Mandy Korpusik
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Mandy Korpusik has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- James Glass
Massachusetts Institute of Technology Cambridge MA USA
4 shared publications - Cheryl H. Gilhooly
JM USDA Human Nutrition Center on Aging, Tufts University Boston MA
2 shared publications - Ryan B. Simpson
Friedman School of Nutrition Science and Policy, Tufts University, Boston, MA 02111, USA
2 shared publications - Sai Krupa Das
Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University Boston Massachusetts USA
2 shared publications - Salima F. Taylor
Friedman School of Nutrition Science and Policy, Tufts University, Boston, USA
2 shared publications - Susan B. Roberts
Geisel School of Medicine, Dartmouth College, Hanover, USA
2 shared publications
Latest Publications
- Open Access
- 30 July 2025
- Published by JMIR Publications Inc. in JMIR Human Factors
- Vol. 13, e81439
A Smart Phone App to Improve Intuitive Eating and Diet Quality: Design and Usability Study (Preprint)
Web-based and mobile phone-based apps have become widely available for dietary self-monitoring; however, their use may increase the risk of disordered eating. College students frequently demonstrate poor nutrient intake despite consumption of sufficient calories. One way to improve diet quality may be via the use of a smartphone app that encourages intuitive eating. The purpose of this study was to improve diet quality among college students through the use of a novel smartphone app that promotes intuitive eating rather than calorie counting and weight loss. The In2Eat iOS mobile app was developed in SwiftUI and stored user data in a Firebase database. A total of 45 college students completed assessments of intuitive eating, diet quality, and disordered eating before and after 4 weeks of using the In2Eat app. Users evaluated the usability of the app with the System Usability Scale (SUS). Engagement with the app was recorded as the total number of days a meal was logged, the total number of meals logged, and the average number of meals logged per day. After our 4-week intervention, dietary qualities that protect against chronic disease increased by 28%, fruit consumption increased by 63%, and skin antioxidant levels increased by 6.1% (Hedges g=0.16; mean difference 0.33, 95% bias corrected and accelerated [BCa] CI 0.04-0.61; P=.03). Global intuitive eating did not change during the user study; however, the unconditional permission to eat subscale increased (Hedges g=-0.28; mean difference 0.28, 95% BCa CI 0.07-0.49; P=.01, adjusted P=.07). Overall, disordered eating also did not change with app use, although dietary restraint decreased (Hedges g=-0.23; mean difference 0.30, 95% BCa CI -0.61 to -0.04; P=.04, adjusted P=.22). The average SUS score for the In2Eat app was 67.2 (SD 15.5). The number of days a meal was logged was positively correlated with SUS scores (r=0.28; P=.06), and the total number of meals logged had a monotonic association with app usability (ρ=0.31; P=.04). When divided according to the low (mean 10.2, SD 5.3), medium (mean 26.3, SD 2.8), and high (mean 33.6, SD 3.8) number of days logging meals, participants with higher days of logging reported the app as more usable (H=6.75; P=.03). A regression analysis showed that 8% of the variance in system usability (R2=0.080; P=.31) was explained by app use; however, none of the individual predictors contributed substantially to the variance. An intuitive eating smartphone app can improve diet quality without increasing disordered eating. Results suggest that participants who logged more meals tended to rate the app as more usable. Further research is needed with a greater sample size after incorporating features to improve the app's usability.
