AI Based Asthma Recognition System Using Respiratory Sound

Abstract

Asthma is a chronic respiratory condition that affects millions of people worldwide, leading to significant healthcare costs and a reduced quality of life for affected individuals. Timely and accurate detection of asthma exacerbations is crucial for effective management and intervention. Traditional methods of asthma diagnosis rely heavily on clinical assessments, which may not always provide real-time, objective data for prompt action. The aim of this project is to develop an AI-based asthma detection system that leverages the analysis of respiratory sound patterns. Current diagnostic tools often lack the ability to capture subtle changes in respiratory sounds that could indicate the onset or worsening of asthma symptoms. By employing advanced machine learning algorithms, this system aims to identify distinctive patterns and anomalies in respiratory sounds associated with asthma, enabling early detection and intervention.

Introduction

I. INTRODUCTION

Asthma stands as a pervasive and challenging chronic respiratory condition affecting millions worldwide, contributing to substantial healthcare costs and diminishing the quality of life for those afflicted. Timely and accurate diagnosis of asthma exacerbations is crucial for effective management, enabling proactive interventions and improved patient outcomes. Traditional diagnostic approaches often rely on clinical assessments and spirometry tests, lacking the real-time, objective, and non-invasive characteristics necessary for comprehensive asthma recognition. This research endeavors to pioneer an innovative paradigm in asthma detection by proposing an AI-based Asthma Recognition System utilizing deep learning methodologies, specifically focusing on the analysis of respiratory sound patterns. Leveraging the transformative power of artificial intelligence, this system aims to fill existing gaps in diagnostic capabilities, offering a promising avenue for early and accurate asthma detection. The prevalence of asthma's diverse manifestations poses a formidable challenge to the development of a universal diagnostic model. Traditional methods struggle to capture the nuanced variations in respiratory sounds that may signify the onset or exacerbation of asthma symptoms. In response, this research harnesses the capabilities of deep learning, employing Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to discern intricate patterns within respiratory sound data. The project's foundation lies in the compilation of a comprehensive and diverse dataset of respiratory sounds, representative of various asthma conditions and severity levels. Deep learning algorithms, adept at feature extraction and pattern recognition, are then trained on this dataset to identify subtle yet distinctive sound patterns associated with asthma. The proposed Asthma Recognition System is poised to transcend the limitations of current diagnostic approaches, providing a real-time and data-driven solution for asthma detection. As this research unfolds, it not only seeks to address the technical challenges of variability in sound patterns and real-time processing but also emphasizes the importance of noise reduction techniques and interference mitigation to enhance the system's accuracy and reliability. Furthermore, the project aims to validate and evaluate the system rigorously, comparing its performance against established diagnostic methods and assessing its potential impact on clinical decision support and healthcare analytics.

In envisioning the convergence of artificial intelligence, deep learning, and respiratory sound analysis, this research aspires to contribute significantly to the evolving landscape of healthcare technology. The proposed Asthma Recognition System holds the promise of revolutionizing asthma diagnosis, offering a non-invasive, real-time, and objective approach that could positively impact the lives of individuals affected by this chronic respiratory condition.

II. LITERATURE SURVEY

1. This Paper presents the research on audio classification and acoustic analysis of inhalers sounds of asthma patient on medication adherence which could use to improve aerosel theory It uses machine learning algorithms for anlaysis and recognition of inhalers sound to detect asthma published by NIKOS DIMITRIS FAKOTAKIS and STAVROS NOUSIAS. "AI Sound recognition on asthma medication adherence"
2. In this research paper, Describes how deep learning could be used in the recognition of respiratory disease just from the respiratory sound. Respiratory audios are important indicators of respiratory health and respiratory disorder. For example, a wheezing sound is a common sign that a patient has an obstructive airway disease like asthma or chronic obstructive pulmonary disease (COPD).Victor Basu, "Respiratory diseases recognition through respiratory sound with the help of deep neural network."- 2020 4th International Conference on Computational Intelligence and Networks(CINE)
3. In this work, they explored the combination of various feature extraction techniques and classifier architectures. they propose an end-to-end pipeline, R-STFT, for classifying lung sounds, combining STFT and the pre-trained ResNet18 image classifier. Zizaho Chen, Xilin liu; " Classify Respiratory Abnormality in Lung Sounds Using STFT and a Fine-Tuned ResNet18 Network"
4.  In this paper large scale of lung sound database was used. With the help of seven indexes describing pathological breath and sound intensities they proposed ai solution, "Artificial Intelligence approach to the monitoring of respiratory sounds in asthmatic patients" Department of Computer Science, Christ University, Bangalore, India

III. AIM & OBJECTIVES

1. Improve Diagnosis of Asthma
2. Early and Accurate Diagnosis
3. Prevent Asthma at early stage
4. Reduce cost of hospital bills
5. Effective solution for detection of asthma

V. APPLICATIONS

The asthma recognition system can be used in the following:

1. Early Diagnosis and Intervention of Asthma
2. Remote Patient Monitoring

VI. FUNCTIONAL & NON-FUNCTIONAL REQUIREMENTS

A. Functional Requirements

1. Respiratory Sound Data Collection: The system should be able to collect and store a diverse dataset of respiratory sound recordings, encompassing various asthma conditions and severity levels.
2. Data Preprocessing: Implement preprocessing algorithms to clean and standardize respiratory sound data, addressing issues such as noise, artifacts, and variations in recording quality.
3. Deep Learning Model Implementation: Develop and implement deep learning models, such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), for the analysis of respiratory sound data.
4. Model Generalization: Ensure that the trained model can generalize well to new and unseen data, accommodating diverse populations and asthma conditions not present in the training dataset.

