USA: +1(669)289-8683 
Sign in
- I need the help with the glossary
( the glossary from chapter 1 and 3 only )
- Also The references need rearrange.
- ( the references follow the APA stile Combine the references in chapter one with the ones that already in the references section )
College of Computing & Informatics (CCI)
SENIOR PROJECT-I REPORT
Author(s):
Project Supervisor:
1
Chronic Illness Prediction Model
Based on the Patient’s Medical
History.
Thesis/Project submitted to:
College of Computing & Informatics, Saudi Electronic University, Riyadh, Saudi Arabia.
In partial fulfillment of the requirements for the degree of:
BACHELOR OF SCIENCE IN INFORMATION TECHNOLOGY
Project Supervisor
2
Project Committee Chair
ABSTRACT
Include a 2-3 paragraph brief about the project, its utility and your contribution. Also highlight
how the project has helped in your professional growth. It should briefly explain the problem and
why you need to solve it, what method did you use and what is the advantages of your proposed
method with reflection on the findings or results.
DEDICATION
This work is dedicated to…
PREFACE
The preface content comes here. Make sure it is not more than one page. Preface should be used to
describe any special clarifications regarding project report. You may also include acknowledgment
paragraph in this section.
REVISION HISTORY
Name
Date
Reason For Changes
Version
5
TABLE OF CONTENTS
CHAPTER 1: INTRODUCTION ………………………………………………………………………………. 7
1.1
Project Background/Overview: ………………………………………………………………………………7
1.2
Problem Description: ……………………………………………………………………………………………7
1.3
Project Scope: ……………………………………………………………………………………………………..8
1.4
Project Objectives: ……………………………………………………………………………………………….9
1.5
Project Structure/Plan: …………………………………………………………………………………………9
CHAPTER 2: LITERATURE REVIEW …………………………………………………………………… 15
CHAPTER 3: METHODOLOGY ……………………………………………………………………………. 50
CHAPTER 4: SYSTEM ANALYSIS ………………………………………………………………………… 52
4.1
Product Features: ……………………………………………………………………………………………… 52
4.2
Functional Requirements: …………………………………………………………………………………… 52
4.3
Nonfunctional Requirements……………………………………………………………………………….. 54
4.4
Analysis Models ………………………………………………………………………………………………… 56
CHAPTER 5: SYSTEM DESIGN ……………………………………………………………………………. 57
CHAPTER 6: DISCUSSION & CONCLUSION ………………………………………………………… 60
6.1
Discussion ………………………………………………………………………………………………………… 60
6.2
Conclusion ……………………………………………………………………………………………………….. 60
REFERENCES ……………………………………………………………………………………………………. 61
APPENDIX: Glossary …………………………………………………………………………………………… 66
6
CHAPTER 1: INTRODUCTION
The introduction should set the context for the project and should provide the reader with a summary of the
key things to look out for in the remainder of the report. When detailing the contributions, it is helpful to
provide pointers to the section(s) of the report that provide the relevant technical details. The introduction
itself should be largely non-technical. It is useful to state the main objectives of the project as part of the
introduction. Should have the following headings:
1.1
Project Background/Overview:
Chronic illnesses are long-term health conditions that affect millions of people
worldwide [1]. However, not all chronic illness prediction models are fair and accurate for
both men and women [2]. Sex is a key factor that influences the risk and outcome of many
chronic diseases. Women have different biological, social, and environmental factors that
affect their health than men [3][4]. Moreover, women often face medical gaslighting, which
is the practice of dismissing or minimizing their symptoms and concerns by healthcare
providers [5][6][7]. This can lead to delayed diagnosis, misdiagnosis, or under treatment of
chronic diseases in women [8][9]. Therefore, it is important to consider sex differences
when developing and evaluating chronic illness prediction models. Our project aims to
create a machine learning model that can predict the risk of developing three chronic
conditions in women: heart diseases, autoimmune diseases, and chronic pain. Our project
will collect and analyze data from a large and diverse sample of women, use sex-specific
features and risk factors, and train and test our model using appropriate methods and
metrics. Our project will provide a reliable and useful tool for women’s health.
1.2
Problem Description:
Chronic illness prediction models are mathematical tools that use data and
algorithms to estimate the probability of developing a chronic condition in the future.
