# Publications

### 2019

 (J) Konstantinos Kyritsis, Christos Diou and Anastasios Delopoulos IEEE Journal of Biomedical and Health Informatics (JBHI), 2019 Jan [Abstract][BibTex][pdf]Overweight and obesity are both associated with in-meal eating parameters such as eating speed. Recently, the plethora of available wearable devices in the market ignited the interest of both the scientific community and the industry towards unobtrusive solutions for eating behavior monitoring. In this paper we present an algorithm for automatically detecting the in-meal food intake cycles using the inertial signals (acceleration and orientation velocity) from an off-the-shelf smartwatch. We use 5 specific wrist micromovements to model the series of actions leading to and following an intake event (i.e. bite). Food intake detection is performed in two steps. In the first step we process windows of raw sensor streams and estimate their micromovement probability distributions by means of a Convolutional Neural Network (CNN). In the second step we use a Long-Short Term Memory (LSTM) network to capture the temporal evolution and classify sequences of windows as food intake cycles. Evaluation is performed using a challenging dataset of 21 meals from 12 subjects. In our experiments we compare the performance of our algorithm against three state-of-the-art approaches, where our approach achieves the highest F1 detection score (0.913 in the Leave-One-Subject-Out experiment). The dataset used in the experiments is available at https://mug.ee.auth.gr/intake-cycle-detection/.@article{kyritsis2019modeling,author={Konstantinos Kyritsis and Christos Diou and Anastasios Delopoulos},title={Modeling Wrist Micromovements to Measure In-Meal Eating Behavior from Inertial Sensor Data},journal={IEEE Journal of Biomedical and Health Informatics (JBHI)},year={2019},month={01},date={2019-01-09},url={http://mug.ee.auth.gr/wp-content/uploads/kyritsis2019modeling.pdf},doi={http://10.1109/JBHI.2019.2892011},publisher's url={https://ieeexplore.ieee.org/abstract/document/8606156},abstract={Overweight and obesity are both associated with in-meal eating parameters such as eating speed. Recently, the plethora of available wearable devices in the market ignited the interest of both the scientific community and the industry towards unobtrusive solutions for eating behavior monitoring. In this paper we present an algorithm for automatically detecting the in-meal food intake cycles using the inertial signals (acceleration and orientation velocity) from an off-the-shelf smartwatch. We use 5 specific wrist micromovements to model the series of actions leading to and following an intake event (i.e. bite). Food intake detection is performed in two steps. In the first step we process windows of raw sensor streams and estimate their micromovement probability distributions by means of a Convolutional Neural Network (CNN). In the second step we use a Long-Short Term Memory (LSTM) network to capture the temporal evolution and classify sequences of windows as food intake cycles. Evaluation is performed using a challenging dataset of 21 meals from 12 subjects. In our experiments we compare the performance of our algorithm against three state-of-the-art approaches, where our approach achieves the highest F1 detection score (0.913 in the Leave-One-Subject-Out experiment). The dataset used in the experiments is available at https://mug.ee.auth.gr/intake-cycle-detection/.}}

### 2019

 (J) Janet van den Boer, Annemiek van der Lee, Lingchuan Zhou, Vasileios Papapanagiotou, Christos Diou, Anastasios Delopoulos and Monica Mars The SPLENDID Eating Detection Sensor: Development and Feasibility Study, 6, (9), pp. 170, 2018 Sep [Abstract][BibTex]The available methods for monitoring food intake---which for a great part rely on self-report---often provide biased and incomplete data. Currently, no good technological solutions are available. Hence, the SPLENDID eating detection sensor (an ear-worn device with an air microphone and a photoplethysmogram [PPG] sensor) was developed to enable complete and objective measurements of eating events. The technical performance of this device has been described before. To date, literature is lacking a description of how such a device is perceived and experienced by potential users. Objective: The objective of our study was to explore how potential users perceive and experience the SPLENDID eating detection sensor. Methods: Potential users evaluated the eating detection sensor at different stages of its development: (1) At the start, 12 health professionals (eg, dieticians, personal trainers) were interviewed and a focus group was held with 5 potential end users to find out their thoughts on the concept of the eating detection sensor. (2) Then, preliminary prototypes of the eating detection sensor were tested in a laboratory setting where 23 young adults reported their experiences. (3) Next, the first wearable version of the eating detection sensor was tested in a semicontrolled study where 22 young, overweight adults used the sensor on 2 separate days (from lunch till dinner) and reported their experiences. (4) The final version of the sensor was tested in a 4-week feasibility study by 20 young, overweight adults who reported their experiences. Results: Throughout all the development stages, most individuals were enthusiastic about the eating