Next Article in Journal
Predicting Employee Turnover: A Systematic Machine Learning Approach for Resource Conservation and Workforce Stability
Previous Article in Journal
Numerical Analysis of Lid-Driven Cavity Flow Induced by Triangular Obstacles
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Proceeding Paper

Design and Implementation of an IoT Based Smart Digestive Health Monitoring Device for Identification of Digestive Conditions †

by
Rajesh Kumar Dhanaraj
1,
Alagumariappan Paramasivam
2,*,
Sankaran Vijayalakshmi
3,
Cyril Emmanuel
4,
Pittu Pallavi
2,
Pravin Satyanarayan Metkewar
1 and
Manoj Ashwin
2
1
Symbiosis Institute of Computer Studies and Research (SICSR), Symbiosis International (Deemed University), Pune 411016, India
2
Department of Biomedical Engineering, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology, Chennai 600062, India
3
Department of Electronics and Communication Engineering, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology, Chennai 600062, India
4
Gleneagles Global Health City, Chennai 600100, India
*
Author to whom correspondence should be addressed.
Presented at the 10th International Electronic Conference on Sensors and Applications (ECSA-10), 15–30 November 2023; Available online: https://ecsa-10.sciforum.net/.
Eng. Proc. 2023, 58(1), 33; https://doi.org/10.3390/ecsa-10-16253
Published: 15 November 2023

Abstract

:
Over the past few decades, there has been a significant rise in wearable healthcare technologies that have been playing a major role all over the world in monitoring health, alerting individuals during deviations from their normal health conditions and assisting them to stay fit and healthy. Due to the modern lifestyle and consumption of unhealthy food products, there has been an adverse effect on digestive health standards. In this work, a wearable device with textile electrodes is designed and developed to analyze the digestive conditions, namely, pre-prandial and post-prandial, using Electrogastrogram (EGG) signals. Further, the proposed device is comprised of textile electrodes as a sensor, an Analog-to-Digital Converter (ADC) with a Programmable Gain Amplifier (PGA), a Microcontroller with an inbuilt WirelessFidelity (WiFi) module, and an Internet of Things (IoT) cloud platform. Also, the EGG signals are acquired under two different conditions, namely, pre-prandial and post-prandial conditions, and then the Long Short Term Memory (LSTM) deep learning model is utilized to classify pre-prandial and post-prandial EGG signals to identify the eating habits of normal individuals. Results demonstrate that the proposed approach is capable of classifying the pre-prandial and post-prandial EGG signals, which, in turn, identify the fasting or ingestion state of normal individuals.

