Detecting And Diagnosing COVID-19 & TB Using Machine Learning

Project Title

Detecting And Diagnosing COVID-19 & TB Using Machine Learning

Project Area of Specialization Artificial IntelligenceProject Summary

Worldwide COVID-19 and Tuberculosis (TB) is one of the very fatal diseases there are many different methods through which we can diagnose these diseases but these methods of diagnosis are expensive and a country like Pakistan has less technology through which the death ratio increases. We will resolve this issue of late diagnosis of these diseases by taking the cough of any person and it will tell that a person is having COVID-19 or TB. Our model will work on classification it will take cough as an input and train through this data. Our model will help in early detection because the major issue which make these diseases deadly is late detection and our model will be easy to use, easily available and less complex so that less literate people can easily use it.

Project Objectives

The test or diagnosis of TB through different methods are too expensive. such as Smear microscopy, spectrum culture and chest radiography but these methods are too expensive, therefore in our project we tried to grab symptoms like cough and by analyzing the cough sound we will do the detection and predict whether person is having TB or not using deep learning. Our system is less expensive and less complex than other methods of TB diagnosis moreover it is easily accessible which will help in early detection.

Project Implementation Method

With breakthroughs in deep learning and neural networks, it is now possible to build simple yet powerful networks that can do classification, forecasting, and object detection tasks that were previously difficult or impossible to accomplish. Convolutional neural networks (CNNs, or ConvNets) are a type of deep neural network used to analyze visual imagery in deep learning. When given the right data and a specific goal, they're quite good at recognizing and classifying objects and patterns in photos.

However, it turns out that these CNNs aren't just for computer vision problems; they may also be used to solve non-image problems. For example, a sound can be transformed into a spectrogram, which is a graph that indicates the amount of each frequency in an audio file at each moment. This method has been shown to outperform state of the art environmental sound detection methods.

How our model works at the beginning we will collect cough samples of users then label it after the pre processing of data we will do feature extraction then feed our data into CNN which will predict disease after the training of our model we will deploy our model on cloud platform. Now our user can record their cough through our mobile app then our model will predict the result which will show to the user on mobile screen.

Benefits of the Project

Expensive cost , shortage and long turnaround time of clinical testing of covid 19 and Tb are the key factors of spread out and deaths occurred  through these diseases therefor the main benefits of our project will be that it will overcome these constrains and help in early detection of these diseases because late detection is one of the main key factor in spread out approximately one undiagnosed TB patient can infect 96 other persons and one Covid 19 patients can infect 100+ persons and these are called super spreaders. Some other major benefits are,

·Early detection

·Cheap

·Less complex

·Availability

·No skilled person needed to interpret results

Technical Details of Final Deliverable

The technical details for the final delivery product are as follows:

• Clean Dataset of Cough of TB and COVID19 Patient.
• Suitable Algorithm for classification of cough.
• A model trained for detecting TB and COVID-19 through a cough.
• Database deployed on the cloud.
• Live Application on which users can interact easily. 
• Documentation.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Medical Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable 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)	50955
Amazon Sage Maker Studio Notebooks	Equipment	1	1000	1000
Web hosting	Equipment	1	4380	4380
Domain Name	Equipment	1	1650	1650
Blue Yeti Professional Multi-Pattern USB Microphone	Equipment	1	29340	29340
Google Play store	Equipment	1	4585	4585
Data Collection	Miscellaneous	1	9000	9000
Printing	Miscellaneous	1	1000	1000