Detection of Pulmonary Diseases through Cough

Project Title

Detection of Pulmonary Diseases through Cough

Project Area of Specialization Artificial IntelligenceProject Summary

Over the past few decades, significant innovations have been made in the field of medicine and medical science. Since the use and employment of basic medicine to treat illness ranging from basic level to chronic diseases, the domain of medical science has undergone significant innovations which can be seen implemented today in the form of advanced healthcare technologies. Lung diseases are one of the most frequent and recurrent medical conditions faced by millions of people all around the world. In the United States alone, tens of millions of people face respiratory and lung related diseases caused by various factors such as smoking, infections, genes, and polluted air intake [1]. The 3rd Major cause of Annual Deaths worldwide are Pulmonary Diseases. Pulmonologists have classified the most common lung diseases as Asthma, Bronchitis, Tuberculosis, Punemnia, COVID etc. Our Final Year Research based project is based upon how to classify a simple cough as a case to be taken. This research project is based upon classification of a cough as either Wet or Dry.

Project Objectives

The research and development carried out for the establishment of an android application for the detection of respiratory and pulmonary diseases were divided into various goals and objectives. The goal of this study is to implement machine learning for the detection of respiratory diseases using cough samples. The research carried out is divided into various objectives which are highlighted below: 1. Literature Review 2. Dataset Gathering 3. Machine Learning Model Training 4. Development of an Android Application 5. Testing of the Application.

Project Implementation Method

ResNet50 is a machine learning algorithm that discriminates different sounds of different vocal cord strengths. The ‘mmmm’ sound is generated when we cough which will give an indication of how strong or weak a person's vocal cords are. Sound events will be detected based on different features. In this work we will record three types of audios from every person giving audio samples: one silent just breathing sound secondly normal talking and lastly cough and from here we can take a mean of the audio pressure level and any sound above mean level will be considered as a cough. So our model will detect all types of cough which are mainly wet and dry coughs and then further categorize and detect diseases like pneumonia, asthma, allergies, tuberculosis, high fever, chest pains. After detecting a disease, we will suggest a natural cure if possible or suggest an appointment with a doctor.

Benefits of the Project

This application will be available to the general public. Even a person who possesses a mobile phone which has internet access can use this application anywhere in the world. Not only will doctors be able to access it but all kinds of people that are concerned about their health will also have access to it. Making it an android application it will be available to maximum users. As all the past work we have read related to cough detection they are primarily detecting only Covid-19 which no doubt can be found and spread through cough primarily but what about other cough diseases which directly affect the lungs, the vital organ in the human body.

Technical Details of Final Deliverable

For a new user, they will open the app and select the sign-up button. Then they will fill the form which popped up Afterwards they will click sign up and their account will be made. Then they will enter their Username and Password and log in.

REQ-SF1-1: When the user is signed up, the added information of that specific user will be updated in our real-time database (Firebase).

REQ-SF1-2: User login information should be verified, and then allowed access.

REQ-SF1-3: Using the forget password link user should be able to change their password.

Description and Priority

The user who hasregistered on the application will enter hissymptoms based on the questionnaire that will be presented based on the domain of the disease. The user will answer the questionnaire and the artificial intelligence algorithm will analyze and diagnose the disease based on the symptoms of the patient

Stimulus/Response Sequences

The user will enter the domain of his potential disease. A questionnaire will be presented. The user will fill the questionnaire based on symptoms

The artificial intelligence algorithm will diagnose the disease based on the symptoms the patient entered and cross-reference it with the symptoms of the disease entered in the system database

RESULT

The user can view the results of the analysis algorithm generated.

Final Deliverable of the Project Software SystemCore Industry MedicalOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
SSD's to store huge amount of data as its in audio format .wav	Equipment	4	17500	70000
Travel	Miscellaneous	1	10000	10000