1.

INTRODUCTION

1.1 ARTIFICIAL INTELLIGENCE

Artificial intelligence emerged as a result of the need to automate repetitive and complex
tasks, technological developments, and increasing reliance on data. The emergence of AI did
not happen overnight, rather with decades of research, problem-solving, and technological
breakthroughs.

Artificial intelligence refers to the performance of tasks by computer systems that require
human intelligence. AI assists in identifying data patterns and making decisions based on the
fetched data. Techniques such as computer vision, robotics, and natural language processing
are used to achieve these tasks. The very aim of AI is to develop machines that mimic human
intelligence and perform various functions with increased efficiency and accuracy. AI can learn
and recognize patterns and relationships from large, multi-dimensional and multi-model data
sets. For instance, by analysing thousands of MRI scans, AI-powered systems can identify early
patterns of brain tumors that help doctors detect cancer before symptoms appear.

AI is not a single, all-encompassing technology. Instead, it consists of various subfields, such

1.2 GROWING DEMAND FOR AI IN HEALTHCARE

Around the globe, the healthcare industry encounters numerous challenges in achieving its
objectives which include improving population health, caregiver experience, and patient
experience of care as well as reducing the cost of healthcare. Due to the increase in the elderly
population, chronic diseases, and the cost of healthcare, the government, payers, regulators,
and providers face various challenges that lead to the demand for innovation and
transformation of models in healthcare delivery. In addition, the global pandemic highlighted
the shortages in the healthcare workforce and the lack of equity in case of care.

The primary factor that drove the adoption of AI in healthcare is that, with the introduction
of Electronic Health Records (EHRs), imaging technology, genomic data, and real-time patient

1
monitoring, an incomprehensible and obscene quantity of data was generated. Traditional
data analysis was not sufficient for the vast data, demanding highly advanced tools. Thus,
machine learning algorithms and other cognitive technologies are used to assist in the
diagnosis, treatment, and overall management of patient care.

The traditional approaches in medicine use a one-size-fits-all model, while AI analyzes unique
patient data, such as genetic information, lifestyle factors, and medical history, to create a
customized treatment plan unique to each patient. This helps increase patient satisfaction,
decrease side effects, and improve treatment effectiveness.

Another reason for employing AI in healthcare is the critical impact early disease diagnosis
has on individuals and society. When diseases are detected at an early stage, treatment
outcomes can be improved, and the negative impact of the disease can be minimized. There
is a high chance of curing and controlling the disease when it is detected early. For example,
identifying diabetes at an early stage enables changes in lifestyle and prompt treatment to
avert severe complications.

Concurrently, detecting diseases early contributes to preventing the spread of disease in the
community at large. When infectious diseases, such as influenza or tuberculosis are detected
early, more effective measures can be taken to stop their spread to other populations.

1.3 AI IN HEALTHCARE FORECASTING

The predictive, diagnosing, and preventive characteristics of various diseases have changed
with the emergence of artificial intelligence in healthcare. Predictive healthcare refers to the
utilization of AI to accommodate a great amount of clinical and genetic information to
enhance the recognition of diseases and carry out preventive actions, which improves the way
of work for healthcare providers. Machine learning and deep learning have paved the way for
viewing data patterns within large datasets, improving the accuracy of diagnosis, as well as
tailoring treatment progress for medical patients.

AI can handle large amounts of data compared to data handled by humans. For instance, data
from Electronic Health Records (EHRs) together with medical imaging data, can be quickly
reviewed to determine symptoms of diseases. Research has shown that machine learning
techniques, such as convolutional neural networks (CNNs), possess a greater accuracy in
identifying medical conditions, such as cancer, heart disease, and neurological disorders.

The use of AI in healthcare has substantially refined the efficacy and precision in diagnosing
diseases. Due to the large amount of medical data, AI helps detect diseases early. This study
delves into how AI-based predictive analytics enhances disease diagnosis and its applications
in the medical field.

2
 2. AI IN EARLY DISEASE DIAGNOSING

2.1 MACHINE LEARNING IN DISEASE DIAGNOSIS

Machine learning is a part of Artificial Intelligence that enables algorithms to make predictions
and decisions on their own without having to be explicitly programmed for each task.