1. Introduction

In general, indulging in food on a regular basis is essential for a healthy digestive system. However, people in different age groups are more subjected to a sedentary lifestyle. Further, they prefer to consume highly processed foods, spicy foods, or excessive meals without hunger, skipping meals, eating at unusual times, and eating before sleep [1]. Also, problems with sleep are associated with digestive disorders, either as a form of symptom or a cause [2]. Digestive disorders, otherwise known as gastrointestinal (GI) tract disorders, are impairments and diseases linked to the Gastrointestinal (GI) tract of the human body. These GI tract disorders are linked to abnormal eating habits and unhealthy lifestyles. Overall, this leads to most of the digestive disorders such as Gastroesophageal Reflux Disease (GERD) [1], chronic diarrhea, and major gastrointestinal cancers, namely Colorectal Cancer (CRC), Esophageal Cancer (EC), Pancreatic Cancer (PC) [3], functional dyspepsia, etc. These are some of the serious digestive disorders observed mostly in people worldwide and, thus, can lead to complications as time progresses.
According to a survey report in 2019 [4], of the 6174 participants, 8.2% had GERD. The disease is more prevalent in urban areas (11.1%) than in rural areas (5.1%). Furthermore, GERD is a very common chronic disease in India and can lead to serious chronic diseases like adenocarcinoma if left unnoticed. The common methods of diagnosing digestive disorders include lab tests, imaging tests, endoscopic methods, etc. Also, the imaging tests include a colorectal transit study, a Computerized Tomography (CT) scan, defecography, a Magnetic Resonance Imaging (MRI) scan, ultrasound scanning, etc. The common imaging method used in digestive diagnostics is computerized tomography. In CT, cross-sectional imagery using the X-ray method is used to detect an accurate reading of the organ’s histology. According to the study conducted by Herbert L. Fred [5], the method is very expensive, and the amount of radiation dose during a test is extremely high. The prolonged use of the device is extremely dangerous for the patient. Endoscopic procedures involve colonoscopy, endoscopic retrograde cholangiopancreatography, and sigmoidoscopy. Endoscopy is the visual examination of the digestive tract using a camera placed at the tip of a flexible tube, and the tube is inserted through the mouth or the anal orifice. Therefore, it can cause extreme discomfort for the patient during the procedure. Blockages due to the flexibility and maneuverability limitations of the tube can also lead to complications and extreme discomfort for the patient. Moreover, it can be used for histological tests of the gut lining, as it only gives imagery as results [6]. These limitations can affect the accuracy of the diagnostics. To prevent such problems, the non-invasive method of electrogastrogram (EGG) is used.
An Electrogastrogram (EGG) is an electrical signal that gives electrical activity to the stomach and can be recorded with the help of a non-invasive technique called Electrogastrography by placing electrodes over the surface of the stomach. With the use of EGG signals, digestive disorders can be detected in an effective manner [7]. In this era of smart technology, health parameters related to the Electrocardiogram (ECG), blood pressure, heart rate, and oxygen saturation are measured [8]. While surfing various literature, it is evident that the EGG shall be used for the purpose of scientific research, and only a few studies are carried out from a practical point of view. So, the need for EGG tracking as a health parameter with the use of real-time Internet of Things technology is highly beneficial and helps individuals to be alert and able to take care of people with digestive disorders.
EGG involves the measurement of the electrical rhythm of the stomach. In a healthy person, the cycle would be 3 cycles per minute. This frequency can either increase or decrease based on the disorders associated with the system. Further, the increase in frequency is known as Tachygastria, and the decrease in frequency is known as Bradygastria. Also, the increase and decrease in frequency can be measured and detected by using Artificial Intelligence systems, especially learning methods. In a study conducted by Raihan et al. [4], the use of various AI algorithms, including support vector machines (SVM), K-Nearest Neighbor (KNN), and Logistic Regression (LR), was utilized. In machine learning, the computer possesses the ability to learn without any background programming, and the ML algorithm is set to find patterns in the data. Furthermore, Electrogastrogram (EGG) signals can be abnormal by being slower or faster than the normal rate. Therefore, the machine learning algorithm can be used to detect changes in the pattern and provide an accurate result that can differentiate between normal and abnormal EGG signals.

