We use cookies to enhance your experience. By continuing to browse this site you agree to our use of cookies. More info.

Most people today sit for long hours with computers leading to unhealthy conditions like shoulder pain, headache, numbness, etc. The article discusses research revealing the use of an inertial sensor built into smartphones to overcome unhealthy sitting behaviors of the office worker.

Image Credit: suriya yapin/Shutterstock.com

Individuals nowadays spend most of their time sitting and sleeping without giving much care to sitting postures. Sitting is not a problem, however sitting in the wrong posture for a longer time on a daily basis can lead to major health issues like back pain, sciatica, and cervical spondylosis.

Conventional methods to analyze patient sitting postures involve letting a patient sit on the chair of the hospital in front of an observer like a therapist or doctor, completing QAs regarding sitting postures. These investigations replaced this with a smart cushion system that combines pressure sensors and IMU to monitor the sitting behavior at the workplace, in the car, and in a wheelchair.

Sitting posture monitoring systems (SPMSs) are also employed to evaluate the real-time postures of the individual and enhance the sitting behaviors. To enhance the analysis of sitting postures, it is necessary to replace the conventional techniques and think of sophisticated techniques. The research analyzed five various sitting postures, as shown in Figure 1.

Figure 1. A framework of the proposed system. Image Credit: Sinha, et al., 2021.

Smartphones feature in daily life, containing inbuilt inertial sensors. These consist of a gyroscope, accelerometer, and magnetometer. Researchers track the body's movement with an IMU sensor attached to the upper rear trunk to track the exact movement of the spine, which in turn helps identify body postures.

Table 1 lists some of the major related articles discussed based on used sensors, classifiers, and their accuracy. 

Table 1. Literature review on the basis of previous papers. Source: Sinha, et al., 2021 .

In the approach discussed, the inertial sensor of the smartphone is employed as a sitting behaviors detector. The research also prioritized data access, cost, compatibility, unobtrusive use, and system deployment.

The system gathered data for the five different static movements of the body while sitting in the chair (Figure 1). The smartphone was attached to the rear upper trunk at second thoracic vertebrae T2 to gather the measurable dataset (Figure 2). The system can identify the postures when the subject moves from the correct sitting posture to incorrect sitting postures.

Figure 2. Wearable sensor location in the human body. Image Credit: Sinha, et al., 2021.

The study created a new dataset for analyzing human sitting behavior while sitting on a chair. A total of five general movements were taken into consideration—Left movement, Right movement, Front movement, Back movement, Straight movement.

The activities were carried out by the people under the supervision of an instructor to generate the dataset effectively. Table 2 shows the total number of instances for each activity, and the same is analyzed with a pie chart in Figure 3.

Figure 3. Pie chart representation for a number of instances. Image Credit: Sinha, et al., 2021.

Table 2. Number of instances per activity. Source: Sinha, et al., 2021.

Researchers used the One Plus 6 smartphone as a hardware platform for dataset gathering. Data pre-processing is a vital step after data collection and before feature extraction. 

The feature vectors are calculated from the collected dataset. The total acceleration of the accelerometer, gyroscope, and magnetometer are also calculated.

The morphological features comprise the study of the morphological features like structure and shape from the dataset of sitting behaviors.

The article introduces two kinds of entropy-based features—wavelet entropy, the measure of relative energies in various signals, and is employed to determine the degree of the disorder and log energy entropy (LEE).

In this study, the researchers employed a filter method called correlation-based feature selection (CFS) as a feature selection algorithm. Table 3 shows the subset of features depending on their contributions.

Table 3. Selected features with contribution ratings. Source: Sinha, et al., 2021 .

The scientists combined the CFS technique and particle swarm optimization (PSO) search technique, a total of 27 features were selected out of 85 calculated features (Table 3).

This article used three classifier techniques and compared them to the most popular recognition techniques like support vector machine (SVM), K-nearest neighbor (KNN), and naive Bayes after feature selection algorithm to recognize the sitting behaviors.

MATLAB R2021a was employed to carry out all calculations and analyses of the dataset of human sitting behaviors of office workers in the office environment.

The performance evaluation of the KNN, naive Bayes, and SVM classifiers for feature selection with a gyroscope, accelerometer, and magnetometer is illustrated in Table 4. This table also depicts the comparison of the performance of each classifier for all five sitting behaviors.

Table 4. Classifier results with feature selection of accelerometer, gyroscope, and magnetometer. Source: Sinha, et al., 2021.

