DEVELOPMENT OF AN AI-DRIVEN APPLICATION TO

DETECT ASTHMA USING MACHINE LEARNING

ALGORITHM.

BY
TAJUDEEN SAMAD OLANREWAJU(RUN/CMP/22/13067)
AJAYI OLUWOLE TEMITOPE(RUN/CMP/21/10829)
OMOTAYO FAVOUR(RUN/CMP/22/13063)
OGBA- IROKALU MIRIAM(RUN/CMP/21/10894)

CHAPTER ONE

ABSTRACT

Asthma is a common chronic respiratory disease that affects millions of

people worldwide, with significant under-diagnosis and under-treatment,

especially in resource-poor settings. Traditional diagnostic methods often

require specialized equipment and expertise that may not be readily

available in these settings. Recent advances in artificial intelligence (AI)

and machine learning (ML) have shown promise in improving disease

detection and diagnosis. This study aims to develop an AI-based

application that uses machine learning algorithms to improve asthma

detection. By analyzing patient symptoms, medical history, and

environmental factors, the proposed system aims to facilitate early

diagnosis, reduce diagnostic errors, and increase access to asthma

detection tools, especially in under-resourced settings.

 INTRODUCTION

1.1 Backgound of Study

Asthma is a long-term disease that affects the airways in the lungs,

making it difficult for people to breathe. It is one of the most common

chronic illnesses, affecting over 300 million people worldwide, and its

prevalence continues to rise (Global Initiative for Asthma [GINA], 2023).

Asthma symptoms, such as wheezing, coughing, and shortness of breath,

can range from mild to life-threatening. The condition can be triggered by

environmental factors, allergens, or physical activity, making it difficult

to manage without proper diagnosis and treatment (Papi et al., 2021).

While asthma cannot be cured, early detection and proper management

can significantly improve the quality of life for those affected.

Diagnosing asthma can be challenging because its symptoms often

overlap with other respiratory diseases, such as chronic obstructive

pulmonary disease (COPD) and bronchitis (Reddel et al., 2022).

Traditional diagnostic methods include spirometry, peak flow

measurements, and bronchial provocation tests, but these require

specialized equipment and trained professionals (Lai et al., 2022). Many

healthcare facilities, especially in low-resource settings, lack access to

these tools, leading to underdiagnoses or misdiagnosis of asthma cases.

Moreover, asthma symptoms can vary from person to person, making it

difficult to rely on a single test for an accurate diagnosis (Gupta et al.,

2022).

Advancements in artificial intelligence (AI) and machine learning (ML)

are transforming the healthcare industry by improving the accuracy and

efficiency of disease detection (Topol, 2019). AI-powered systems can

analyze large datasets, identify patterns, and assist healthcare providers in

making more informed decisions. In recent years, machine learning has

been successfully applied to the detection of respiratory diseases such as

pneumonia, tuberculosis, and COPD, showing promise for its application

in asthma diagnosis as well (Shen et al., 2021).

Machine learning algorithms, such as decision trees, support vector

machines (SVM), and neural networks, can process patient data,

including symptoms, medical history, and environmental factors, to

predict the likelihood of asthma (Yu et al., 2023). These algorithms can

analyze thousands of cases in seconds, helping doctors detect asthma

earlier and with greater accuracy than traditional methods. Additionally,

AI can be integrated into mobile applications and wearable devices,

allowing individuals to monitor their breathing patterns and receive real-

time alerts if asthma symptoms worsen (Laguarta et al., 2020).

This study aims to develop an AI-driven application that uses machine

learning algorithms to detect asthma more effectively. The proposed

system will analyze patient symptoms, medical history, and

environmental factors to identify potential asthma cases. By leveraging

AI technology, this research seeks to improve early diagnosis, reduce

misdiagnosis, and make asthma detection more accessible, particularly in

areas where specialized healthcare services are limited.

1.2 Statement of the Problem

As mentioned earlier, Asthma is a chronic respiratory disease that affects

millions of people around the world, causing economic burden,

significant morbidity, and mortality. Despite its prevalence, asthma still

remains undertreated and underdiagnosed, particularly in resource-

constrained settings. Current diagnostic methods, such as spirometry and

clinical evaluations, require specialized expertise, expensive and can be

time-consuming therefore, predictive models are needed to solve this

problem and aid medical practioners in early diagnosis and treatment,

reducing the mortality rate caused by the disease,hence the study.

1.3 Aim and Objectives

The study aims to improve the existing machine-learning algorithm

(Models), which would be used to develop a model for detecting Asthma.

