Smartphones and Deep Learning Can Be Used to Detect Obesity

In an article recently published in the International Journal of Environmental Research and Public Health, researchers introduced a novel deep learning framework designed to identify obesity in adolescents through smartphone inertial measurements. This approach leverages gait pattern analysis using deep learning models to detect obesity indicators. The proposed method offers a scalable, accessible solution for continuous, non-intrusive obesity risk monitoring.

Smartphones and Deep Learning Can Be Used to Detect Obesity

Image Credit: Creativa Images/Shutterstock.com

Background

Obesity in adolescents is a growing concern due to its association with various health issues, including osteoarthritis, sleep apnea, cancer, and mental illnesses. Traditional methods, such as body mass index (BMI), offer a basic classification of overweight and obese individuals but fail to capture detailed biomechanical changes, such as those seen in gait patterns.

Gait analysis, the study of walking patterns, is widely used in clinical diagnoses, robotics, sports, and biomechanics. However, traditional methods like motion capture systems, which require markers, are costly and confined to lab-based settings, limiting their practicality for large-scale or real-time monitoring. Portable and unobtrusive alternatives, such as wearable sensors—including accelerometers, gyroscopes, and inertial measurement units—allow for gait analysis in real-world environments.

About the Research

In this paper, the authors propose a deep learning-based novel framework to identify obesity in adolescents using smartphone inertial sensors for gait data collection. Their main goal is to determine whether gait patterns captured by smartphone sensors can accurately differentiate between normal and obese adolescents.

The study used three deep learning models: convolutional neural networks (CNNs), long short-term memory networks (LSTMs), and a combination of CNN and LSTM models. These models were selected for their ability to capture both spatial and temporal patterns in sequential data, making them suitable for analyzing complex gait patterns related to obesity.

Data was then collected from 244 first-year high school students using a mobile health app installed on smartphones. Each participant placed the smartphone in a pocket at the center of their body and walked a 78-foot course back and forth. Additionally, nine-channel rotation matrices from the smartphone’s rotation vector were used to analyze the gait data.

The researchers analyzed data from 173 participants, comprising 92 from the normal weight group and 46 from the overweight or obese group. To prepare the data for model training, they first reduced noise using a moving average filter, followed by time series segmentation through a sliding window method. For model evaluation, they allocated 80 % of the data for training and 20 % for testing.

The models were trained using the PyTorch framework with the Adam optimizer to minimize the cost function. Training was conducted over 50 epochs with a learning rate of 0.003. Model performance was assessed using several metrics, including accuracy, precision, recall, and the F1-score, which represents the harmonic mean of precision and recall.

Research Findings

The outcomes showed that the hybrid CNN-LSTM model outperformed both the CNN and LSTM models in classifying obesity based on gait patterns. It achieved an accuracy of 97 %, compared to 96.31 % for the LSTM model and 95.81 % for the CNN model. These results indicate that the hybrid model consistently performed better because it could handle sequential and spatial features.

While the CNN model performed well during training, it struggled to generalize to new data, likely due to its limitations in addressing the sequential nature of gait patterns. In contrast, the LSTM model effectively captured temporal dependencies but lacked the CNN’s ability to extract spatial features, resulting in slightly lower performance compared to the hybrid model. This study underscored the advantages of integrating deep learning models with smartphone technology for continuous, non-intrusive obesity risk monitoring in adolescents.

Additionally, the researchers noted significant differences in gait patterns between the obese and normal groups. The obese group exhibited shorter step lengths, slower walking speeds, and greater variability, consistent with previous research linking obesity to alterations in gait patterns.

Applications

This research has significant implications for developing new methods to detect and prevent obesity. Using smartphone inertial sensors and deep learning models, the study presents a noninvasive and cost-effective approach to monitoring gait patterns and identifying individuals at risk of obesity. This technology can be applied in clinical, community, and public health settings. It could offer personalized feedback on gait patterns and track changes over time, helping raise awareness of weight changes and supporting obesity prevention by providing quick alerts for better health monitoring.

Conclusion

In summary, using deep learning models combined with smartphone inertial sensors proved effective for identifying obesity through gait analysis. The hybrid CNN-LSTM model showed superior accuracy in classifying the obese based on gait patterns, highlighting the importance of integrating temporal and spatial feature extraction.

Future work should include more diverse samples to check the results and explore various smartphone models to ensure robustness. Overall, this research could significantly contribute to public health efforts to combat obesity.

Journal Reference

Degbey, G.-S.; Hwang, E.; Park, J.; Lee, S. Deep Learning-Based Obesity Identification System for Young Adults Using Smartphone Inertial Measurements. Int. J. Environ. Res. Public Health 2024, 21, 1178. DOI: 10.3390/ijerph21091178, https://www.mdpi.com/1660-4601/21/9/1178

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Muhammad Osama

Written by

Muhammad Osama

Muhammad Osama is a full-time data analytics consultant and freelance technical writer based in Delhi, India. He specializes in transforming complex technical concepts into accessible content. He has a Bachelor of Technology in Mechanical Engineering with specialization in AI & Robotics from Galgotias University, India, and he has extensive experience in technical content writing, data science and analytics, and artificial intelligence.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Osama, Muhammad. (2024, September 12). Smartphones and Deep Learning Can Be Used to Detect Obesity. AZoRobotics. Retrieved on October 09, 2024 from https://www.azorobotics.com/News.aspx?newsID=15251.

  • MLA

    Osama, Muhammad. "Smartphones and Deep Learning Can Be Used to Detect Obesity". AZoRobotics. 09 October 2024. <https://www.azorobotics.com/News.aspx?newsID=15251>.

  • Chicago

    Osama, Muhammad. "Smartphones and Deep Learning Can Be Used to Detect Obesity". AZoRobotics. https://www.azorobotics.com/News.aspx?newsID=15251. (accessed October 09, 2024).

  • Harvard

    Osama, Muhammad. 2024. Smartphones and Deep Learning Can Be Used to Detect Obesity. AZoRobotics, viewed 09 October 2024, https://www.azorobotics.com/News.aspx?newsID=15251.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.