Nutrient and Food Group Prediction as Orchestrated by an Automated Image Recognition System in a Smartphone App (CALO mama): Validation Study

Validation of artificial intelligence-based application to estimate nutrients in daily meals

BACKGROUND

Diet modification is a mainstay for the successful management of metabolic syndrome and potentially may reduce the risk of cardiovascular disease. Accurate estimation of essential nutrients in daily meals is currently challenging to quantify. HAKARIUM (AstraZeneca Co., Ltd., Osaka, Japan) is a recently introduced artificial intelligence (AI)-based application that can estimate each nutrient component through photographs, although its applicability to real-world practice remains unknown.

METHODS

Lunchtime meals served for healthy individuals at a single university cooperative society between September 2023 and February 2024 were analyzed. Nutrient components, including energy in the form of calories, protein, and salts, were estimated by the HAKARIUM application and compared with the actual nutrient values that were officially calculated and presented by the university cooperative society.

RESULTS

A total of 62 meals were included. Actual values of energy, protein, and salt content per meal were 382 (358, 431) kcal, 17.1 (13.9, 18.9) g, and 2.9 (2.6, 3.1) g, respectively. AI-estimated values of energy, protein, and salt content per meal were 636 (493, 835) kcal, 25.7 (19.7, 36.3) g, and 4.2 (3.5, 4.6) g, respectively. Most of the values were within the limits of agreement with significant correlations between the two variables, respectively (r > 0.80, p < 0.05 for all).

CONCLUSION

AI-based estimation of nutrient components had relatively good agreement with actually calculated values.

Increased vegetable consumption in Japan using an incentivized health communication campaign with a quiz

Dietary habits, particularly vegetable consumption, play a crucial role in preventing noncommunicable diseases. However, despite international guidelines advocating daily vegetable intake, adherence remains low across many populations. As a result, more focused efforts to boost vegetable consumption at the population level are essential. This study aimed to assess the impact of a health communication campaign (HCC) in City A, which combined information dissemination and incentives to promote vegetable consumption. In 2021, a new app-based vegetable quiz was introduced as part of the ongoing campaign, which had been implemented since 2017. Participants earned 10 points per correct quiz answer, which could be redeemed for product certificates, with a maximum of 30 points. To evaluate the effectiveness of the quiz, we analysed vegetable intake data from 786 quiz users. A multiple regression analysis was conducted to consider factors such as sex, age, body mass index, pre-campaign points, prior vegetable intake, and frequency of food recording during the campaign. We ensured robustness of the results by analysing data from 605 individuals whose vegetable intake had been tracked one year earlier, during a non-incentivized version of the campaign. The results demonstrated that participants who completed all three quizzes consumed 10.7% more vegetables than non-participants. Year-over-year comparisons further showed a significant increase in vegetable intake among frequent quiz participants compared to the previous year, highlighting the positive impact of gamified quizzes on vegetable consumption. These findings suggest that incentivized HCC, especially those incorporating gamification elements, can be highly effective in encouraging healthier eating habits.

Real-World Data on Alcohol Consumption Behavior Among Smartphone Health Care App Users in Japan: Retrospective Study

BACKGROUND

Although many studies have used smartphone apps to examine alcohol consumption, none have clearly delineated long-term (>1 year) consumption among the general population.

OBJECTIVE

The objective of our study is to elucidate in detail the alcohol consumption behavior of alcohol drinkers in Japan using individual real-world data. During the state of emergency associated with the COVID-19 outbreak, the government requested that people restrict social gatherings and stay at home, so we hypothesize that alcohol consumption among Japanese working people decreased during this period due to the decrease in occasions for alcohol consumption. This analysis was only possible with individual real-world data. We also aimed to clarify the effects of digital interventions based on notifications about daily alcohol consumption.

METHODS

We conducted a retrospective study targeting 5-year log data from January 1, 2018, to December 31, 2022, obtained from a commercial smartphone health care app (CALO mama Plus). First, to investigate the possible size of the real-world data, we investigated the rate of active users of this commercial smartphone app. Second, to validate the individual real-world data recorded in the app, we compared individual real-world data from 9991 randomly selected users with government-provided open data on the number of daily confirmed COVID-19 cases in Japan and with nationwide alcohol consumption data. To clarify the effects of digital interventions, we investigated the relationship between 2 types of notification records (ie, "good" and "bad") and a 3-day daily alcohol consumption log following the notification. The protocol of this retrospective study was approved by the Ethics Committee of the Kyoto University Graduate School and Faculty of Medicine (R4699).