detection sensor. However, it was stressed multiple times that it was critical that the device be discreet and comfortable to wear for a longer period. In the final study, the eating detection sensor received an average grade of 3.7 for wearer comfort on a scale of 1 to 10. Moreover, experienced discomfort was the main reason for wearing the eating detection sensor <2 hours a day. The participants reported having used the eating detection sensor on 19/28 instructed days on average. Conclusions: The SPLENDID eating detection sensor, which uses an air microphone and a PPG sensor, is a promising new device that can facilitate the collection of reliable food intake data, as shown by its technical potential. Potential users are enthusiastic, but to be successful wearer comfort and discreetness of the device need to be improved.@article{2018Boer,author={Janet van den Boer and Annemiek van der Lee and Lingchuan Zhou and Vasileios Papapanagiotou and Christos Diou and Anastasios Delopoulos and Monica Mars},title={The SPLENDID Eating Detection Sensor: Development and Feasibility Study},journal={The SPLENDID Eating Detection Sensor: Development and Feasibility Study},volume={6},number={9},pages={170},year={2018},month={09},date={2018-09-04},doi={https://doi.org/10.2196/mhealth.9781},issn={2291-5222},publisher's url={http://mhealth.jmir.org/2018/9/e170/},abstract={The available methods for monitoring food intake---which for a great part rely on self-report---often provide biased and incomplete data. Currently, no good technological solutions are available. Hence, the SPLENDID eating detection sensor (an ear-worn device with an air microphone and a photoplethysmogram [PPG] sensor) was developed to enable complete and objective measurements of eating events. The technical performance of this device has been described before. To date, literature is lacking a description of how such a device is perceived and experienced by potential users. Objective: The objective of our study was to explore how potential users perceive and experience the SPLENDID eating detection sensor. Methods: Potential users evaluated the eating detection sensor at different stages of its development: (1) At the start, 12 health professionals (eg, dieticians, personal trainers) were interviewed and a focus group was held with 5 potential end users to find out their thoughts on the concept of the eating detection sensor. (2) Then, preliminary prototypes of the eating detection sensor were tested in a laboratory setting where 23 young adults reported their experiences. (3) Next, the first wearable version of the eating detection sensor was tested in a semicontrolled study where 22 young, overweight adults used the sensor on 2 separate days (from lunch till dinner) and reported their experiences. (4) The final version of the sensor was tested in a 4-week feasibility study by 20 young, overweight adults who reported their experiences. Results: Throughout all the development stages, most individuals were enthusiastic about the eating detection sensor. However, it was stressed multiple times that it was critical that the device be discreet and comfortable to wear for a longer period. In the final study, the eating detection sensor received an average grade of 3.7 for wearer comfort on a scale of 1 to 10. Moreover, experienced discomfort was the main reason for wearing the eating detection sensor <2 hours a day. The participants reported having used the eating detection sensor on 19/28 instructed days on average. Conclusions: The SPLENDID eating detection sensor, which uses an air microphone and a PPG sensor, is a promising new device that can facilitate the collection of reliable food intake data, as shown by its technical potential. Potential users are enthusiastic, but to be successful wearer comfort and discreetness of the device need to be improved.}} (J) Christos Diou, Pantelis Lelekas and Anastasios Delopoulos Journal of Imaging, 4, (11), pp. 125, 2018 Oct [Abstract][BibTex]Background: Evidence-based policymaking requires data about the local population’s socioeconomic status (SES) at detailed geographical level, however, such information is often not available, or is too expensive to acquire. Researchers have proposed solutions to estimate SES indicators by analyzing Google Street View images, however, these methods are also resource-intensive, since they require large volumes of manually labeled training data. (2) Methods: We propose a methodology for automatically computing surrogate variables of SES indicators using street images of parked cars and deep multiple instance learning. Our approach does not require any manually created labels, apart from data already available by statistical authorities, while the entire pipeline for image acquisition, parked car detection, car classification, and surrogate variable computation is fully automated. The proposed surrogate variables are then used in linear regression models to estimate the target SES indicators. (3) Results: We implement and evaluate a model based on the proposed surrogate variable at 30 municipalities of varying SES in Greece. Our model has R2=0.76 and a correlation coefficient of 0.874 with the true unemployment rate, while it achieves a mean absolute percentage error of 0.089 and mean absolute error of 1.87 on a held-out test set. Similar results are also obtained for other