And Table 5 shows the confusion matrix of all five sitting behaviors by the SVM classifier.

Table 5. Confusion matrix of SVM classifier of selected features of accelerometer, gyroscope, and magnetometer. Source: Sinha, et al., 2021

Table 6 shows the results of classifiers for feature selection employing accelerometer and gyroscope sensors and Figure 4.

Depicts the overall comparison of classifier results with all input postures.

Figure 4. Different classifier results with feature selection of accelerometer and gyroscope. Image Credit: Sinha, et al., 2021.

Table 6. Classifiers result with feature selection of accelerometer and gyroscope. Source: Sinha, et al., 2021 .

Table 7 illustrates the confusion matrix for all considered sitting behaviors of office workers by employing an accelerometer and gyroscope with the SVM classifier.

Table 7. Confusion matrix of SVM classifier of selected features of accelerometer and gyroscope. Source: Sinha, et al., 2021.

Table 8 shows the result of all the applied classifiers by employing the accelerometer sensor and the confusion matrix in Table 9 is generated for five various sitting behaviors confused with each other. 

Table 8. Classifier results with feature selection of accelerometer. Source: Sinha, et al., 2021

Table 9. Confusion matrix of KNN (K = 3) classifier of selected features of accelerometer. Source: Sinha, et al., 2021

The current study used smartphone technology as the sensor for analysis of the sitting behavior of test subjects. Based on the results, it was noted that the SVM classifier outdid the other classifiers. However, when considering the other classifiers, the performance is not that precise while considering all the postures.

The study analyzed five general sitting behaviors of office workers by employing the inertial sensor inbuilt in a smartphone with the help of machine learning classification techniques. Researchers have also performed a comparative analysis of different activity recognition techniques and found that 99.90% accuracy was achieved for all sitting behaviors on the office chair by the SVM classifier.

Continue reading: Wireless 3D Printed Wearable Sensor to Track Health and Body Function.

Sinha, V. K., Patro, K. K., Pławiak, P., Prakash, A. J. (2021) Smartphone-Based Human Sitting Behaviors Recognition Using Inertial Sensor. Sensors, 21(19), p. 6652. Available at: https://doi.org/10.3390/s21196652.

Megan graduated from The University of Manchester with a B.Sc. in Genetics, and decided to pursue an M.Sc. in Science and Health Communication due to her passion for combining science with content creation. As part of her studies, Megan partnered with Jodrell Bank Discovery Centre as a Digital Marketing Assistant, producing content and updating sections of their website. In her spare time, she loves to travel, exploring each location's culture and history - including the local cuisine. Her other interests include embroidery, reading fiction, and practicing her Japanese language skills.

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

Craig, Megan. (2021, November 05). Analysis of Different Human Sitting Behaviors Using Inertial Sensor. AZoSensors. Retrieved on August 08, 2022 from https://www.azosensors.com/article.aspx?ArticleID=2352.

Craig, Megan. "Analysis of Different Human Sitting Behaviors Using Inertial Sensor". AZoSensors. 08 August 2022. <https://www.azosensors.com/article.aspx?ArticleID=2352>.

Craig, Megan. "Analysis of Different Human Sitting Behaviors Using Inertial Sensor". AZoSensors. https://www.azosensors.com/article.aspx?ArticleID=2352. (accessed August 08, 2022).

Craig, Megan. 2021. Analysis of Different Human Sitting Behaviors Using Inertial Sensor. AZoSensors, viewed 08 August 2022, https://www.azosensors.com/article.aspx?ArticleID=2352.

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

AZoSensors speaks with Mr. Takeshi Konno from Fujitsu about the release of their new millimeter-wave sensor system that aims to improve patient privacy and safety in healthcare settings.

NASA's newly launched satellite, CAPSTONE, hopes to verify an optimal lunar orbit and is a significant step towards establishing the Lunar Gateway outpost for NASA's future Artemis missions.

AZoSensors speaks with the Institute of Railway Research about the integration of an ultrasonic sensor into railway track lubrication monitoring and maintenance.

The BM-25 diaphragm switch is a point level detector for bin and chutes.

OTT HydroMet’s CHM8k is a ceilometer that can measure aerosol backscatter profiles up to 10 km high. In the daylight, it can also collect data on cloud bases up to 8 km high.

The MFE600E Intelligent Electromagnetic Flowmeter from Micro Sensor is best suited for use in industrial fields.

AZoSensors.com - An AZoNetwork Site

Owned and operated by AZoNetwork, © 2000-2022