Evaluating the Quality and Comparative Validity of Manual Food Logging and Artificial Intelligence-Enabled Food Image Recognition in Apps for Nutrition Care

For artificial intelligence (AI) to support nutrition care, high quality and accuracy of its features within smartphone applications (apps) are essential. This study evaluated popular apps' features, quality, behaviour change potential, and comparative validity of dietary assessment via manual logging and AI. The top 200 free and paid nutrition-related apps from Australia's Apple App and Google Play stores were screened (n = 800). Apps were assessed using MARS (quality) and ABACUS (behaviour change potential). Nutritional outputs from manual food logging and AI-enabled food-image recognition apps were compared with food records for Western, Asian, and Recommended diets. Among 18 apps, Noom scored highest on MARS (mean = 4.44) and ABACUS (21/21). From 16 manual food-logging apps, energy was overestimated for Western (mean: 1040 kJ) but underestimated for Asian (mean: -1520 kJ) diets. MyFitnessPal and Fastic had the highest accuracy (97% and 92%, respectively) out of seven AI-enabled food image recognition apps. Apps with more AI integration demonstrated better functionality, but automatic energy estimations from AI-enabled food image recognition were inaccurate. To enhance the integration of apps into nutrition care, collaborating with dietitians is essential for improving their credibility and comparative validity by expanding food databases. Moreover, training AI models are needed to improve AI-enabled food recognition, especially for mixed dishes and culturally diverse foods.

Carbohydrate Estimation Accuracy of Two Commercially Available Smartphone Applications vs Estimation by Individuals With Type 1 Diabetes: A Comparative Study

BACKGROUND

Despite remarkable progress in diabetes technology, most systems still require estimating meal carbohydrate (CHO) content for meal-time insulin delivery. Emerging smartphone applications may obviate this need, but performance data in relation to patient estimates remain scarce.

OBJECTIVE

The objective is to assess the accuracy of two commercial CHO estimation applications, SNAQ and Calorie Mama, and compare their performance with the estimation accuracy of people with type 1 diabetes (T1D).

METHODS

Carbohydrate estimates of 53 individuals with T1D (aged ≥16 years) were compared with those of SNAQ (food recognition + quantification) and Calorie Mama (food recognition + adjustable standard portion size). Twenty-six cooked meals were prepared at the hospital kitchen. Each participant estimated the CHO content of two meals in three different sizes without assistance. Participants then used SNAQ for CHO quantification in one meal and Calorie Mama for the other (all three sizes). Accuracy was the estimate's deviation from ground-truth CHO content (weight multiplied by nutritional facts from recipe database). Furthermore, the applications were rated using the Mars-G questionnaire.

RESULTS

Participants' mean ± standard deviation (SD) absolute error was 21 ± 21.5 g (71 ± 72.7%). Calorie Mama had a mean absolute error of 24 ± 36.5 g (81.2 ± 123.4%). With a mean absolute error of 13.1 ± 11.3 g (44.3 ± 38.2%), SNAQ outperformed the estimation accuracy of patients and Calorie Mama (both P > .05). Error consistency (quantified by the within-participant SD) did not significantly differ between the methods.

CONCLUSIONS

SNAQ may provide effective CHO estimation support for people with T1D, particularly those with large or inconsistent CHO estimation errors. Its impact on glucose control remains to be evaluated.

Smartphone applications for nutrition Support: A systematic review of the target outcomes and main functionalities

INTRODUCTION

A proper nutrition is essential for human life. Recently, special attention on this topic has been given in relation to three health statuses: obesity, malnutrition and specific diseases that can be related to food or treated with specific diets. Mobile technology is often used to assist users that wish to regulate their eating habits, and identifying which fields of application have been explored the most by the app developers and which main functionalities have been adopted can be useful in view of future app developments.