The specific objectives are to:

i. Acquire a adataset for Asthma dataset

ii. Implementation of various machine learning dataset.

iii. Develop model capable of distinguishing asthma from other

respiratory diseases.

iv. Evaluate the performance of the algorithms and the developed model

1.4 Scope of the Study

Asthma is the most chronic disease among children affecting millions of

people and causing thousands of deaths. Most deaths occur in low-and
lower-middle-income countries, where under-diagnosis and under-
treatment is a challenge. The idea is to diagnose the disease at an early
stage in order to reduce the mortality rate causeed by Asthma.

1.5 Significance of the Study

This project will help reduce over-reliance on subjective symptoms,

limited diagnostic tools, and the unavailability of specialized healthcare

providers in many regions. By developing an AI-driven application

powered by a descriptive machine learning (ML) model, this study seeks

to offer a transformative solution to bridge this gap.

1.6 Definition of Terms

1. Asthma

Asthma is a chronic inflammatory disease of the airways that causes

recurring episodes of shortness of breath, coughing, wheezing, and chest

tightness. It is a complex condition that involves the interaction of

multiple cellular elements, which includes mast cells, T Lymphocytes,

airway epithelial cells, and eosinophils (Holgate & Polosa, 2006).

According to (WHO) World Health Organization, asthma affects

approximately 340 million people worldwide (WHO, 2020).

2. Artificial Intelligence (AI)

Artificial Intelligence (AI) refers to the development of computer systems

that can perform tasks that typically require human intelligence, such as

decision-making, learning, and problem-solving. AI has been increasingly

applied in healthcare to improve treatment, patient outcomes, and

diagnosis (Rajkomar et al., 2019).

3. Machine Learning (ML)

Machine Learning (ML) is a subset of AI that involves training

algorithms to learn patterns and make decisions or predictions based on

available data. Machine Learning (ML) has been widely used in

healthcare to identify patterns that can inform clinical decision-making

and analyse large datasets (Koh & Tan, 2011).

4. Descriptive Machine Learning

Descriptive Machine Learning refers to a type of Machine Learning that

focuses on identifying relationships and patterns in data to gain

understanding and insights. It is often used in exploratory data analysis to

identify anomalies, correlations and trends (Fayyad et al., 1996).

5. Asthma Detection

Asthma detection refers to the process of identifying individuals who are

likely to have asthma based on various factors, such as environmental

exposure, symptoms, and medical history. Accurate asthma detection is

critical for early prevention of complications, intervention, and treatment

(GINA, 2020).

1.7 Proposed Methodology

1. A brief description is given on how to achieve this project's objectives.

The first objective involves the acquiring of an Asthma disease datasets

that would be used to develop and train the machine learning model
project. Publicly available asthma datasets (e.g., Kaggle, UCI Machine

Learning Repository) will be used, as well as collaboration with

healthcare providers for clinical data. The machine learning model will be

designed to predict Asthma in humans. Factors such as the sample set

(split data for training and testing), domain and scaling will be considered

before choosing the datasets.

2. The second objective will be the cleaning, normalization, and feature

extraction from raw data. Handling missing data and balancing datasets to

aid the development of a machine-learning algorithm in which a random

forest classifier will be developed to be later used in objective. This will

be achieved when the datasets obtained are processed using data

processing techniques by separating the labels from the features for the

specification of the machine learning model as well as various Python

libraries(e.g Scikit-learn, Pandas), which are used for identifying and

classifying the data errors in the datasets. The data will be divided into

two parts 20% of the data for the testing and 80% for the training of the

datasets, and the model will be trained based on factors like data size and

data variability. And the machine learning algorithm used in the random

forest classifier because the data will be multi-class datasets.

3. To achieve objective three, evaluation of the model will be developed

with machine learning algorithms and would be done based on the

performance (accuracy), precision, recall, test harness(train/test split).

The model will be built and evaluated using python programming

language.

4. There will then be a development of a user-friendly application (e.g.,

mobile or web-based) integrating the ML model.

5. There will be validation and testing by employing statistical methods

to evaluate performance. Hence, conducting usability testing with

healthcare professionals.

1.8 Implementation and Expected Result

The implementation involves acquiring relevant datasets, applying

various machine learning algorithms, and developing a model capable of

distinguishing asthma from other respiratory diseases. The expected

outcome is a user-friendly application that can assist healthcare providers

and individuals in the timely and accurate detection of asthma, thereby

improving patient outcomes and reducing the disease's burden.