socioeconomic indicators, related to education level and occupational prestige. (4) Conclusions: The proposed methodology can be used to estimate SES indicators at the local level automatically, using images of parked cars detected via Google Street View, without the need for any manual labeling effort@article{Diou2018JI,author={Christos Diou and Pantelis Lelekas and Anastasios Delopoulos},title={Image-Based Surrogates of Socio-Economic Status in Urban Neighborhoods Using Deep Multiple Instance Learning},journal={Journal of Imaging},volume={4},number={11},pages={125},year={2018},month={10},date={2018-10-23},doi={http://10.3390/jimaging4110125},issn={2313-433X},publisher's url={https://www.mdpi.com/2313-433X/4/11/125},abstract={Background: Evidence-based policymaking requires data about the local population’s socioeconomic status (SES) at detailed geographical level, however, such information is often not available, or is too expensive to acquire. Researchers have proposed solutions to estimate SES indicators by analyzing Google Street View images, however, these methods are also resource-intensive, since they require large volumes of manually labeled training data. (2) Methods: We propose a methodology for automatically computing surrogate variables of SES indicators using street images of parked cars and deep multiple instance learning. Our approach does not require any manually created labels, apart from data already available by statistical authorities, while the entire pipeline for image acquisition, parked car detection, car classification, and surrogate variable computation is fully automated. The proposed surrogate variables are then used in linear regression models to estimate the target SES indicators. (3) Results: We implement and evaluate a model based on the proposed surrogate variable at 30 municipalities of varying SES in Greece. Our model has R2=0.76 and a correlation coefficient of 0.874 with the true unemployment rate, while it achieves a mean absolute percentage error of 0.089 and mean absolute error of 1.87 on a held-out test set. Similar results are also obtained for other socioeconomic indicators, related to education level and occupational prestige. (4) Conclusions: The proposed methodology can be used to estimate SES indicators at the local level automatically, using images of parked cars detected via Google Street View, without the need for any manual labeling effort}} (J) Maryam Esfandiari, Vasilis Papapanagiotou, Christos Diou, Modjtaba Zandian, Jenny Nolstam, Per Södersten and Cecilia Bergh JoVE, (135), 2018 May [Abstract][BibTex]Subjects eat food from a plate that sits on a scale connected to a computer that records the weight loss of the plate during the meal and makes up a curve of food intake, meal duration and rate of eating modeled by a quadratic equation. The purpose of the method is to change eating behavior by providing visual feedback on the computer screen that the subject can adapt to because her/his own rate of eating appears on the screen during the meal. The data generated by the method is automatically analyzed and fitted to the quadratic equation using a custom made algorithm. The method has the advantage of recording eating behavior objectively and offers the possibility of changing eating behavior both in experiments and in clinical practice. A limitation may be that experimental subjects are affected by the method. The same limitation may be an advantage in clinical practice, as eating behavior is more easily stabilized by the method. A treatment that uses this method has normalized body weight and restored the health of several hundred patients with anorexia nervosa and other eating disorders and has reduced the weight and improved the health of severely overweight patients.@article{Esfandiari2018,author={Maryam Esfandiari and Vasilis Papapanagiotou and Christos Diou and Modjtaba Zandian and Jenny Nolstam and Per Södersten and Cecilia Bergh},title={Control of Eating Behavior Using a Novel Feedback System},journal={JoVE},number={135},year={2018},month={05},date={2018-05-08},doi={http://10.3791/57432},publisher's url={https://www.jove.com/video/57432/control-of-eating-behavior-using-a-novel-feedback-system},abstract={Subjects eat food from a plate that sits on a scale connected to a computer that records the weight loss of the plate during the meal and makes up a curve of food intake, meal duration and rate of eating modeled by a quadratic equation. The purpose of the method is to change eating behavior by providing visual feedback on the computer screen that the subject can adapt to because her/his own rate of eating appears on the screen during the meal. The data generated by the method is automatically analyzed and fitted to the quadratic equation using a custom made algorithm. The method has the advantage of recording eating behavior objectively and offers the possibility of changing eating behavior both in experiments and in clinical practice. A limitation may be that experimental subjects are affected by the method. The same limitation may be an advantage in clinical practice, as eating behavior is more easily stabilized by the method. A treatment that uses this method has normalized body weight and restored the health of several hundred patients with anorexia nervosa and other eating disorders and has reduced the weight and improved the health of severely overweight patients.