METHODS

We selected 322 articles mentioning nutrition support apps through a literature database search, all of which have undergone an initial screening. After the exclusion of papers that were already reviews, not presenting apps or not focused on nutrition, not relevant or not developed for human subjects, 100 papers were selected for subsequent analyses that aimed at identifying the main treated conditions, outcome measures and functionalities implemented in the Apps.

RESULTS

Of the selected studies, 33 focus on specific diseases, 24 on obesity, 2 on malnutrition and 41 on other targets (e.g., weight/diet control). Type 2 diabetes is the most targeted disease, followed by gestational diabetes, hypertension, colorectal cancer and CVDs which all were targeted by more than one app. Most Apps include self-monitoring and coaching functionalities, educational content and artificial intelligence (AI) tools are slightly less common, whereas counseling, gamification and questionnaires are the least implemented. Body weight and calories/nutrients were the most common general outcome measures, while glycated hemoglobin (HbA1c) was the most common clinical outcome. No statistically significant differences in the effectiveness of the different functionalities were found.

CONCLUSION

The use of mobile technology to improve nutrition has been widely explored in the last years, especially for weight control and specific diseases like diabetes; however, other food-related conditions such as Irritable Bowel Diseases appear to be less targeted by newly developed smartphone apps and their related studies. All different kinds of functionalities appear to be equally effective, but further specific studies are needed to confirm the results.

AI-based digital image dietary assessment methods compared to humans and ground truth: a systematic review

OBJECTIVE

Human error estimating food intake is a major source of bias in nutrition research. Artificial intelligence (AI) methods may reduce bias, but the overall accuracy of AI estimates is unknown. This study was a systematic review of peer-reviewed journal articles comparing fully automated AI-based (e.g. deep learning) methods of dietary assessment from digital images to human assessors and ground truth (e.g. doubly labelled water).

MATERIALS AND METHODS

Literature was searched through May 2023 in four electronic databases plus reference mining. Eligible articles reported AI estimated volume, energy, or nutrients. Independent investigators screened articles and extracted data. Potential sources of bias were documented in absence of an applicable risk of bias assessment tool.

RESULTS

Database and hand searches identified 14,059 unique publications; fifty-two papers (studies) published from 2010 to 2023 were retained. For food detection and classification, 79% of papers used a convolutional neural network. Common ground truth sources were calculation using nutrient tables (51%) and weighed food (27%). Included papers varied widely in food image databases and results reported, so meta-analytic synthesis could not be conducted. Relative errors were extracted or calculated from 69% of papers. Average overall relative errors (AI vs. ground truth) ranged from 0.10% to 38.3% for calories and 0.09% to 33% for volume, suggesting similar performance. Ranges of relative error were lower when images had single/simple foods.

CONCLUSIONS

Relative errors for volume and calorie estimations suggest that AI methods align with - and have the potential to exceed - accuracy of human estimations. However, variability in food image databases and results reported prevented meta-analytic synthesis. The field can advance by testing AI architectures on a limited number of large-scale food image and nutrition databases that the field determines to be adequate for training and testing and by reporting accuracy of at least absolute and relative error for volume or calorie estimations.

Exploring Changes in Attitudes, Behaviors, and Self-Measured Health Data Through Lifestyle Modification Support by Community Pharmacists: Suito-Ogaki Selfcare (SOS) Trial

Purpose

Contributing to public health by supporting people's health is the social mission of community pharmacists. This multicenter, prospective case series study aimed to evaluate changes in people's behavior and health states through community pharmacists' self-care support for healthy lifestyles.

Methods

The participants were recruited from voluntary adults aged ≥20 years who agreed to participate in the study, at community pharmacies in Gifu, Japan, between June and September 2021. Participants self-managed their lifestyles for six months while recording their health data, including blood pressure (BP), daily using devices (home BP monitor, body composition monitor, and activity meter) and a diet-recording app. They received lifestyle modification support at pharmacies at least once per month. Participants' subjective health status, attitudes, and behavioral changes were evaluated using self-report questionnaires. Due to the exploratory nature of this study, data were primarily analyzed descriptively.