The machine learning algorithmic models are trained with vast amounts of medical data,
including disease symptoms, which helps detect symptoms of a disease early. It aids in
predicting common conditions and treating them on time. Diseases such as cancer, diabetes,
cardiovascular diseases, and Alzheimer's disease, among others, can be detected early with
the help of ML algorithmic models. The performance of each ML algorithmic model has its
advantages and limitations. For each medical need, different algorithms are used. Often, a
combination of algorithms is used for maximum advantage.

One of the most popular machine learning algorithms is the "decision tree". A decision tree
is a flowchart that shows different choices and their possible results that help in making
decisions, where the Root Node is the starting point that represents the entire dataset.
Branches are lines that connect nodes. It shows the flow from one decision to another.
Internal nodes are points where decisions are made based on input features. Leaf nodes
represent the outcome or prediction.

Decision trees are easy to interpret and understand. However, they are less accurate
compared to other sophisticated algorithms as they are prone to overfitting when the tree is
deep. Decision trees are effective for diseases like eye diseases, kidney diseases, and
cardiovascular diseases.

Figure 1 shows a decision tree model for diagnosing diabetes using factors like blood sugar
levels, symptoms, BMI, and family history. Based on the path followed, the AI system or doctor
decides whether the patient is diabetic, prediabetic, or at low risk.

2.2 AI IN MEDICAL IMAGING

Medical imaging encompasses various techniques like X-rays, Computed Tomography (CT)
scans, and Magnetic Resonance Imaging (MRI) to create visual representations of the human
body, and help in diagnosis, monitoring, and treatment of medical conditions. The primary
reason for AI's emergence in medical imaging is the growth of radiological imaging data and
the decline in the number of available trained readers.

3
 figure 1: decision tree for diabetes diagnosis using patient data

Two AI methods are widely used in medical imaging. The first method relies on engineered
features extracted from regions of interest on the basis of expert knowledge. Examples of
these features in cancer characterization include tumour volume, shape, texture, intensity and
location. The most robust features are selected and fed into machine learning classifiers. Tools
like PyRadiomics, Support Vector Machine (SVM), Random Forest, 3DSlicer, and ITK-SNAP are
used here.

Figure 2 demonstrates how machine learning and feature extraction is used in medical
imaging, through an example.

The second method uses deep learning and does not require region annotation rather,
localization is usually sufficient. It comprises several layers where feature extraction, selection
and ultimate classification are performed simultaneously during training. As layers learn
increasingly higher-level features (Box 1), earlier layers might learn abstract shapes such as
lines and shadows, while other deeper layers might learn entire organs or objects. Algorithms
like Convolutional Neural Network (CNN) and labellmg are used here.

Figure 3 demonstrates how deep learning is used in medical imaging.

4
 figure 2: AI Detecting Lung Cancer from CT Scans using ML and feature extraction

figure 3: AI detecting diseases from medical imaging techniques using deep learning

5
 2.3 AI IN BLOOD TESTS AND GENETIC PROFILING

The integration of AI into blood test analysis presents a transformative potential for
healthcare, enabling faster, more accurate diagnosis, and allowing for real-time monitoring of
patients. Traditional blood analysis methods often rely on human interpretation, which can be
prone to error and subjectivity. AI algorithms, on the other hand, can process large datasets,
detect complex patterns, and provide precise diagnostics based on routine laboratory data. AI
aids in the automatic interpretation of lab results by locating deviations in cholesterol,
hemoglobin, glucose levels, and biomarkers. Machine learning models analyze the hidden
patterns in blood test data and flag potential health risks such as anemia, diabetes, etc. Deep
learning techniques, such as Convolutional Neural Networks (CNNs), are also used in image-
based blood smear analysis to detect conditions like leukemia and malaria. Figure 4
demonstrates how CNN is used in detecting diseases.

figure 4: Convolutional Neural Network in blood smear analysis.

In genetic profiling, AI plays a crucial role in analyzing DNA sequences to predict hereditary
diseases before symptoms appear. AI-driven tools scan genetic variations, such as single
nucleotide polymorphisms (SNPs) and mutations, to assess the risk of disorders like cancer,

6
Alzheimer’s, and rare genetic diseases. By processing genome sequencing data, AI can identify
mutations in genes like BRCA1 and BRCA2, which are linked to breast cancer, or APOE,
associated with Alzheimer’s. Advanced AI models, including DeepVariant by Google and
AlphaFold by DeepMind, enhance genetic analysis, providing precise insights into disease risk.
Machine learning algorithms further help classify DNA sequences, assisting in personalized
medicine by tailoring treatments based on an individual’s genetic makeup.

7
8