}} (J) George Mamalakis, Christos Diou, Andreas Symeonidis and Leonidas Georgiadis Neural Computing and Applications, 2018 Jul [Abstract][BibTex]In this work, we propose a methodology for reducing false alarms in file system intrusion detection systems, by taking into account the daemon's file system footprint. More specifically, we experimentally show that sequences of outliers can serve as a distinguishing characteristic between true and false positives, and we show how analysing sequences of outliers can lead to lower false positive rates, while maintaining high detection rates. Based on this analysis, we developed an anomaly detection filter that learns outlier sequences using k-nearest neighbours with normalised longest common subsequence. Outlier sequences are then used as a filter to reduce false positives on the {\$}{\$}FI^2DS{\$}{\$}FI2DSfile system intrusion detection system. This filter is evaluated on both overlapping and non-overlapping sequences of outliers. In both cases, experiments performed on three real-world web servers and a honeynet show that our approach achieves significant false positive reduction rates (up to 50 times), without any degradation of the corresponding true positive detection rates.@article{Mamalakis2018,author={George Mamalakis and Christos Diou and Andreas Symeonidis and Leonidas Georgiadis},title={Of daemons and men: reducing false positive rate in intrusion detection systems with file system footprint analysis},journal={Neural Computing and Applications},year={2018},month={07},date={2018-07-05},doi={http://10.1007/s00521-018-3550-x},issn={1433-3058},publisher's url={https://link.springer.com/article/10.1007/s00521-018-3550-x},abstract={In this work, we propose a methodology for reducing false alarms in file system intrusion detection systems, by taking into account the daemon\'s file system footprint. More specifically, we experimentally show that sequences of outliers can serve as a distinguishing characteristic between true and false positives, and we show how analysing sequences of outliers can lead to lower false positive rates, while maintaining high detection rates. Based on this analysis, we developed an anomaly detection filter that learns outlier sequences using k-nearest neighbours with normalised longest common subsequence. Outlier sequences are then used as a filter to reduce false positives on the {\\$}{\\$}FI^2DS{\\$}{\\$}FI2DSfile system intrusion detection system. This filter is evaluated on both overlapping and non-overlapping sequences of outliers. In both cases, experiments performed on three real-world web servers and a honeynet show that our approach achieves significant false positive reduction rates (up to 50 times), without any degradation of the corresponding true positive detection rates.}} (J) Ioannis Sarafis, Christos Diou and Anastasios Delopoulos CoRR, abs/1809.06124, 2018 Sep [Abstract][BibTex][pdf]Weighted SVM (or fuzzy SVM) is the most widely used SVM variant owning its effectiveness to the use of instance weights. Proper selection of the instance weights can lead to increased generalization performance. In this work, we extend the span error bound theory to weighted SVM and we introduce effective hyperparameter selection methods for the weighted SVM algorithm. The significance of the presented work is that enables the application of span bound and span-rule with weighted SVM. The span bound is an upper bound of the leave-one-out error that can be calculated using a single trained SVM model. This is important since leave-one-out error is an almost unbiased estimator of the test error. Similarly, the span-rule gives the actual value of the leave-one-out error. Thus, one can apply span bound and span-rule as computationally lightweight alternatives of leave-one-out procedure for hyperparameter selection. The main theoretical contributions are: (a) we prove the necessary and sufficient condition for the existence of the span of a support vector in weighted SVM; and (b) we prove the extension of span bound and span-rule to weighted SVM. We experimentally evaluate the span bound and the span-rule for hyperparameter selection and we compare them with other methods that are applicable to weighted SVM: the K-fold cross-validation and the $\xi - \alpha$ bound. Experiments on 14 benchmark data sets and data sets with importance scores for the training instances show that: (a) the condition for the existence of span in weighted SVM is satisfied almost always; (b) the span-rule is the most effective method for weighted SVM hyperparameter selection; (c) the span-rule is the best predictor of the test error in the mean square error sense; and (d) the span-rule is efficient and, for certain problems, it can be calculated faster than K-fold cross-validation.