2. Literature Survey

Due to poor electrical activity in the stomach, the majority of people suffer from digestive disorders. Gopu et al. (2008) have proposed a method of recording electrogastrogram signals to avoid the use of endoscopy by using cutaneous Ag-AgCl electrodes and a Signal Conditioning Unit to improve signal quality. The signals are filtered to remove noise, and the converted digital output is sent to a microcontroller [9]. A method presented by Haddab et al. (2009) represents the EGG signal acquisition, using neural networks for noise filtering of motion artifacts with the actual signal and then transmits to the medical care unit through GSM communication [10]. In contrast to passive electrode systems for the acquisition of electrogastrogram signals, an active electrode setup was proposed by Gopu et al. (2010), which showed a higher sensitivity and reliability that helped in the diagnosis of gastric disorders such as ulcers and dyspepsia through preprocessing using Principal Component Analysis (PCA) with the support of a wavelet transform for analysis [11].
Gharibans et al. (2018) discussed that EGG signals can be recorded using multi-channel systems of wearable type and also used some signal processing methods for the removal of artifacts, overtaking the limitations of single-channel measurement and the presence of signal artifacts that resulted in inconsistency in data reliability. This proposed approach shows an increased scope for diagnosing and treating GI disorders in an effective manner [12]. The development of both two- and three-electrode systems and a comparison made by Alagumariappan et al. (2018) for recording the electrical activity of the stomach showed that the three-electrode system showed a higher information content, ensuring progression in the accurate diagnosis of any abnormalities related to electrogastrogram signals [13]. Gharibans et al. (2019) revealed that abnormality in gastrointestinal function is a multifactorial and potential cause of gastroparesis and functional dyspepsia, for which non-invasive cutaneous high-resolution recording of EGG helps in identification of those above-mentioned symptoms [14].
Several researchers have extracted features from acquired EGG signals and classified various digestive diseases [15,16,17,18,19,20]. Alagumariappan et al. (2020) have discussed the role of electrogastrograms in prior detection of digestion abnormalities in diagnosing Type 2 Diabetes with the help of extracting features by pre-processing the recorded EGG signals using Empirical Mode Decomposition and generic algorithms in picking up good features and relating all these features with the digestive system’s mobility [15]. A user-friendly, non-invasive, and wearable approach for monitoring gastrointestinal problems is proposed by Kumar et al. (2020). They used LabVIEW software for analyzing signals and the moving average algorithm in MATLAB for accurate electro-gastrographic extremities [16]. A study conducted by Paramasivam et al. (2021) explores that there is a positive impact of yoga asanas on the digestion process. This is identified through Fast Fourier Transform (FFT) by recording EEG signals before and after yoga, for which the normal frequency range of EGG signals is aligned with post-yoga recorded signals [17].
The objective of this work is to design and develop an Internet of Things-based smart wearable device to alert/notify people once they have skipped their food habits on time. In this paper, the proposed work is organized into four different sections. The first section deals with a brief introduction to the digestion process, its associated electrical signals, and the techniques used to assess the progress of digestion. Further, Section 2 deals with literature relevant to EGG techniques, objectives, and the organization of the research paper. Section 3 explains the proposed methodology, and Section 4 focuses on the results and their analyses. The conclusions reached through analysis are presented in Section 5.

3. Methodology

In this work, a wearable device fabricated with three textile electrodes is designed and developed. Further, the digestive conditions, namely pre-prandial and post-prandial, using EGG signals are analyzed. Figure 1 shows an overall block diagram of the proposed approach.
In the proposed approach, participants without any previous history of digestive health complications are selected. Furthermore, two different things, namely consent forms and questionnaires, are obtained from the participants. The experimental procedures are explained clearly, and informed consent is obtained, whereas the questionnaires are obtained to ensure that the participants are not having any medical complications. After obtaining informed consent and questionnaires, the participants are selected for EGG signal acquisition. Also, the EGG signals are acquired from selected participants for two different conditions, namely pre-prandial and post-prandial conditions. These acquired EGG signals are preprocessed, and the unwanted frequency components are removed. Further, the preprocessed pre-prandial and post-prandial EGG signals are given to a deep learning model for the learning process. Once the deep learning model is trained, it provides decision support about digestive habits, namely pre-prandial and post-prandial conditions.

3.1. Proposed EGG Device

Figure 2 shows the block diagram for the proposed device. The proposed device is a wearable device that is used to monitor digestive habits, namely pre-prandial and post-prandial conditions, effectively, which leads to a healthy life.
The proposed device consists of components such as Textile Electrodes, an ADC with programmable gain amplifier, a Microcontroller unit, a battery unit, and an Internet of Things cloud platform.

3.1.1. Textile Electrodes

A conductive thread made up of stainless-steel material is utilized in this work to fabricate electrodes for EGG signal acquisition. Also, the thread is stitched in three distinct places of the innerwear, which forms three fabric electrodes, and its position is determined according to the three-electrode placement protocol suggested by [21]. Furthermore, the wire is tapped from three fabric electrodes, and once the innerwear is worn by the individuals, the fabric electrodes pick up the EGG signals. These acquired EGG signals are given to the Analog to Digital Converter (ADC) for further processing.