These models can help health care providers and patients to prevent, diagnose, and treat
chronic diseases. However, not all chronic illness prediction models are fair and accurate
1
Noncommunicable diseases (who.int)
Sex and gender differences and biases in artificial intelligence for biomedicine and healthcare | npj Digital
Medicine (nature.com)
3
Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment – PubMed (nih.gov)
4
Sex, gender, and pain: a review of recent clinical and experimental findings – PubMed (nih.gov)
5
Gender Disparity in Analgesic Treatment of Emergency Department Patients with Acute Abdominal Pain – Chen 2008 – Academic Emergency Medicine – Wiley Online Library
6
Obstetric gaslighting and the denial of mothers’ realities – ScienceDirect
7
Trust, Distrust, and ‘Medical Gaslighting’ | The Philosophical Quarterly | Oxford Academic (oup.com)
8
Back pain and the resolution of diagnostic uncertainty in illness narratives – PubMed (nih.gov)
9
Provider judgments of patients in pain: seeking symptom certainty – PubMed (nih.gov)
2
7
for both men and women. There are several sources of bias that can affect the
performance and validity of these models for different sexes [10].
One source of bias is the data quality and quantity. Many chronic illness prediction
models are based on data that are collected from male-dominated populations or clinical trials that
exclude or underrepresent women. This can lead to incomplete or inaccurate data that do not
reflect the biological, social, and environmental factors that influence women’s health. For
example, some prediction models for cardiovascular diseases use risk factors that are more
common or relevant for men, such as smoking,
Another source of bias is algorithm design and evaluation. Many chronic illness
prediction models are developed and tested using statistical methods that assume a
homogeneous or balanced distribution of data across sexes. This can lead to biased or
misleading results that favor the majority or dominant group, which is often men [11].
To overcome these biases and create a sex-based chronic illness prediction model
for women, we intend to follow several steps in our project. First, we will collect and
analyze medical records data from a large and diverse sample of women who have been
diagnosed with or are at risk of developing three chronic conditions: heart diseases,
autoimmune diseases, and chronic pain. We will ensure that the data are complete,
accurate, and representative of the population of interest. Second, we will use machine
learning techniques to identify the most relevant features and risk factors for predicting
the incidence of these conditions in women. We will use sex-specific variables and
interactions that capture the biological and medical influences on women’s health. Third,
we will train and test our prediction model using appropriate methods that account for
variance of data across sexes. We will use metrics that measure the accuracy, fairness,
and robustness of our model for our target demographic – women. We will also compare
our model with existing chronic illness prediction models to evaluate its performance and
validity.
1.3
Project Scope:
The aim of the project is to develop a machine learning model that can predict the
risk of chronic conditions such as heart diseases, autoimmune diseases, and chronic pain
in women. The model will use the following data sources and methods:
–
Data sources: The model will use medical records of 35–65-year-old people of
female patients. The medical records will include information such as demographics,
medical history, medications, and diagnoses.
–
Methods: The model will use a supervised learning approach to train and test the
model on the data. The model will use a classification algorithm to predict the
probability of having a chronic disease based on the input features. The model will
10
Sex and gender differences and biases in artificial intelligence for biomedicine and healthcare | npj Digital
Medicine (nature.com)
11
Prevalence of sexual dimorphism in mammalian phenotypic traits | Nature Communications
8
use cross-validation and performance metrics such as accuracy, precision, recall,
and F1-score to evaluate the model’s performance and generalizability.
–
1.4
Benefits: The project will benefit healthcare providers by providing them with a tool
that can help them prevent or manage chronic diseases in their female patients. The
project will also benefit patients by providing them with personalized, preventative,
and unbiased healthcare services that can improve their quality of life and reduce
their healthcare costs.
Project Objectives:
The objective of this project is to develop and evaluate a chronic diseases
prediction model for women using machine learning techniques and medical records data.
The model will aim to predict the risk of developing three chronic conditions: heart
diseases, autoimmune diseases, and chronic pain, based on women’s records. The model
will be validated using a test dataset of women’s medical records. The objective of this
project is to provide a useful tool for women and health care providers to improve the
prevention, diagnosis, and treatment of chronic diseases in women.
1.5
Project Structure/Plan:
1.5.1
Textual plan:
Project management is one of the most important stages of work in the project, if
not its foundation. It works on setting the beginning and end lines. It is the process of
organizing, planning, and implementing tasks, resources, and timetables to ensure the
achievement of the project objectives.
Planning is one of the first stages of project management. It determines the
general form of the project and sets its basics, which includes defining the idea of the
project and the participating team, in addition to the project goals, objectives, scope, and
strategies to achieve them. This is done by providing a plan for how to implement,
monitor and control the project. The following are the main activities in the planning
stage:
•
Choosing work team: which involves choosing the team and submitting the names of the
team members in the form. It starts on 4/9 and ends on 4/9, with a duration of 1 days. It
requires no resources.
•
Determining the initial idea, which involves deciding on the initial idea by using nominal
group technique. It starts on 5/9 and ends on 7/9, with a duration of 3 days. It requires
Researching the new technologies in the field, team’s input and brainstorming sessions as
resources.
•
Writing the introduction to our project, which involves writing the background/overview
and problem description, determining the scope, and setting the objective. It starts on 7/9
9
and ends on 13/9, with a duration of 7 days. It requires scholar literatures, Journals, and
research as resources.