- 30 July 2025
- Published by JMIR Publications Inc. in JMIR Human Factors
- Vol. 13, e81439
Open AccessA Smart Phone App to Improve Intuitive Eating and Diet Quality: Design and Usability Study (Preprint)
Web-based and mobile phone-based apps have become widely available for dietary self-monitoring; however, their use may increase the risk of disordered eating. College students frequently demonstrate poor nutrient intake despite consumption of sufficient calories. One way to improve diet quality may be via the use of a smartphone app that encourages intuitive eating. The purpose of this study was to improve diet quality among college students through the use of a novel smartphone app that promotes intuitive eating rather than calorie counting and weight loss. The In2Eat iOS mobile app was developed in SwiftUI and stored user data in a Firebase database. A total of 45 college students completed assessments of intuitive eating, diet quality, and disordered eating before and after 4 weeks of using the In2Eat app. Users evaluated the usability of the app with the System Usability Scale (SUS). Engagement with the app was recorded as the total number of days a meal was logged, the total number of meals logged, and the average number of meals logged per day. After our 4-week intervention, dietary qualities that protect against chronic disease increased by 28%, fruit consumption increased by 63%, and skin antioxidant levels increased by 6.1% (Hedges g=0.16; mean difference 0.33, 95% bias corrected and accelerated [BCa] CI 0.04-0.61; P=.03). Global intuitive eating did not change during the user study; however, the unconditional permission to eat subscale increased (Hedges g=-0.28; mean difference 0.28, 95% BCa CI 0.07-0.49; P=.01, adjusted P=.07). Overall, disordered eating also did not change with app use, although dietary restraint decreased (Hedges g=-0.23; mean difference 0.30, 95% BCa CI -0.61 to -0.04; P=.04, adjusted P=.22). The average SUS score for the In2Eat app was 67.2 (SD 15.5). The number of days a meal was logged was positively correlated with SUS scores (r=0.28; P=.06), and the total number of meals logged had a monotonic association with app usability (ρ=0.31; P=.04). When divided according to the low (mean 10.2, SD 5.3), medium (mean 26.3, SD 2.8), and high (mean 33.6, SD 3.8) number of days logging meals, participants with higher days of logging reported the app as more usable (H=6.75; P=.03). A regression analysis showed that 8% of the variance in system usability (R2=0.080; P=.31) was explained by app use; however, none of the individual predictors contributed substantially to the variance. An intuitive eating smartphone app can improve diet quality without increasing disordered eating. Results suggest that participants who logged more meals tended to rate the app as more usable. Further research is needed with a greater sample size after incorporating features to improve the app's usability.
- Open Access
- 6 December 2021
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 23 (12) , e26988
Use of Natural Spoken Language With Automated Mapping of Self-reported Food Intake to Food Composition Data for Low-Burden Real-time Dietary Assessment: Method Comparison Study
Background: Self-monitoring food intake is a cornerstone of national recommendations for health, but existing apps for this purpose are burdensome for users and researchers, which limits use. Objective: We developed and pilot tested a new app (COCO Nutritionist) that combines speech understanding technology with technologies for mapping foods to appropriate food composition codes in national databases, for lower-burden and automated nutritional analysis of self-reported dietary intake. Methods: COCO was compared with the multiple-pass, interviewer-administered 24-hour recall method for assessment of energy intake. COCO was used for 5 consecutive days, and 24-hour dietary recalls were obtained for two of the days. Participants were 35 women and men with a mean age of 28 (range 20-58) years and mean BMI of 24 (range 17-48) kg/m2. Results: There was no significant difference in energy intake between values obtained by COCO and 24-hour recall for days when both methods were used (mean 2092, SD 1044 kcal versus mean 2030, SD 687 kcal, P=.70). There were also no significant differences between the methods for percent of energy from protein, carbohydrate, and fat (P=.27-.89), and no trend in energy intake obtained with COCO over the entire 5-day study period (P=.19). Conclusions: This first demonstration of a dietary assessment method using natural spoken language to map reported foods to food composition codes demonstrates a promising new approach to automate assessments of dietary intake.
- 6 December 2021
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 23 (12) , e26988
Open AccessUse of Natural Spoken Language With Automated Mapping of Self-reported Food Intake to Food Composition Data for Low-Burden Real-time Dietary Assessment: Method Comparison Study
Background: Self-monitoring food intake is a cornerstone of national recommendations for health, but existing apps for this purpose are burdensome for users and researchers, which limits use. Objective: We developed and pilot tested a new app (COCO Nutritionist) that combines speech understanding technology with technologies for mapping foods to appropriate food composition codes in national databases, for lower-burden and automated nutritional analysis of self-reported dietary intake. Methods: COCO was compared with the multiple-pass, interviewer-administered 24-hour recall method for assessment of energy intake. COCO was used for 5 consecutive days, and 24-hour dietary recalls were obtained for two of the days. Participants were 35 women and men with a mean age of 28 (range 20-58) years and mean BMI of 24 (range 17-48) kg/m2. Results: There was no significant difference in energy intake between values obtained by COCO and 24-hour recall for days when both methods were used (mean 2092, SD 1044 kcal versus mean 2030, SD 687 kcal, P=.70). There were also no significant differences between the methods for percent of energy from protein, carbohydrate, and fat (P=.27-.89), and no trend in energy intake obtained with COCO over the entire 5-day study period (P=.19). Conclusions: This first demonstration of a dietary assessment method using natural spoken language to map reported foods to food composition codes demonstrates a promising new approach to automate assessments of dietary intake.