3.1.2. ADC with PGA

In this work, an ADS1115-based ADC with a PGA module is used to convert and amplify the EGG signals with less power consumption. The ADS1115 consumes 150 micro-amps of current. Also, the utilized ADS1115 operates from 2 volts to 5 volts and has 4 ADC channels, which can perform two differential input operations. In general, the range of the EGG signals is in micro-volts, and due to this, it is important to amplify the acquired EGG signals to perform further processing. The adopted ADC converter with PGA module performs two different operations, namely amplification and AD conversion. Firstly, the acquired EGG signals are amplified from the micro-volt range to the volt range by setting the gain of the amplifier through programming. Further, the acquired EGG signals are amplified and converted from analog to digital, and they are fed to the microcontroller unit through the Inter-Integrated Circuits (I2C) protocol.

3.1.3. Microcontroller Unit

A Raspberry Pi 3 Model B+ based Microcontroller is used to perform computing operations that are portable, consume less power, and can be connected to the cloud with the help of Internet connectivity. The preprocessing and deep learning algorithms are coded inside the Pi controller using open-source Python programming software. The acquired EGG signals are amplified and preprocessed to remove noise. Further, these preprocessed signals are given to a deep LSTM-based deep learning model for training and testing. Also, the Pi controller updates the decision support produced by the trained deep learning model to the Internet of Things (IoT) cloud platform.

3.1.4. ThingSpeak Internet of Things Cloud Platform

In general, Internet of Things cloud platforms, namely ThingSpeak, are used to visualize and analyze dynamic data remotely. A user account is created on the open-source ThingSpeak Internet of Things cloud platform, and the individual’s food habits, namely pre-prandial and post-prandial conditions, are monitored. The microcontroller unit accesses the ThingSpeak user account with the help of an Application Programming Interface (API) key and stores the decision support, which can be viewed by the individual or doctor personnel remotely at any time. Also, the day-wise individual’s pre-prandial and post-prandial conditions are logged, which helps the individual lead a healthy life.

3.2. Data Acquisition and Analysis

In this work, the Empirical Mode Decomposition (EMD) technique is used to decompose the acquired pre-prandial and post-prandial EGG signals into multiple components called Intrinsic Mode Functions (IMFs) [15]. The frequency of these IMFs is derived using the Fast Fourier Transform (FFT). By analyzing the frequency of all the IMFs, the unwanted IMFs are eliminated and the remaining IMFs are concatenated, which produces a resultant noise-free EGG signal. The frequency components of the sampled EGG signal can be extracted using the FFT algorithm. Also, in this work, the FFT analysis is used to represent the acquired EGG signal in the frequency domain. Further, the FFT of the acquired EGG signal can be computed by the expression (1).
f ( x ) = n = 0 M 1 e 2 π j x n M y ( n )
where f(x) requires a sum of M terms. Also, the frequency with the maximum amplitude will be considered the dominant frequency, which is the fundamental frequency of the particular EGG signal.
LSTM is a type of Recurrent Neural Network (RNN) architecture used in deep learning. Recurrent Neural Network (RNN) tends to have the problem of exploding and vanishing gradients; therefore, it is much more difficult to train. This problem occurs when the gradient either becomes too small or too large during back propagation. This happens in the RNN because they have a recurrent connection that allows them to store information from previous time steps. The LSTM is designed specifically to prevent the problem of an exploding and vanishing gradient. The LSTM architecture uses two types of nonlinear activation functions, namely the logistic sigmoid function and the hyperbolic tangent function.
The sigmoid activation function converts any x coordinate into a y coordinate between 0 and 1. This is used as a gate activation function. Further, the sigmoid activation function is given by the expression (2):
σ x = e x e x + 1
The hyperbolic tangent function converts any x coordinate value and converts it into a y coordinate value between −1 and 1. Further, the function is given by the expression (3):
x = e x e x e x + e x = tanh x
As the name suggests, there are two types of memory in the LSTM architecture, namely long-term memory and short-term memory. The long-term memory is also called the cell state, and it can be modified by arithmetic functions, but there are no weights or biases that can modify the function. The short-term memory is also known as the hidden state. Unlike long-term memory, short-term memory has weights and biases that can modify its function. Also, the LSTM architecture consists of memory blocks, and these memory blocks are a set of recurrently connected sub-networks. Furthermore, the memory block maintains its state over time and regulates the flow of information. A vanilla LSTM unit is composed of four components, such as a cell, an input gate, an output gate, and a forget gate [22].