•
Writing a work plan, which involves writing the introduction, literature review,
methodology, data analysis, results, discussion, and conclusion sections of the article. It
starts on 14/9 and ends on 20/9, with a duration of 7 days. It requires team’s input, SEU
academic calendar and quick research about the timeline of similar projects.
•
Literature review, which involves discussing and describing 25, or more, literature
resources relating to the project being carried out. It starts on 24/9 and ends on 27/9, with
a duration of 4 days. It requires literature searches, scholar literatures, journals, and
research as resources.
•
Choosing the methodology, which involves choosing the learning approach to be used to
train and test the model on the data, choosing the algorithm to be used to predict the
probability of having a chronic disease based on the input features, and choosing the
performance metrics to be used to evaluate the model’s performance and generalizability.
The analysis stage is the second stage of project planning (in SDLC) is the stage
where the product team conducts a thorough examination of the business requirements
and the system specifications for the software project. This stage will It starts on 22/10
and ends on 1/11, with a duration of 11 days. It involves the following activities:
•
Product Features, which Listing and explaining major product features to be developed in
the section.
•
Functional Requirements, which involves Specifying the expected behavior of the model
under different conditions and scenarios, as well as the constraints and assumptions that
the model should follow.
•
Nonfunctional requirements, which involves specifying the criteria and standards that the
model should meet or exceed in terms of its quality, reliability, usability, security, and
maintainability. The three earlier activities require researching the features of similar
technologies, group input, brainstorming sessions, and supervisor’s feedback as
resources.
•
Making analysis model, which involves building models for demonstration purposes or
development project. It requires use case templates, use case diagrams, and use case
tools. This stage will start on 26/10 and ends on 1/11, with a duration of 7 days.
The design phase in the SDLC is the third stage where we will be making the
following design model: component diagram, deployment diagram, design level sequence
diagram, complete class diagram, entity-relationship diagram. This stage will start on
2/11 and ends on 15/11, with a duration of 14 days.
The discussion and conclusion stage is where we will be revising the previous
chapters and improving on them, describing the crux and importance of the project, its
10
novelty and its current applications, and suggesting future work that can be derived out of
this project. This stage will start on 2/11 and end on 5/12 with a duration of 11 days.
1.5.2
Activity
Planning
Choose work
team
Tabular plan:
Description
Creating any project must
begin with planning, which
includes a group of tasks,
which are as follows:
The teamwork’s names
were submitted in form
Start
Finish
Duration
Sun 3/9
Week3
Wed
4/10
Week7
31 days
4/9
4/9
1d
Determine the
initial idea
Deciding on the initial idea
using the nominal group
technique.
5/9
7/9
3d
Writing the
introduction
to our project
Writing the
background/overview and
problem description.
Determining the scope
and setting the objective.
7/9
13/9
7d
Write work
plan and initial
schedule
Writing a work plan
involves outlining the
specific tasks, timelines,
and resources required to
achieve a project’s
objectives.
14/9
20/9
literature
review
Discussing and describing
25, or more, literature
resources relating to the
project being carried out.
methodology
Choosing the learning
approach to be used to
train and test the model
on the data.
Choosing the algorithm to
be used to predict the
probability of having a
7d
Resources
Requirements
No resources
needed
– Researching the
new
technologies in
the field
– Group meeting
– Brain storming
sessions.
– Scholar
literatures.
– Journals.
– Researches.
– Reviewing
similar projects.
– Team meeting.
– SEU academic
calendar.
– Quick research
about the
timeline of
similar projects.
–
24/9
27/9
4d
–
28/9
4/10
7d
Scholar
literatures.
Journals.
Researches.
articles, blogs,
tutorials, and
books that
explain different
machine
learning
methodologies
11
chronic disease based on
the input features.
Choosing the performance
metrics to be used to
evaluate the model’s
performance and
generalizability.
Analysis
Product
Features
functional
Requirements
Nonfunctional
Requirements
The deliverables of this
activity represent the
documents of the project
requirements.
To ensure obtaining these
documents, these tasks
must be carried out
Listing and explaining
major product features to
be developed in the
section.
Specifying the expected
behavior of the model
under different conditions
and scenarios, as well as
the constraints and
assumptions that the
model should follow.
specifying the criteria and
standards that the model
should meet or exceed in
terms of its quality,
reliability, usability,
security, and
maintainability.
–
–
Sun
Wed
22/10
Week
10
1/11
Week
11
11 days
–
22/10
25/10
4d
–
Analysis
Models
Design
Building models for
demonstration purposes
or development project.
Making the following
design model: component
diagram, deployment
diagram, design level
26/10
Sun
2/11
Week
12
1/11
Wed
15/11
Week
13
researching
different
machine
learning
algorithms.
Researching
machine
learning libraries
and frameworks.