Results

Fifty-four participants aged 20 to 77 (mean age: 49.6 years; female participant proportion: 55.6%) participated in this study. Their mean weekly BP shifted almost horizontally from baseline to week 24 (systolic BP: 118.8 to 121.5 mmHg; diastolic BP: 76.1 to 77.5 mmHg). At six months, 38.9% and 35.2% of the participants reported better overall health and mental health, respectively, than at baseline. Over 85% of the participants became more proactive in improving their lifestyles regarding salt intake, diet, weight loss, and exercise, although drinking and smoking habits were more challenging to change. All the participants reported that they intended to continue to improve their lifestyle.

Conclusion

The participants' responses suggested that community pharmacists' support helped increase participants' health awareness and promote their health-enhancing behaviors. However, its impact on health parameters should be further examined in future studies. More vigorous, tailored self-care support may be worth considering in developing a more effective, community-fitted health/well-being support system in Japan.

A Smartphone Healthcare Application, CALO mama Plus, to Promote Weight Loss: A Randomized Controlled Trial

Mobile applications are increasingly used in healthcare. We have developed a smartphone healthcare application, CALO mama Plus, that can register daily diet, exercise, mood, and sleep quality, calculate dietary intake, and provide advice using artificial intelligence technology. This 3-month randomized controlled trial tested the hypothesis that CALO mama Plus could promote body weight reduction in Japanese adults with overweight or obesity. We recruited office workers as participants. The key eligibility criteria were an age of 20–65 years and a body mass index of 23–40 kg/m². The primary outcome was body weight change over 3 months. We enrolled 141 participants and randomly assigned them to the intervention (n = 72) and control (n = 69) groups. The intervention group used CALO mama Plus, and the control group did not receive any intervention. The change in body weight was −2.4 ± 4.0 kg and −0.7 ± 3.3 kg in the intervention and control groups, respectively. An analysis of covariance adjusted for related variables showed a significant between-group difference in body weight change (−1.60 kg; 95% confidence interval −2.83 to −0.38; p = 0.011). The present study suggests that CALO mama Plus effectively promotes weight loss.

Can Digital Health Solutions Fill in the Gap for Effective Guideline Implementation in Cardiovascular Disease Prevention: Hope or Hype?

PURPOSE OF REVIEW

To evaluate the use of digital health solutions in improving the implementation of cardiovascular disease (CVD) prevention guidelines and review current evidence supporting it.

RECENT FINDINGS

Healthy diet guideline recommendations can be improved by text-messaging programs and apps for reinforcing healthy food intake and reducing unhealthy nutrients purchase and consumption. Wearable activity trackers are also effective in increasing physical activity levels. Text-messaging programs for smoking cessation have demonstrated benefits in increasing quitting rates; however, evidence on smartphone apps for smoking cessation is still lacking. Smartphone apps have the potential to improve medication adherence; however, better quality evidence is needed. Digital pills are another promising digital solution to improve medication adherence. The use of digital health solutions in CVD prevention is an evolving field and, to date, there is an increasing body of evidence that supports that such technologies are effective in improving CVD prevention guideline implementation.

Multimedia Data-Based Mobile Applications for Dietary Assessment

Diabetes mellitus (DM) and obesity are chronic medical conditions associated with significant morbidity and mortality. Accurate macronutrient and energy estimation could be beneficial in attempts to manage DM and obesity, leading to improved glycemic control and weight reduction, respectively. Existing dietary assessment methods are subject to major errors in measurement, are time consuming, are costly, and do not provide real-time feedback. The increasing adoption of smartphones and artificial intelligence, along with the advances in algorithms and hardware, allowed the development of technologies executed in smartphones that use food/beverage multimedia data as an input, and output information about the nutrient content in almost real time. Scope of this review was to explore the various image-based and video-based systems designed for dietary assessment. We identified 22 different systems and divided these into three categories on the basis of their setting for evaluation: laboratory (12), preclinical (7), and clinical (3). The major findings of the review are that there is still a number of open research questions and technical challenges to be addressed and end users-including health care professionals and patients-need to be involved in the design and development of such innovative solutions. Last, there is a clear need that these systems should be validated under unconstrained real-life conditions and that they should be compared with conventional methods for dietary assessment.