3.2.1. Forget Gate

This step determines how much of the information must be deleted from its previous cell state. This gate uses the sigmoid activation function for its operation, and this is the first block in the LSTM algorithm.

3.2.2. Input Gate

This step is used to update the value of the previous LSTM cell by combining the input value with its biases and weights and the last LSTM output. This gate also uses the sigmoid activation function for its operations.

3.2.3. Cell

This step combines the values of the input value, the input gate value, the forget gate value, and the previous cell value.

3.2.4. Output Gate

This step combines the current input value, the output of the LSTM unit, and the cell value of the last LSTM unit.
The input and output values of the LSTM are combined with the cell values of the last unit and the current unit to obtain the block input value and the block output value, respectively. The values are introduced into the LSTM unit in the form of block input, and the output value is received in the form of block output. In this work, a total of 200 EGG signals are acquired from normal individuals, out of which 100 EGG signals are acquired under the pre-prandial condition and 100 EGG signals are acquired under the post-prandial condition. Furthermore, 80% of the total EGG signals are utilized to train the LSTM deep learning model, and the remaining 20% of the EGG signals are utilized to test the proposed LSTM deep learning model. Also, the proposed deep learning model is incorporated into the developed wearable device, and the device updates the decision support in the Internet of Things cloud.

4. Results and Discussion

Figure 3a,b show a typical EGG signal acquired from normal individuals under pre-prandial and post-prandial conditions. It is observed that the x-axis shows the amplitude of the acquired EGG signal in volts, and the y-axis shows the sample data points acquired at different points in time. Also, it is seen in Figure 3 that the typical EGG signal acquired from normal individuals under pre-prandial and post-prandial conditions has no significant variation by visual examination except for the change in amplitude of both EGG signals.
This study is conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Committee of Gleneagles Global Health City, Chennai, India. (Reference number: BMHR/2023/0055). For this proposed study, a total of 200 EGG signals were acquired under pre-prandial and post-prandial conditions from normal individuals with their proper consent.
The EGG signals are analyzed using FFT, and it is observed that there are no significant changes in frequency of the acquired pre-prandial and post-prandial EGG signals. Moreover, 80 pre-prandial EGG signals and 80 post-prandial EGG signals are given to the proposed LSTM deep learning model for the training process. Also, 20 pre-prandial EGG signals and 20 post-prandial EGG signals are given to the proposed LSTM deep learning model for testing. The confusion matrix generated after the testing process is shown in Figure 4. Figure 4 shows the confusion matrix of binary classes, namely pre-prandial and post-prandial, generated by the proposed LSTM deep learning model. Also, it is observed that parametric values such as True Positive (TP), False Positive (FP), True Negative (TN), and False Negative (FN) are given appropriately in terms of true values versus predicted values. By using the above-discussed parametric values, the four different performance metrics are calculated, and the four different performance metrics, such as accuracy, F1_Score, precision, and recall, of the proposed LSTM deep learning model are presented in Table 1.
From Table 1, it is evident that the accuracy of the proposed LSTM model is 96% and the recall of the proposed LSTM deep learning model is 98%. Further, the precision and F1_Score of the proposed LSTM deep learning model are 94.2% and 96%, respectively. Also, it is evident that the proposed LSTM deep learning model is capable of identifying an individual’s food habits. Figure 5 shows the ThingSpeak user account for monitoring pre-prandial and post-prandial conditions. Further, the food habits can be monitored by oneself or another person, especially a doctor, by giving proper access. It is shown that the location of the person can also be visualized on the user page. The predicted output of the proposed LSTM deep learning model is logged to the field 2 Chart of the user account using the API key. Furthermore, it is also seen that in the field 2 Chart, the individual food habits are logged with respect to date and time. From the literature, it is evident that the skipping/late consumption of food leads to various digestive abnormalities; however, it is evident that the proposed Internet of Things-based smart digestive health monitoring device is highly efficient at identifying the individual’s food habits/consumption results in maintaining a healthy life since digestion plays a vital role in every human’s life.