7d
–
14 days
–
Researching the
features of
similar
technologies in
Group meeting
Brain storming
sessions.
Supervisor’s
feedback
Use case
templates
Use case
diagrams
Use case tools
Design model
templates
Design model
diagrams
12
Discussion &
Conclusion
1.5.3
sequence diagram,
complete class diagram,
entity-relationship
diagram
Revising the previous
chapters and improving on
them.
Describe he crux and
importance of the project,
its novelty and its current
applications.
suggesting future work
that can be derived out of
this project
Sun
26/11
Week
14
Tue
5/12
Week
15
11 days
–
Design model
tools
–
Our on work
Supervisor’s
feedback
AMD diagram:
13
1.5.4
Gantt Chart:
14
CHAPTER 2: LITERATURE REVIEW
Machine learning algorithms in early detection of diabetes.
Machine learning algorithms have arisen as a powerful tool in the early detection and prediction
of diabetes, capitalizing on their capacity for data analytics and pattern recognition. According to
(Jeyamurugan et al., 2023), their studies employed various classifiers such as Naïve Bayes, support vector
machine (SVM), logistic regression, and decision trees, comparing their performance in diabetes
prediction. Their investigations meticulously curated extensive diabetes datasets, undertaking rigorous
preprocessing to rectify null values, and subsequently trained machine learning models. A hybrid
ensemble model, fusing artificial neural networks (ANN) with logistic regression, emerged as the most
accurate predictor. Additionally, their studies revealed that the application of deep neural networks
(DNNs) demonstrated remarkable success in distinguishing between type 1 diabetes (T1D) and type 2
diabetes (T2D). Integrating logistic regression and decision tree algorithms into a hybrid neural network
model (Jeyamurugan et al., 2023) achieved high accuracy in predicting diabetes. Incorporating machine
learning algorithms presents several advantages, including their proficiency in handling vast and complex
datasets, adeptness at identifying pertinent features for prediction, and adaptability to evolving patterns
and trends. By analyzing diverse patient attributes like glucose levels, blood pressure, body mass index,
and family history, these algorithms generate precise predictions, enabling early intervention and
personalized management of diabetes. Hence, the integration of machine learning algorithms into diabetes
prediction stands as a pivotal advancement in healthcare, offering the potential for improved outcomes
through early detection, tailored treatment plans, and preventive measures. These studies underscore
diverse approaches, from ensemble models to deep neural networks, all showing promise in achieving
precise diabetes predictions. As this field continues to evolve, further research and innovations in machine
learning applications hold the potential to revolutionize diabetes care and significantly impact public
health.
Advanced machine learning techniques for cardiovascular disease.
Machine learning (ML) holds promise in revolutionizing cardiovascular disease (CVD) detection.
Techniques like Support Vector Machines, Decision Trees, Random Forests, and Gradient Boosting
Machines have shown effectiveness. (Baghdadi et al., 2023), Their studies showcased deep learning and
transfer learning’s potential. One of their studies achieved over 90% accuracy in predicting heart attack
risk in patients with no CVD history. Another achieved over 95% accuracy in diagnosing coronary artery
disease in patients with chest pain. However, the challenges include acquiring high-quality patient data
and ensuring the interpretability of ML algorithms for clinical use. Future research should focus on
robust, interpretable ML techniques to enhance CVD early detection and diagnosis.
15
Machine learning in the prediction, diagnosis, and management of diabetes.
In their comprehensive review, (Afsaneh et al., 2022) delve into recent advancements in machine
learning (ML) and deep learning (DL) models for diabetes prediction, diagnosis, and management. The
authors present compelling evidence of the potential of these models across various facets of diabetes
care. ML and DL models demonstrate promise in predicting diabetes risk, particularly in high-risk
individuals with prediabetes or familial predisposition. A recent study achieved over 90% accuracy in
forecasting type 2 diabetes risk from a dataset of 100,000 patients. These models also exhibit proficiency
in diagnosing diabetes in symptomatic individuals, with a striking example being the accurate diagnosis
of diabetic retinopathy, a severe complication that can lead to blindness. The DL model achieved an
impressive accuracy rate surpassing 95%, drawing from a dataset of over 50,000 patients. Moreover, ML
and DL models hold potential in personalizing treatment plans, as evidenced by a study reducing average
blood glucose levels by 10% in patients with type 1 diabetes from a dataset of over 10,000 individuals. At
the same time, promising, challenges persist, including the need for extensive, high-quality patient
datasets and the development of interpretable algorithms for clinician comprehension. Future research
endeavors should prioritize the creation of strong, reliable, and interpretable models to revolutionize
diabetes prediction, diagnosis, and management. (Afsaneh et al. 2023) The review underscores the
transformative potential of ML and DL models in diabetes care, paving the way for personalized,
effective, and accurate approaches to prediction, diagnosis, and management.
