid Clustering, symbolic aggregate approximation (SAX), a Random Forest classifier (trained on segmented motifs) and a time-point classifier fusion technique to show reliable classification of feeding gestures (75% F-measure), and a 94% accuracy of feeding gesture count in the unstructured eating experiment, resulting in a root mean square error of 2.9 feeding gestures. Mapping feeding gesture count to caloric intake, we obtain a rough estimate of whether participants overate while watching television.

正文组织架构：

1. Introduction

2. Related works

3. Devices and experiments

3.1 Wrist-worn sensor and mobile phone system

3.2 Data collection and video labeling

3.3 Highly structured test

3.4 Structured test

3.5 Unstructured test

3.6 Defining overeating episodes: Harris Benedict equation

3.7 Feeding gestures and calories

4. Methodology

4.1 Upper bound on feeding gesture classification

4.2 Data collection

4.3 Data preprocessing

4.4 Motif-based segmentation

        4.4.1 Defining dynamic energy signal

        4.4.2 Motif generation based on K-Spectral Centroid Clustering

        4.4.3 Motif matching and segmentation based on symbolic aggregate approximation (SAX)

4.5  Feature extraction and selection

4.6 Classification and feeding gesture count estimation

4.7  Defining the criterion (prediction measure)

4.8 Evaluation

4.9 Limitations

5. Results

5.1 Upper bound on feeding gestures

5.2 Motif generation

5.3 Classification results based on motif generation

5.4 Feeding gesture estimation

5.5 Overeating prediction

6. Discussion and future work

7. Conclusion

正文部分内容摘录：

1. Biological Problem: What biological problems have been solved in this paper?

2. Main discoveries: What is the main discoveries in this paper?

• showing a correlation between feeding gesture count and caloric intake in unstructured eating (r=.79, p-value=.007). 
• In this effort we prove that we can predict overeating through feeding gesture counts and the Harris Benedict principle. 

3. ML(Machine Learning) Methods: What are the ML methods applied in this paper?

• a Random Forest classifier (trained on segmented motifs) 
• 
• To detect and count feeding gestures, we designed a framework in two steps. First, we extracted motifs from exact segments (defined by ground truth) to build a database of motifs. Secondly, we perform K-Spectral Centroid Clustering to extract motif templates and perform motif matching to search for candidate feeding gesture segments. Thirdly, motif matching is performed using a symbolic aggregate approximation (SAX) method, followed by feature extraction, Random Forest classification, and a unique decision-level fusion method combining majority voting and multiple overlapping window segments to detect and count feeding gestures.
• This enabled us to obtain an upper bound on our feeding gesture classification (assuming we can detect hand-up and hand-down motion). 根据手势的注释对是否为进食进行分类?
• 如下所示的table3是所有的类别吧？数据呢，长什么样？是图片吗？那这不是类似于自然语言中的给图片添加caption的功能吗？the in-lab structured data segments 这是用于随机森林的数据集》
• 

4. ML Advantages: Why are these ML methods better than the traditional methods in these biological problems?

• We tested nine widely adopted machine learning algorithms that build generative and discriminative models from the training set including Nearest Neighbors, Gaussian Process, Decision Tree, Random Forest, Neural Net, AdaBoost, Naive Bayes, QDA, and Linear SVM. We tested a personalized model and generalized models. Personalized models were designed using a 70%/30% train/test split ratio. For generalized models, we tested 10-fold cross validation, Leave One Participant Out (LOPO), and a 70%/30% train/test split ratio. Among these classifiers, Random Forest achieved optimal results, which is consistent with prior literature 

5. Biological Significance: What is the biological significance of these ML methods’ results?

6. Prospect: What are the potential applications of these machine learning methods in biological science?

7. Mine Question(Optional)