5. Conclusions

In general, the acid produced by the stomach can sometimes leak into the esophagus because of improper closure of the cardiac sphincter, causing a burning sensation in the esophagus with symptoms such as regurgitations, belching, and coughing. The main cause of this disease is basically linked to the patient’s lifestyle and eating habits. In this work, a wearable device was designed and developed to monitor the food intake habits of normal individuals to maintain a healthy lifestyle. Further, the EGG signals were acquired from normal individuals for pre-prandial and post-prandial conditions, and the LSTM deep learning model was utilized to identify the food intake habits of the normal individuals. Results demonstrate that the proposed LSTM model is good at classifying pre-prandial and post-prandial conditions and exhibits an accuracy of 96%. Also, the ThingSpeak Internet of Things cloud platform helps normal individuals monitor their food intake habits day-to-day in a remote manner anytime since the data are being logged regularly to the ThingSpeak Internet of Things user account. Since the proposed device is compact and can be integrated into usual clothing, it shall be used to effectively monitor digestive habits, namely pre-prandial and post-prandial conditions, which leads to a healthy life.

Author Contributions

R.K.D., A.P., S.V., and C.E. conceptualized the idea for this manuscript. R.K.D. provided the resources. A.P. designed and developed the hardware for data acquisition. P.P. and M.A. carried out the investigation and data curation of the acquired data. S.V. validated the data and results acquired. A.P. prepared the original draft. S.V., C.E., and P.S.M. reviewed and edited the original draft. C.E. and P.S.M. supervised the work, and R.K.D. administered the work. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Ethics Committee of Gleneagles Global Health City, Chennai, India. (Reference number: BMHR/2023/0055).

Informed Consent Statement

Informed consent was obtained from all subjects involved in this study.