Diabetes risk prediction using machine learning approaches.
Their comprehensive survey, (Firdous et al., 2022) spotlighted the pivotal role of machine
learning in predicting diabetes risk. They emphasize the criticality of early detection and prevention of
diabetes, a chronic ailment fraught with severe complications. The authors delineate two overarching
categories of machine learning approaches: supervised and unsupervised learning. Supervised algorithms
glean insights from labeled patient datasets, discerning features most indicative of diabetes. Conversely,
unsupervised algorithms analyze unlabeled patient data, discerning patterns that can cluster individuals
into risk-associated groups. Numerous studies demonstrate machine learning algorithms’ high accuracy in
prognosticating diabetes risk. However, challenges loom, notably the imperative for expansive, highquality patient data and the development of interpretable algorithms for clinical comprehension. The
authors advocate for future research to mitigate these challenges and craft robust, reliable, and
interpretable algorithms. With continued refinement and development, machine learning holds substantial
promise in revolutionizing early diabetes detection and prevention. Specific algorithms like logistic
regression, support vector machines, decision trees, random forests, gradient boosting machines, and
neural networks have demonstrated remarkable accuracy, even with relatively modest datasets.
Leveraging machine learning, clinicians can effectively identify high-risk individuals, enabling targeted
preventive interventions for a more
proactive approach to diabetes management.
16
Real-Time Diabetes Prediction with CNN-LSTM
In recent years, the application of machine learning in diabetes prediction has garnered significant
attention. Various algorithms have been employed, including logistic regression, support vector machines,
decision trees, and random forests. Deep learning algorithms have emerged as a powerful tool for diabetes
prediction, given their ability to discern intricate patterns within data. This has led to the utilization of
deep learning techniques such as convolutional neural networks (CNNs), recurrent neural networks
(RNNs), and long short-term memory (LSTM) networks. CNNs excel in extracting features from image
data, while RNNs and LSTM networks specialize in recognizing sequential patterns. (Madan et al., 2022)
their article introduces an innovative diabetes prediction model amalgamating a CNN with a Bi-LSTM
network. This novel approach surpasses existing state-of-the-art accuracy, sensitivity, and specificity
models. The authors also contemplate the potential application of this model in real- time settings,
offering promising prospects for creating tools geared toward early diabetes detection and prevention. The
integration of advanced deep learning techniques showcases a notable leap forward in the domain of
diabetes prediction.
ML Evaluation for Predicting Cardiovascular Disease
(Asif et al., 2021) their study extensively assesses twelve distinct machine learning algorithms for
their efficacy in predicting cardiovascular disease (CVD). By taking advantage of a dataset of over
300,000 patients, including demographic, medical, and laboratory data, the algorithms were trained to
discern CVD presence. Performance metrics encompassed accuracy, precision, recall, specificity, and F1
score. The ensemble classifiers (EVCH and EVCS) emerged as the most proficient, achieving over 92%
accuracy. Other algorithms also demonstrated commendable performance, ranging between 80% to 90%
accuracy. The study aligns with prior research, underscoring ensemble classifiers as the prime choice for
CVD prediction. However, it is vital to acknowledge that the evaluation was confined to a specific
dataset, and algorithm performance may vary in diverse settings. Thus, a broader assessment across
various datasets is imperative before clinical implementation. (Asif et al.,2021) research constitutes a
valuable contribution to machine learning in CVD prediction, emphasizing ensemble classifiers’
superiority, which necessitates further evaluation in diverse datasets.
Development of a predictive model for integrated medical and long-term care resource
consumption based on health behaviour: application of healthcare big data of patients
with circulatory diseases.
The study, conducted by Tomoyuki Takura, Keiko Hirano Goto, and Asao Honda, is a research
article published in BMC Medicine in 2021. The study used healthcare big data of patients with
circulatory diseases to develop the Adherence Score for Healthcare Resource Outcome (ASHRO) model.
The authors employed machine learning techniques with random forests and K-fold cross-validation to
select and integrate explanatory variables and to set weights. The authors also used logistic regression
analysis to examine the association between the ASHRO score and clinical outcomes (Takura, Hirano
Goto, & Honda, 2021, p.3).
17
The study population consisted of 48,456 individuals, mean age 68.3 years, who underwent specific
health check-ups in Japan between 2011 and 2012. Of these individuals, 38% were women and 62% were
men. The study population had a high prevalence of hypertension, dyslipidemia, and diabetes, which are
common chronic conditions in Japan (Takura, Hirano Goto, & Honda, 2021, p.7).
The results of the study showed that the ASHRO score was significantly associated with clinical
outcomes, including hospitalization, emergency department visits, and death. The authors also found that
the ASHRO score was a better predictor of clinical outcomes than traditional risk factors, such as age,
sex, and comorbidities. The authors concluded that the ASHRO score could be a useful tool for predicting
healthcare resource consumption and improving patient outcomes (Takura, Hirano Goto, & Honda, 2021,
p.3).