B. Non-Functional Requirements

1. Performance

a. Response Time: The system should provide real-time processing with low-latency response times to ensure timely identification of asthma-related patterns.

b. Throughput: The system should be capable of handling a high volume of respiratory sound data efficiently, especially in scenarios involving simultaneous monitoring of multiple individuals.

2. Reliability

a. Availability: The system should have high availability, minimizing downtime to ensure continuous monitoring and timely alerts.

b. Fault Tolerance: The system should be designed to handle errors gracefully and maintain functionality in the presence of faults.

3. Scalability

The system should be scalable to accommodate an increasing number of users, data inputs, and connected devices without a significant degradation in performance.

4. Security

a. Data Encryption: Ensure that sensitive patient data is encrypted during transmission and storage to protect against unauthorized access.

b. Access Control: Implement access controls to restrict system access to authorized personnel, ensuring data privacy and compliance with healthcare regulations.

VII. SYSTEM REQUIREMENTS

A. Hardware Requirements

1. AMD/Intel i3 Processor or above Processor
2. 4GB RAM or above RAM
3. 80 GB or above Hard Disk
4. Graphics Card: Intel HD620 or above
5. A mobile device or Desktop

B. Software Requirements

1. Windows 8.1 or above
2. python
3. html
4. CSS
5. Anaconda IDE

Conclusion

The development and implementation of an AI-based asthma monitoring system represent a significant leap forward in the field of respiratory health. Through the utilization of advanced deep learning techniques, this research aimed to create a non-invasive, real-time solution for the early detection and continuous monitoring of asthma-related patterns in respiratory sounds. The key findings and contributions of this study underscore the potential transformative impact of such a system on asthma management and healthcare practices. The comprehensive dataset compilation, encompassing a diverse range of respiratory sounds, served as a foundational element for training the deep learning model. The selected neural network architecture, incorporating Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), demonstrated its efficacy in extracting relevant features from complex respiratory sound data. The model\'s ability to generalize well to unseen data was a critical aspect, ensuring its adaptability to various populations and asthma conditions. As technology continues to evolve, the proposed AI-based asthma monitoring system holds promise not only for immediate clinical applications but also for contributing valuable data to healthcare analytics, research studies, and the advancement of personalized medicine. By addressing the challenges associated with asthma management, this research paves the way for a more proactive and patient-centric approach to respiratory health.

References

[1] Rocha BM, Mendes L, Couceiro R, Henriques J, Carvalho P, Paiva R. Detection of Explosive Cough Events in Audio Recordings by Internal Sound Analysis Internal Sound Analysis. 39th Annu Int Conf IEEE Eng Med Biol Soc 2017;2761-6 [2] L. Mendes, Ioannis Vogiatzis, Eleni Perantoni, Evangelos Kaimakamis, Ioanna Chouvarda, N. Maglaveras, Jorge Henriques, Paulo de Carvalho, Rui Pedro Paiva. Detection of crackle events using a multi-feature approach. 38th Annu Int Conf IEEE Eng Med Biol Soc. IEEE; 2016;367983. [3] Guntupalli KK, Alapat PM, Bandi VD, Kushnir I. Validation of automatic wheeze detection in patients with obstructed airways and in healthy subjects. J Asthma. 2008;45(10):9037. [4] L. G. Heaney and R. Horne, “Non-adherence in difficult asthma: Time to take it seriously,” Thorax, vol. 67, no. 3, p. 268–270, 2012. [5] G. Kaufman and Y. Birks, “Strategies to improve patients’ adherence to medication,” Nursing Standard (through 2013), vol. 23, no. 49, p. 51, [6] WHO. The top 10 causes of death. [Online]. Available: https://www. who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death [7] L. Pham, H. Phan, A. Schindler, R. King, A. Mertins, and I. McLoughlin, “Inception-based network and multi-spectrogram ensemble applied to predict respiratory anomalies and lung diseases,” in 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC). IEEE, 2021, pp. 253–256. [Online]. Available: https://doi.org/10.1109/EMBC46164.2021.9629857 [8] Y. Kim, Y. Hyon, S. S. Jung, S. Lee, G. Yoo, C. Chung, and T. Ha, “Respiratory sound classification for crackles, wheezes, and rhonchi in the clinical field using deep learning,” Scientific Reports, vol. 11, no. 1, p. 17186, Aug 2021. [Online]. Available: https://doi.org/10.1038/s41598-021-96724-7 [9] N. Jakovljevic and T. Lon ´ car-Turukalo, “Hidden markov model ? based respiratory sound classification,” in International Conference on Biomedical and Health Informatics. Springer, 2017, pp. 39–43. [Online]. Available: https://doi.org/10.1007/978-981-10-7419-6 7 [10] A. H. Falah and J. Jondri, “Lung sounds classification using stacked autoencoder and support vector machine,” in 2019 7th International Conference on Information and Communication Technology (ICoICT), 2019, pp. 1–5. [Online]. Available: https://doi.org/10.1109/ICoICT. 2019.8835278

Copyright

Copyright © 2023 Aditya Somani, Tushar Mahale, Vinayak Vallakatti, Siddhant Patil, Prof. Abhay Gaidhani. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.