Data Availability Statement

The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical restrictions.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Sharma, D.C.; Naqvi, A.; Chawla, S.; Gulati, C.; Kumar, S.; Singh, N.; Zehra, S. A Survey on Association Between Lifestyle Related Causes of Gastroesophageal Reflux Disease. J. Exp. Zool. India 2022, 25, 961–964. [Google Scholar]
  2. Vernia, F.; Di Ruscio, M.; Ciccone, A.; Viscido, A.; Frieri, G.; Stefanelli, G.; Latella, G. Sleep disorders related to nutrition and digestive diseases: A neglected clinical condition. Int. J. Med. Sci. 2021, 18, 593. [Google Scholar] [CrossRef] [PubMed]
  3. Jardim, S.R.; de Souza, L.M.P.; de Souza, H.S.P. The Rise of Gastrointestinal Cancers as a Global Phenomenon: Unhealthy Behavior or Progress? Int. J. Environ. Res. Public Health 2023, 20, 3640. [Google Scholar] [CrossRef] [PubMed]
  4. Raihan, M.M.S.; Shams, A.B.; Preo, R.B. Multi-Class Electrogastrogram (EGG) Signal Classification Using Machine Learning Algorithms. In Proceedings of the 2020 23rd International Conference on Computer and Information Technology (ICCIT), Dhaka, Bangladesh, 19–21 December 2020; pp. 1–6. [Google Scholar] [CrossRef]
  5. Fred, H.L. Drawbacks and limitations of computed tomography: Views from a medical educator. Tex. Heart Inst. J. 2004, 31, 345–348. [Google Scholar] [PubMed]
  6. Atar, M.; Kadayifci, A. Transnasal endoscopy: Technical considerations, advantages and limitations. World J. Gastrointest. Endosc. 2014, 6, 41–48. [Google Scholar] [CrossRef] [PubMed]
  7. Alagumariappan, P.; Krishnamurthy, K.; Jawahar, P.M. Selection of surface electrodes for electrogastrography and analysis of normal and abnormal electrogastrograms using Hjorth information. Int. J. Biomed. Eng. Technol. 2020, 32, 317–330. [Google Scholar] [CrossRef]
  8. Sangeethalakshmi, K.; Preethi, U.; Pavithra, S. Patient health monitoring system using Internet of Things. Mater. Today Proc. 2023, 80, 2228–2231. [Google Scholar]
  9. Gopu, G.; Neelaveni, R.; Porkumaran, K. Acquisition and analysis of electrogastrogram for digestive system disorders using a novel approach. In Proceedings of the 2008 International Conference on Electrical and Computer Engineering, Dhaka, Bangladesh, 20–22 December 2008; pp. 65–69. [Google Scholar]
  10. Haddab, S.; Laghrouche, M. Microcontroller-based system for electrogastrography monitoring through wireless transmission. Meas. Sci. Rev. 2009, 9, 122. [Google Scholar] [CrossRef]
  11. Gopu, G.; Neelaveni, R.; Pokumaran, K.; Shekar, M.G. An enhanced technique for recording and analysis of electrogastrogram using active electrodes. Sri Lanka J. Bio-Med. Inform. 2010, 1, 21. [Google Scholar] [CrossRef]
  12. Gharibans, A.A.; Smarr, B.L.; Kunkel, D.C.; Kriegsfeld, L.J.; Mousa, H.M.; Coleman, T.P. Artifact rejection methodology enables continuous, noninvasive measurement of gastric myoelectric activity in ambulatory subjects. Sci. Rep. 2018, 8, 5019. [Google Scholar] [CrossRef] [PubMed]
  13. Alagumariappan, P.; Krishnamurthy, K. An approach based on information theory for selection of systems for efficient recording of electrogastrograms. In Proceedings of the International Conference on Computing and Communication Systems: I3CS 2016, NEHU, Shillong, India, 11–13 November 2018; Springer: Singapore, 2018; pp. 463–471. [Google Scholar]
  14. Gharibans, A.A.; Coleman, T.P.; Mousa, H.; Kunkel, D.C. Spatial patterns from high-resolution electrogastrography correlate with severity of symptoms in patients with functional dyspepsia and gastroparesis. Clin. Gastroenterol. Hepatol. 2019, 17, 2668–2677. [Google Scholar] [CrossRef] [PubMed]
  15. Alagumariappan, P.; Krishnamurthy, K.; Kandiah, S.; Cyril, E.; Rajinikanth, V. Diagnosis of Type 2 Diabetes Using Electrogastrograms: Extraction and Genetic Algorithm–Based Selection of Informative Features. JMIR Biomed. Eng. 2020, 5, e20932. [Google Scholar] [CrossRef]
  16. Kumar, G.P.; Prakash, S.O.; Sangeetha, B.; Asha, R.; Suganthi, L.; Divya, B. Wireless Real-Time Electrogastrography Monitoring System. J. Comput. Theor. Nanosci. 2020, 17, 3724–3732. [Google Scholar] [CrossRef]
  17. Paramasivam, A.; Najumnissa Jamal, D.; Emmanuel, C.; Bhaskar, K.B.; Mohit Jaisingh, M.; Kannan, R. Analysis of Influence of Yoga-Asana on the Digestive Process Using Electrogastrograms. In Proceedings of the International Conference on Computing and Communication Systems: I3CS 2020, NEHU, Shillong, India, 28–30 April 2020; Springer: Singapore, 2021; pp. 423–429. [Google Scholar]
  18. Paramasivam, A.; Kamalanand, K.; Emmanuel, C.; Mahadevan, B.; Sundravadivelu, K.; Raman, J.; Jawahar, P.M. Influence of electrode surface area on the fractal dimensions of electrogastrograms and fractal analysis of normal and abnormal digestion process. In Proceedings of the 2018 International Conference on Recent Trends in Electrical, Control and Communication (RTECC), Chennai, India, 20–22 March 2018; pp. 245–250. [Google Scholar]
  19. Rajagopal, A.; Alagumariappan, P.; Krishnamurthy, K. Development of an automated decision support system for diagnosis of digestive disorders using electrogastrograms: An approach based on empirical mode decomposition and K-means algorithm. In Disruptive Technology: Concepts, Methodologies, Tools, and Applications; IGI Global: Hershey, PA, USA, 2020; pp. 661–678. [Google Scholar]
  20. Chowdhury, S.D.; George, G.; Ramakrishna, K.; Ramadass, B.; Pugazhendhi, S.; Mechenro, J.; Jeyaseelan, L.; Ramakrishna, B.S. Prevalence and factors associated with gastroesophageal reflux disease in southern India: A community-based study. Indian J. Gastroenterol. 2019, 38, 77–82. [Google Scholar] [CrossRef] [PubMed]
  21. Parkman, H.P.; Hasler, W.L.; Barnett, J.L.; Eaker, E.Y. Electrogastrography: A document prepared by the gastric section of the American Motility Society Clinical GI Motility Testing Task Force. Neurogastroenterol. Motil. 2003, 15, 89–102. [Google Scholar] [CrossRef] [PubMed]
  22. Van Houdt, G.; Mosquera, C.; Nápoles, G. A review on the long short-term memory model. Artif. Intell. Rev. 2020, 53, 5929–5955. [Google Scholar] [CrossRef]
Figure 1. Overall block diagram of the proposed approach.
Figure 1. Overall block diagram of the proposed approach.
Engproc 58 00033 g001
Figure 2. Block diagram of the proposed wearable device.
Figure 2. Block diagram of the proposed wearable device.
Engproc 58 00033 g002
Figure 3. Typical EGG signal acquired under (a) pre-prandial conditions; (b) post-prandial conditions.
Figure 3. Typical EGG signal acquired under (a) pre-prandial conditions; (b) post-prandial conditions.
Engproc 58 00033 g003
Figure 4. Confusion matrix for pre-prandial and post-prandial conditions.
Figure 4. Confusion matrix for pre-prandial and post-prandial conditions.
Engproc 58 00033 g004
Figure 5. ThingSpeak user account for monitoring pre-prandial and post-prandial conditions.
Figure 5. ThingSpeak user account for monitoring pre-prandial and post-prandial conditions.
Engproc 58 00033 g005
Table 1. Performance metrics of the LSTM deep learning model.
Table 1. Performance metrics of the LSTM deep learning model.
Performance MetricsPercentage (%)
Accuracy96
Precision94.2
Recall98
F1_Score96
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Dhanaraj, R.K.; Paramasivam, A.; Vijayalakshmi, S.; Emmanuel, C.; Pallavi, P.; Metkewar, P.S.; Ashwin, M. Design and Implementation of an IoT Based Smart Digestive Health Monitoring Device for Identification of Digestive Conditions. Eng. Proc. 2023, 58, 33. https://doi.org/10.3390/ecsa-10-16253