However, the study has some limitations. The authors acknowledged that the study was limited to patients
with circulatory diseases and that the generalizability of the findings to other patient populations is
unclear. The authors also noted that the study did not perform highly reliable external validation, which is
a limitation of the research. Additionally, the authors acknowledged that the cross-validation technique
used in the study may lead to overfitting, and further research is needed to improve internal validation
methods (Takura, Hirano Goto, & Honda, 2021, p.13).
Prediction of prognosis in elderly patients with sepsis based on machine learning
(random survival forest)
The study was conducted by Luming Zhang, Tao Huang, Fengshuo Xu, Shaojin Li, Shuai Zheng,
Jun Lyu, and Haiyan Yin and published in BMC Emergency Medicine. The authors aimed to develop a
machine learning model that could predict the prognosis of elderly patients with sepsis. The study used
data from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, which is a large
public database of electronic health records from patients admitted to intensive care units (Zhang et al.,
2022, p.1).
A total of 6,503 patients were enrolled in this study. The study population was evenly split between men
and women, and the mean age was 77.00 years (Zhang et al., 2022, p.3). The authors used a random
survival forest model to predict the survival time of patients with sepsis. The model was trained on a
subset of the MIMIC-IV database and then tested on a separate validation set (Zhang et al., 2022, p.2).
The authors evaluated the performance of the model using several metrics, including the concordance
index (C-index) (Zhang et al., 2022, p.1).
The random survival forest model achieved a higher C-index on the validation set than the other models,
showing that it had good predictive performance for elderly patients with sepsis (Zhang et al., 2022, p.3).
Overall, this study provides evidence that machine learning models can be used to predict the prognosis
of elderly patients with sepsis. The authors suggest that these models could be used to identify high-risk
patients and guide clinical decision-making. However, the study lacks external validation (Zhang et al.,
2022, p.8).
Machine learning based efficient prediction of positive cases of waterborne diseases.
18
The study, conducted by Mushtaq Hussain, Mehmet Akif Cifci, Tayyaba Sehar, Said Nabi, Omar
Cheikhrouhou, Hasaan Maqsood, Muhammad Ibrahim, and Fida Mohammad is a research published in
BMC Medicine in 2023. The study aims to develop a warning system or dashboard based on ML that
enables the health department to obtain useful information from patient records, such as ranking the
different types of waterborne diseases in various regions of Pakistan and then making predictions based
on patient records (Hussain et al, 2023, p.2).
The study collected typhoid and malaria patient data for the years 2017 – 2020 from Ayub Medical
Hospital in Abbottabad, Pakistan (Hussain et al, 2023, p.5). Different ML models were trained and tested
on the dataset using the tenfold cross-validation method (Hussain et al, 2023, p.3). The study explored
several research questions, including identifying the ML model with the greatest performance predicting
positive cases based on patient history, discovering the input includes patient symptoms most relevant for
patients with waterborne diseases, and investigating the number of positive cases of patients in the
hospital database using ML models (Hussain et al, 2023, p.2).
The study employed actual patient data from Ayub Medical Hospital in Abbottabad, Pakistan, to forecast
positive cases of waterborne disease using explainable ML techniques. The study used common ML
algorithms such as Decision Trees, Random Forest, Support Vector Machine, Logistic Regression, and KNearest Neighbor to predict the waterborne disease-positive cases. These algorithms are easily
explainable, interpretable, implemented, and used in many fields with good performance (Hussain et al,
2023, p.9).
The study’s findings suggest that ML techniques can be used to predict positive cases of waterborne
diseases with high accuracy. The study also identified the most relevant patient symptoms for patients
with waterborne diseases, which can help in early detection and prevention of outbreaks. The study’s
contributions to our understanding of waterborne diseases include the development of a warning system
or dashboard based on ML that enables the health department to obtain useful information from patient
records and the use of actual patient data to forecast positive cases of waterborne disease using
explainable ML techniques.
Dense phenotyping from electronic health records enables machine learning-based
prediction of preterm birth.
The study, conducted by Abin Abraham, Brian Le, Idit Kosti, Peter Straub, Digna R. VelezEdwards, Lea K. Davis, J. M. Newton, Louis J. Muglia, Antonis Rokas, Cosmin A. Bejan, Marina Sirota
& John A. Capra is a research published in BMC Medicine in 2022. The study aims to appliy machine
learning to diverse data from EHRs to predict singleton preterm birth. predicting preterm birth risk.
