Obesity and nutrition-related diseases are currently growing challenges for medicine. A precise and timesaving method for food intake monitoring is needed. For this purpose, an approach based on the classification of sounds produced during food intake is presented. Sounds are recorded non-invasively by miniature microphones in the outer ear canal. A database of 51 participants eating seven types of food and consuming one drink has been developed for algorithm development and model training. The database is labeled manually using a protocol with introductions for annotation. The annotation procedure is evaluated using Cohen's kappa coefficient. The food intake activity is detected by the comparison of the signal energy of in-ear sounds to environmental sounds recorded by a reference microphone. Hidden Markov models are used for the recognition of single chew or swallowing events. Intake cycles are modeled as event sequences in finite-state grammars. Classification of consumed food is realized by a finite-state grammar decoder based on the Viterbi algorithm. We achieved a detection accuracy of 83% and a food classification accuracy of 79% on a test set of 10% of all records. Our approach faces the need of monitoring the time and occurrence of eating. With differentiation of consumed food, a first step toward the goal of meal weight estimation is taken.

The analysis of the food intake behavior has the potential to provide insights into the development of obesity and eating disorders. As an elementary part of this analysis, chewing strokes have to be detected and counted. Our approach for food intake analysis is the evaluation of chewing sounds generated during the process of eating. These sounds were recorded by microphones applied to the outer ear canal of the user. Eight different algorithms for automated chew event detection were presented and evaluated on two datasets. The first dataset contained food intake sounds from the consumption of six types of food. The second dataset consisted of recordings of different environmental sounds. These datasets contained 68,094 chew events in around 18 h recording data. The results of the automated chew event detection were compared to manual annotations. Precision and recall over 80% were achieved by most of the algorithms. A simple noise reduction algorithm using spectral subtraction was implemented for signal enhancement. Its benefit on the chew event detection performance was evaluated. A reduction of the number of false detections by 28% on average was achieved by maintaining the detection performance. The system is able to be used for calculation of the chewing frequency in laboratory settings.

No abstract

Analyzing food intake behavior is necessary to prevent obesity and overweight. Detecting and counting chewing strokes is an elementary part of this analysis. In our project, sounds of food intake were recorded using a microphone in the outer ear canal. The records contained sounds of 51 healthy subjects chewing 8 types of food. We evaluated seven different algorithms to detect chew events in sound records. Results of the automated detection were compared to manual annotations. Best performances (precision and recall over 76 %) were achieved by detecting chew events in six different frequency bands and fusing these results. With this method for counting the number of chews, an important step towards the estimation of bite weight has been done

There is a need of experimental studies on biomarkers in patients with anorexia nervosa (P AN), especially in the context of stress, in order to foster understanding in illness maintenance. To this end, the cortisol response to an acute stressor was investigated in n = 26 P AN (BMI: 19.3 ± 3.4 kg/m 2), age, and gender matched to n = 26 healthy controls (HC; BMI: 23.08 ± 3.3 kg/m 2). For this purpose, salivary cortisol parameters were assessed in two experimental conditions: (1) rest/ no intervention and (2) stress intervention (TSST; Trier Social Stress Test). In addition, psychological indicators of stress were assessed (Primary Appraisal Secondary Appraisal, Visual Analogue Scale, and Trier Inventory for the assessment of Chronic Stress), as well as psychological distress, depression, and eating disorder (ED) symptoms. A 2 × 2 × 8 ANOVA demonstrated elevated cortisol levels in P AN in the resting condition. In the stress intervention no significant group effect in terms of cortisol (F (1, 50) = 0.69; p = 0.410; η 2 p ¼ 0:014). A significant condition (F (1, 50) = 20.50; p = 0.000; η 2 p ¼ 0:291) and time effect (F(2.71, 135.44) = 11.27; p = 0.000; η 2 p ¼ 0:20) were revealed, as well as two significant interaction effects. First: Condition × group (F (1, 50) = 4.17, p = 0.046; η 2 p ¼ 0:077) and second: Condition × time (F (2.71, 135.44) = 16.07, p = 0.000, η 2 p ¼ 0:24:). In terms of AUC G , no significant differences between both groups were exhibited. Regardless, significant results were evinced in terms of an increase (AUC i : F(1, 50) = 20.66, p = 0.015, η 2 p ¼ 0:113), baseline to peak (+20 min post-TSST: t 5 = 16.51 (9.02), p = 0.029) and reactivity (M PAN = 0.73 vs. M HC = 4.25, p = 0.036). In addition, a significant correlation between AUC G and BMI: r (24) = −0.42, p = 0.027 was demonstrated, but not between AUC i and BMI (r (24) = −0.26, p = 0.20). Psychological indices suggested higher levels of chronic and perceived stress in P AN relative to HC. However, stress perception in the stress condition (VAS) was comparable. Additional analyses demonstrated that ED-symptoms are highly correlated with psychological distress and depression, but not with BMI. In addition, it could be demonstrated that reactivity is rather related to EDsymptoms and psychological burden than to BMI. In conclusion, P AN showed elevated basal cortisol levels at rest and exhibited a blunted cortisol reactivity to the TSST as evinced by salivary cortisol parameters. Further, it was shown that weight recovery influences reversibility of hypercortisolemia, i.e., cortisol levels normalize with weight gain. However, HPAA (hypothalamus-pituitary-adrenal axis) irregularities in terms of reactivity persist even at a BMI ≤ 19.3 (±3.4). Our data suggest that pronounced psychological burden in P AN , have a greater impact on the HPAA functionality (secondary to the ED) than BMI itself.