- Open Access
- 27 January 2021
- Published by JMIR Publications Inc. in JMIR Preprints
Use of Natural Spoken Language With Automated Mapping of Self-reported Food Intake to Food Composition Data for Low-Burden Real-time Dietary Assessment: Method Comparison Study (Preprint)
BACKGROUND Self-monitoring food intake is a cornerstone of national recommendations for health, but existing apps for this purpose are burdensome for users and researchers, which limits use. OBJECTIVE We developed and pilot tested a new app (COCO Nutritionist) that combines speech understanding technology with technologies for mapping foods to appropriate food composition codes in national databases, for lower-burden and automated nutritional analysis of self-reported dietary intake. METHODS COCO was compared with the multiple-pass, interviewer-administered 24-hour recall method for assessment of energy intake. COCO was used for 5 consecutive days, and 24-hour dietary recalls were obtained for two of the days. Participants were 35 women and men with a mean age of 28 (range 20-58) years and mean BMI of 24 (range 17-48) kg/m2. RESULTS There was no significant difference in energy intake between values obtained by COCO and 24-hour recall for days when both methods were used (mean 2092, SD 1044 kcal versus mean 2030, SD 687 kcal, P=.70). There were also no significant differences between the methods for percent of energy from protein, carbohydrate, and fat (P=.27-.89), and no trend in energy intake obtained with COCO over the entire 5-day study period (P=.19). CONCLUSIONS This first demonstration of a dietary assessment method using natural spoken language to map reported foods to food composition codes demonstrates a promising new approach to automate assessments of dietary intake.
- 27 January 2021
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessUse of Natural Spoken Language With Automated Mapping of Self-reported Food Intake to Food Composition Data for Low-Burden Real-time Dietary Assessment: Method Comparison Study (Preprint)
BACKGROUND Self-monitoring food intake is a cornerstone of national recommendations for health, but existing apps for this purpose are burdensome for users and researchers, which limits use. OBJECTIVE We developed and pilot tested a new app (COCO Nutritionist) that combines speech understanding technology with technologies for mapping foods to appropriate food composition codes in national databases, for lower-burden and automated nutritional analysis of self-reported dietary intake. METHODS COCO was compared with the multiple-pass, interviewer-administered 24-hour recall method for assessment of energy intake. COCO was used for 5 consecutive days, and 24-hour dietary recalls were obtained for two of the days. Participants were 35 women and men with a mean age of 28 (range 20-58) years and mean BMI of 24 (range 17-48) kg/m2. RESULTS There was no significant difference in energy intake between values obtained by COCO and 24-hour recall for days when both methods were used (mean 2092, SD 1044 kcal versus mean 2030, SD 687 kcal, P=.70). There were also no significant differences between the methods for percent of energy from protein, carbohydrate, and fat (P=.27-.89), and no trend in energy intake obtained with COCO over the entire 5-day study period (P=.19). CONCLUSIONS This first demonstration of a dietary assessment method using natural spoken language to map reported foods to food composition codes demonstrates a promising new approach to automate assessments of dietary intake.
- 23 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 27 (8) , 1321-1334
Deep Learning for Database Mapping and Asking Clarification Questions in Dialogue Systems
A dialogue system will often ask followup clarification questions when interacting with a user if the agent is unsure how to respond. In this new work, we explore deep reinforcement learning (RL) for asking followup questions when a user records a meal description, and the system needs to narrow down the options for which foods the person has eaten. We build off of prior work in which we use novel convolutional neural network (CNN) models to bypass the standard feature engineering used in dialogue systems to handle the text mismatch between natural language user queries and structured database entries, demonstrating that our model learns semantically meaningful embedding representations of natural language. In this new nutrition domain, the followup clarification questions consist of possible attributes for each food that was consumed; for example, if the user drinks a cup of milk, the system should ask about the percent milkfat. We investigate an RL agent to dynamically follow up with the user, which we compare to rule-based and entropy-based methods. On a held-out test set, assuming the followup questions are answered correctly, deep RL significantly boosts top-5 food recall from 54.9% without followup to 89.0%. We also demonstrate that a hybrid RL model achieves the best perceived naturalness ratings in a human evaluation.