The study applied machine learning algorithms on 35,282 records of deliveries. They extracted sets of
features from EHRs, including demographic variables (age, race), clinical keywords from obstetric notes,
clinical lab tests ran during the pregnancy, and predicted genetic risk (polygenic risk score for preterm
birth). They trained machine learning models to predict preterm birth at various gestational ages using
19
different subsets of features and compared their performance to existing risk factors. (Abraham et al,
2022, p.2)
The study found that machine learning models based on billing codes alone can predict preterm birth risk
at various gestational ages and outperform comparable models trained using known risk factors. The
models stratify deliveries into interpretable groups based on the presence or absence of specific billing
codes, such as hypertension, diabetes, and infections. The models also identify novel risk factors, such as
certain medications and laboratory results, that are not currently used in clinical practice (Abraham et al,
2022, p.4).
The study provides a proof of concept that machine learning algorithms can use the dense phenotype
information collected during pregnancy in EHRs to predict preterm birth. The prediction accuracy across
clinical contexts and compared to existing risk factors suggests such modeling strategies can be clinically
useful. However, the lower performance on Hispanic and Asian women highlights the need to ensure that
future approaches do not introduce or amplify biases against specific groups or types of preterm birth
(Abraham et al, 2022, p.11).
A Predictive and Preventive Model for Onset of Alzheimer’s Disease
The study, conducted by Udit Singhania, Balakrushna Tripathy, Mohammad Kamrul Hasan, Noble C.
Anumbe, Dabiah Alboaneen, Fatima Rayan Awad Ahmed, Thowiba E. Ahmed Manasik M. Mohamed
Nour, is research published in Frontiers in Public Health in 2021. The authors introduce their predictive
and preventive model for Alzheimer’s disease, which is based on machine learning algorithms and
incorporates a range of demographic, clinical, and genetic factors.
The authors used the OASIS Cross-Sectional and Longitudinal datasets, which contain demographic,
clinical, and genetic data from individuals with and without Alzheimer’s disease. They preprocessed the
data by removing missing values and outliers, and performed feature selection to identify the most
predictive factors for Alzheimer’s disease.
The authors then used machine learning algorithms, including logistic regression, decision tree, and
random forest, to develop their predictive model. They evaluated the performance of the model using
metrics such as accuracy, sensitivity, specificity, and area under the curve (AUC). The results of the study
showed that the predictive model achieved high accuracy in predicting the onset of Alzheimer’s disease,
with an AUC of 0.904 (Singhania et al., 2021, p.6).
The authors also developed a preventive model, which used machine learning algorithms to identify
individuals at high risk of developing Alzheimer’s disease. The preventive model achieved high accuracy
in identifying individuals at risk, with an AUC of 0.92 (Singhania et al., 2021, p.8).
The authors conclude by discussing the potential implications of their research for the future of
Alzheimer’s disease diagnosis and treatment. They suggest that their model could be used to identify
individuals at high risk of developing Alzheimer’s disease, allowing for earlier interventions and
preventive measures. They also highlight the need for further research to validate and refine their model,
and to explore the potential of machine learning techniques for other areas of healthcare.
20
An optimal multi-disease prediction framework using hybrid machine learning
techniques.
The study aims to develop an optimal multi-disease prediction framework using hybrid machine
learning techniques, specifically for predicting lifestyle diseases like heart disease and diabetes. The
methodology includes genetic algorithm-based recursive feature elimination (GA-RFE) and AdaBoost,
which are used to identify the optimal subset of features for multi-disease prediction. The study compares
the proposed approach with benchmark machine learning techniques using k-fold cross-validation. Key
contributions include a study of lifestyle diseases prevalent in the US, a machine learning-based
predictive model for early heart disease and diabetes detection, data preprocessing techniques, and the
proposed modified feature selection method, GA-RFE. AdaBoost and other predictive models are trained
and evaluated for multi-disease prediction, and a comparative analysis is conducted to assess the
effectiveness of the proposed model. The study is organized into sections, including an introduction,
related works, preliminaries, the proposed methodology, experimental setup, results and discussions, and
a summary. The goal is to contribute to healthcare informatics research by developing an efficient and
accurate framework for predicting multiple lifestyle diseases.
An Effective Machine Learning-Based Model for an Early Heart Disease Prediction.
This research article presents an effective machine learning-based model for early heart disease
prediction. Heart disease (HD) is a significant global health issue, requiring accurate diagnosis and early
prevention. The authors propose a machine learning-based prediction model (MLbPM) that uses data
scaling methods, split ratios, best parameters, and machine learning algorithms. The model’s performance
is tested on a University of California Irvine HD dataset. The results show a 96.7% accuracy when
logistic regression, robust scaler, best parameter, and 70:30 split ratio is considered. MLbPM outperforms
other compared works in terms of accuracy. The study emphasizes the need for accurate prediction and
diagnosis in the early detection and treatment of HD. Heart disease (HD) is a leading cause of death
worldwide, with nearly 17.9 million deaths occurring annually. Early detection and classification of HD
can prevent deaths, especially in rural areas where doctors and equipment are limited. Machine learning
(ML) algorithms can be used to analyze data and recognize patterns with minimal human intervention.