AMA Style

Dhanaraj RK, Paramasivam A, Vijayalakshmi S, Emmanuel C, Pallavi P, Metkewar PS, Ashwin M. Design and Implementation of an IoT Based Smart Digestive Health Monitoring Device for Identification of Digestive Conditions. Engineering Proceedings. 2023; 58(1):33. https://doi.org/10.3390/ecsa-10-16253

Chicago/Turabian Style

Dhanaraj, Rajesh Kumar, Alagumariappan Paramasivam, Sankaran Vijayalakshmi, Cyril Emmanuel, Pittu Pallavi, Pravin Satyanarayan Metkewar, and Manoj Ashwin. 2023. "Design and Implementation of an IoT Based Smart Digestive Health Monitoring Device for Identification of Digestive Conditions" Engineering Proceedings 58, no. 1: 33. https://doi.org/10.3390/ecsa-10-16253

APA Style

Dhanaraj, R. K., Paramasivam, A., Vijayalakshmi, S., Emmanuel, C., Pallavi, P., Metkewar, P. S., & Ashwin, M. (2023). Design and Implementation of an IoT Based Smart Digestive Health Monitoring Device for Identification of Digestive Conditions. Engineering Proceedings, 58(1), 33. https://doi.org/10.3390/ecsa-10-16253

Article Metrics

Back to TopTop