- 23 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 27 (8) , 1321-1334
Deep Learning for Database Mapping and Asking Clarification Questions in Dialogue Systems
A dialogue system will often ask followup clarification questions when interacting with a user if the agent is unsure how to respond. In this new work, we explore deep reinforcement learning (RL) for asking followup questions when a user records a meal description, and the system needs to narrow down the options for which foods the person has eaten. We build off of prior work in which we use novel convolutional neural network (CNN) models to bypass the standard feature engineering used in dialogue systems to handle the text mismatch between natural language user queries and structured database entries, demonstrating that our model learns semantically meaningful embedding representations of natural language. In this new nutrition domain, the followup clarification questions consist of possible attributes for each food that was consumed; for example, if the user drinks a cup of milk, the system should ask about the percent milkfat. We investigate an RL agent to dynamically follow up with the user, which we compare to rule-based and entropy-based methods. On a held-out test set, assuming the followup questions are answered correctly, deep RL significantly boosts top-5 food recall from 54.9% without followup to 89.0%. We also demonstrate that a hybrid RL model achieves the best perceived naturalness ratings in a human evaluation.
- 17 April 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 25 (7) , 1450-1461
Spoken Language Understanding for a Nutrition Dialogue System
Food logging is recommended by dieticians for prevention and treatment of obesity, but currently available mobile applications for diet tracking are often too difficult and time-consuming for patients to use regularly. For this reason, we propose a novel approach to food journaling that uses speech and language understanding technology in order to enable efficient self-assessment of energy and nutrient consumption. This paper presents ongoing language understanding experiments conducted as part of a larger effort to create a nutrition dialogue system that automatically extracts food concepts from a user's spoken meal description. We first summarize the data collection and annotation of food descriptions performed via Amazon Mechanical Turk (AMT), for both a written corpus and spoken data from an in-domain speech recognizer. We show that the addition of word vector features improves conditional random field (CRF) performance for semantic tagging of food concepts, achieving an average F1 test score of 92.4 on written data; we also demonstrate that a convolutional neural network (CNN) with no hand-crafted features outperforms the best CRF on spoken data, achieving an F1 test score of 91.3. We illustrate two methods for associating foods with properties: segmenting meal descriptions with a CRF, and a complementary method that directly predicts associations with a feed-forward neural network. Finally, we conduct an end-to-end system evaluation through an AMT user study with worker ratings of 83% semantic tagging accuracy.
- 17 April 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ACM Transactions on Audio, Speech, and Language Processing
- Vol. 25 (7) , 1450-1461
Spoken Language Understanding for a Nutrition Dialogue System
Food logging is recommended by dieticians for prevention and treatment of obesity, but currently available mobile applications for diet tracking are often too difficult and time-consuming for patients to use regularly. For this reason, we propose a novel approach to food journaling that uses speech and language understanding technology in order to enable efficient self-assessment of energy and nutrient consumption. This paper presents ongoing language understanding experiments conducted as part of a larger effort to create a nutrition dialogue system that automatically extracts food concepts from a user's spoken meal description. We first summarize the data collection and annotation of food descriptions performed via Amazon Mechanical Turk (AMT), for both a written corpus and spoken data from an in-domain speech recognizer. We show that the addition of word vector features improves conditional random field (CRF) performance for semantic tagging of food concepts, achieving an average F1 test score of 92.4 on written data; we also demonstrate that a convolutional neural network (CNN) with no hand-crafted features outperforms the best CRF on spoken data, achieving an F1 test score of 91.3. We illustrate two methods for associating foods with properties: segmenting meal descriptions with a CRF, and a complementary method that directly predicts associations with a feed-forward neural network. Finally, we conduct an end-to-end system evaluation through an AMT user study with worker ratings of 83% semantic tagging accuracy.