Ease of access and user friendliness make diet-tracking apps an important ally in their users’ efforts to lose and manage weight. To foster motivation for long-term use and to achieve goals, it is necessary to better understand users’ opinions and needs for dietary self-monitoring. Motivating users to use an app over time could help them better achieve their nutrition goals.
The number of mentions of positive and negative trigrams in user reviews also showed a trend of positive evaluation dominance among users leaving reviews. The top 50 most frequent positive trigrams appeared 12,723 times, while the top 50 most frequent negative trigrams were mentioned 1270 times in our dataset of 72,084 user reviews. Although this study provides valuable insight into user opinions, it is not without limitations. Owing to feasibility constraints, we focused on available reviews and introduced a set of constraints that allowed us to structure and summarize the otherwise diverse user-generated content in the form of app reviews. Future research could apply other text-mining approaches for data collection, cleaning, and analysis. In performing similar studies, it may be beneficial to differentiate users and their motivations for using the diet-tracking app.
- It eventually led to me just not tracking even though I paid for membership.
- Topic modeling is another text-mining and NLP method that is commonly used to discover latent topics in a corpus of text.
- Technological advancements, including those related to mobile devices, are enabling developments of an increasing number of tools to help individuals take control of their health and nutrition.
- I tried it to give it a chance thinking it was what I was looking for, but it wasn’t.
- While Susie doesn’t track food for herself, she does recommend daily logging for most of her clients.
- In reviews, users are text producers for other potential consumers, businesses, and society at large.
Topic Selection Process
This library helped us to build a mathematical model that could classify each review by topic. The list of possible topics was determined during model training, and we predetermined the number of possible topics. To find the most appropriate number of topics, we used the coherence score. Topic coherence measures the degree of semantic similarity between the highly scored words in the topic, which can help to distinguish between topics that are semantically interpretable and topics that are artifacts of statistical inference [49]. This value is given after each model training process and helped us determine the performance of our trained model. After data preprocessing, a new dataset was obtained with cleaned data that could be used for both topic modeling and n-grams identification.
Apps for Cardiovascular Health: Nutrition and Weight Loss Apps
Most of the identified topics included the use of positive words when describing apps in the reviews. In their feedback, users often use words such as “love,” “nice,” “easy,” “good,” and “amaze” to describe the apps. Positively rated topics were more common than negatively rated topics. Users who leave feedback for diet-tracking apps positively rate the possibility to track their food intake; use food scanners and create/access food databases in the apps; and consider the apps to be user-friendly, convenient, and easy to use overall. Weight loss was another important topic, appearing in 10% of user reviews (Table 3).
Healthie:
A total of 72,084 user reviews in English were identified in this step using the Python library langdetect. Every mini-course will help you gain specific knowledge, tools, and skills that will help you change your habits, lose weight, and make progress far beyond the scale. You can make additional in-app purchases that range from $4.99 to $89.99.
Effortlessly Scan Food
As consumers increasingly rely on apps to support their daily activities, they also generate invaluable feedback for both developers and potential users through app reviews and ratings. These reviews typically contain information that is valuable for app evaluation, including user opinions about the app, information about their experiences with the app, and bug complaints or feature suggestions [22]. A previous study showed that almost a quarter (23.3%) of app reviews contain an app feature request or app assessment [23]. In our study, we focused on the user perspective, and aimed to evaluate the diet-tracking apps and their features that are most frequently commented on by users in app reviews.
Links to NCBI Databases
Finding the best number of topics that would give optimal results required several trials, starting with a randomly selected number of topics until we narrowed down to the model with the best score. For example, if our model found 11 topics in the dataset, for each review in our dataset, the model would provide us with the probabilities of how likely the review is to belong to each of the 11 topics. After using MFP for many years, I recently found Lose It to be much easier to use on a day to day basis as far as data entry and scanning labels.
Although users rated the apps they use very highly on average (the overall rating for all apps was 4.4 out of 5, with individual app ratings ranging from 4.1 to 4.7), some features could still be improved to enhance the user experience. Owing to their presence and relevance in the dieting field, diet-tracking apps have attracted the interest of many researchers who have used app evaluation strategies in an attempt to better understand and evaluate app features [29-32]. The influence of diet-tracking apps on users’ food choices and their opinions about these apps have been tested using experimental and survey data collection methods [10,15,18,33]. Previous studies also assessed the consistency of information provided by different apps, and recommended further collaboration and harmonization of information [34,35]. The availability and use of mobile apps in health and nutrition management are increasing.
Methods
With the increase in publicly available user-generated content due to the proliferation of internet-assisted communication, researchers have developed several automated approaches to identify, summarize, and classify the available information [26,36]. The development of new tools allows researchers to obtain unimeal scam more information about users’ opinions and sentiments in their writing. There is a trend to shift the focus of opinion mining from studying long texts to shorter user posts on various social media platforms and websites [22].
I’ve never figured out how it uses the integration with Garmin to “add calories to the day” because you did a big ride. It eventually led to me just not tracking even though I paid for membership. It has a very good database and tracking functionality for macros. It also has diet plans and recipes that I’ve actually found useful. One thing that took some getting used to is that Garmin doesn’t sync over the calories from a workout, instead it syncs Garmin Active Calories and compares it to the you calorie budget. I suppose this makes more sense as it more accurately accounts for nonexercise time as well.
Healthy eating.
The low inclusion of behavior change strategies in diet tracking apps may hinder their ability to help users achieve their long-term diet and nutrition goals [11,12]. However, diet-tracking apps that successfully employ behavior change strategies can have a positive effect on their users’ motivation, habits, and diet and nutrition outcomes [13-16]. These apps have also proven to be helpful in behavioral control and weight management [15].
What App for Tracking Nutrition?
This text can then be used to predict and understand user preferences and behaviors [26]. The aim of this study was to identify the key topics and issues that users highlight in their reviews of diet-tracking apps on Google Play Store. In addition, only apps that had the highest download numbers in the market were selected for this study.
GitHub – davidhealey/waistline: Libre calorie counter app for Android. Built…
We then converted the text to lowercase, performed an extensive spell check of every review, and made necessary corrections using the Speller Python library. Words such as “I,” “are,” “and,” and “the” were considered “stop words” and removed, as such common words tend to dominate the results. We further removed any special characters and numbers from the reviews. This Information Guide may contain information and/or instructional materials developed by Michigan Medicine for the typical patient with your condition.