@article{Sarafis2018CoRR,author={Ioannis Sarafis and Christos Diou and Anastasios Delopoulos},title={Span error bound for weighted SVM with applications in hyperparameter selection (preprint)},journal={CoRR},volume={abs/1809.06124},year={2018},month={09},date={2018-09-17},url={https://arxiv.org/pdf/1809.06124.pdf},publisher's url={https://arxiv.org/abs/1809.06124},abstract={Weighted SVM (or fuzzy SVM) is the most widely used SVM variant owning its effectiveness to the use of instance weights. Proper selection of the instance weights can lead to increased generalization performance. In this work, we extend the span error bound theory to weighted SVM and we introduce effective hyperparameter selection methods for the weighted SVM algorithm. The significance of the presented work is that enables the application of span bound and span-rule with weighted SVM. The span bound is an upper bound of the leave-one-out error that can be calculated using a single trained SVM model. This is important since leave-one-out error is an almost unbiased estimator of the test error. Similarly, the span-rule gives the actual value of the leave-one-out error. Thus, one can apply span bound and span-rule as computationally lightweight alternatives of leave-one-out procedure for hyperparameter selection. The main theoretical contributions are: (a) we prove the necessary and sufficient condition for the existence of the span of a support vector in weighted SVM; and (b) we prove the extension of span bound and span-rule to weighted SVM. We experimentally evaluate the span bound and the span-rule for hyperparameter selection and we compare them with other methods that are applicable to weighted SVM: the K-fold cross-validation and the $\\xi - \\alpha$ bound. Experiments on 14 benchmark data sets and data sets with importance scores for the training instances show that: (a) the condition for the existence of span in weighted SVM is satisfied almost always; (b) the span-rule is the most effective method for weighted SVM hyperparameter selection; (c) the span-rule is the best predictor of the test error in the mean square error sense; and (d) the span-rule is efficient and, for certain problems, it can be calculated faster than K-fold cross-validation.}} (J) Vasilis Papapanagiotou, Christos Diou, Ioannis Ioakimidis, Per Sodersten and Anastasios Delopoulos IEEE Journal of Biomedical and Health Informatics, PP, (99), pp. 1-1, 2018 Mar [Abstract][BibTex][pdf]The structure of the cumulative food intake (CFI) curve has been associated with obesity and eating disorders. Scales that record the weight loss of a plate from which a subject eats food are used for capturing this curve; however, their measurements are contaminated by additive noise and are distorted by certain types of artifacts. This paper presents an algorithm for automatically processing continuous in-meal weight measurements in order to extract the clean CFI curve and in-meal eating indicators, such as total food intake and food intake rate. The algorithm relies on the representation of the weight-time series by a string of symbols that correspond to events such as bites or food additions. A context-free grammar is next used to model a meal as a sequence of such events. The selection of the most likely parse tree is finally used to determine the predicted eating sequence. The algorithm is evaluated on a dataset of 113 meals collected using the Mandometer, a scale that continuously samples plate weight during eating. We evaluate the effectiveness for seven indicators, and for bite-instance detection. We compare our approach with three state-of-the-art algorithms, and achieve the lowest error rates for most indicators (24 g for total meal weight). The proposed algorithm extracts the parameters of the CFI curve automatically, eliminating the need for manual data processing, and thus facilitating large-scale studies of eating behavior.@article{Vassilis2018,author={Vasilis Papapanagiotou and Christos Diou and Ioannis Ioakimidis and Per Sodersten and Anastasios Delopoulos},title={Automatic analysis of food intake and meal microstructure based on continuous weight measurements},journal={IEEE Journal of Biomedical and Health Informatics},volume={PP},number={99},pages={1-1},year={2018},month={03},date={2018-03-05},url={http://mug.ee.auth.gr/wp-content/uploads/papapanagiotou2018automated.pdf},doi={http://10.1109/JBHI.2018.2812243},publisher's url={https://ieeexplore.ieee.org/document/8306871/},abstract={The structure of the cumulative food intake (CFI) curve has been associated with obesity and eating disorders. Scales that record the weight loss of a plate from which a subject eats food are used for capturing this curve; however, their measurements are contaminated by additive noise and are distorted by certain types of artifacts. This paper presents an algorithm for automatically processing continuous in-meal weight measurements in order to extract the clean CFI curve and in-meal eating indicators, such as total food intake and food intake rate. The algorithm relies on the representation of the weight-time series by a string of symbols that correspond to events such as bites or food additions. A context-free grammar is next used to model a meal as a sequence of such events. The selection of the most likely parse tree is finally used to determine the predicted eating sequence. The algorithm is evaluated on a dataset of 113 meals collected using the Mandometer, a scale that continuously samples plate weight during eating. We evaluate the effectiveness for seven indicators, and for bite-instance detection. We compare our approach with three state-of-the-art algorithms, and achieve the lowest error rates for most indicators (24 g for total meal weight). The proposed algorithm extracts the parameters of the CFI curve automatically, eliminating the need for manual data processing, and thus facilitating large-scale studies of eating behavior.}}