This work proposes an ML-based prediction model (MLbPM) that uses the best data scaling method, split
ratio, and parameter to predict HD. The model combines four ML algorithms and three data scaling
methods, with the default parameters and split ratios to find the best match for HD prediction. The main
contributions of this work include proposing a model called MLbPM, improving HD prediction accuracy,
reducing false predictions, and identifying an appropriate prediction algorithm among the four ML
algorithms to accurately classify the given University of California Irvine (UCI) HD dataset. The model
performance is validated using metrics like accuracy, F1score, precision, recall, and a receiver operating
characteristic (ROC) curve. Related works on HD prediction have shown different performance outcomes
due to different ML algorithms. For example, Tu et al. obtained an accuracy of 81.14% using bagging and
78.9% using decision tree (DT), Srinivas et al. achieved an accuracy of 84.14% using a naive approach,
and Chaurasia and Pal achieved an accuracy of 83.49% using classification and regression trees (CART)
21
and DT. Future work could focus on improving HD prediction accuracy and enhancing the effectiveness
of ML algorithms in HD prediction.
A Comparative Analysis for Diabetic Prediction Based on Machine Learning Techniques.
A comparative analysis for diabetic prediction using machine learning techniques is proposed in
the Journal of Basrah Research. The study aims to improve the accuracy, accuracy, and performance of
machine learning algorithms for predicting diabetes mellitus at an early stage. The proposed work uses
five common machine learning algorithms: K-Nearest Neighbors (KNN), Naïve Bayes (NB), Logistic
Regression (LR), Random Forest (RF), and Support Vector Machine (SVM) for obtaining early
prediction, high accuracy, and performance compared to related works. The experimental results indicate
that SVM gets the highest accuracy (83%) on diabetic prediction based on the Pima Indian diabetes
dataset (PIDD). Diabetes is a common chronic disease that occurs when blood sugar levels are
abnormally high, leading to complications such as heart disease, nerve damage, kidney failure, and stroke.
Long-term diabetes can cause macrovascular and microvascular complications, which are the most
frequent health concerns. A healthy lifestyle, including a healthy workout routine and a well-balanced
diet, can help prevent or control diabetes. The National Diabetes Prevention Program, coordinated by the
Centers for Disease Control and Prevention, is a healthier lifestyle program that can help prevent Type 2
diabetes.
The healthcare industry collects vast amounts of medical data, including patient records, hospital
information, and lab results. Early diagnosis of diseases can be imprecise and vulnerable, leading to
inaccurate decisions and potentially preventing patients from receiving appropriate treatment. Machine
learning and data mining have become crucial tools in the healthcare domain, supporting doctors in
making accurate diagnoses and preventing medical errors. This paper presents a practical study analysing
the behaviour of various classification algorithms when implemented to a Pima Indian Diabetes Dataset
(PIDD) data set. The main benefit is to protect patients’ lives by providing a tool that helps doctors make
accurate decisions and better diagnose for early detection of diabetes. The results recommend the use of
SVM for early diabetes diagnosis based on accuracy (83%), precision (79%), and recall (70%). The paper
also discusses the methodology of the proposed system, performance evaluation, and experimental
results.
Recent applications of machine learning and deep learning models in the prediction,
diagnosis, and management of diabetes: a comprehensive review.
Diabetes is a chronic metabolic illness characterized by increased blood glucose levels, which can
lead to critical detriment to other organs. Predicting, diagnosing, and managing diabetes is essential to
prevent harmful effects and recommend more effective treatments. Machine learning algorithms have
been developed for this purpose, and this review surveys recently proposed machine learning (ML) and
deep learning (DL) models for controlling blood glucose and diabetes. The reported results reveal that
22
ML and DL algorithms are promising approaches for controlling blood glucose and diabetes, but they
should be improved and employed in large datasets to confirm their applicability.
T1DM is a chronic autoimmune disease caused by elevated blood glucose levels, causing insulin
deficiency. It is associated with the attendance of autoantibodies several years before the start of
symptoms, as they can be considered a biomarker of autoimmunity. Autoantibodies targeting insulin, zinc
transporter 8 (ZNT8)6–8, or insulinoma-associated protein 2 (IA) target insulin, zinc transporter 8
(ZNT8)6–8, or insulinoma-associated protein 2 (IA). T1DM is influenced by environmental agents,
genetic risk factors, and environmental factors such as the timing of food introduction, viral infections,
and gestational infections.
In conclusion, machine learning and deep learning models have shown promise in controlling blood
glucose and diabetes, but they need to be improved and employed in large datasets to ensure their
applicability. T2DM is the most common type of diabetes, resulting in over 90% of cases. It is caused by
impaired insulin secretion by pancreatic β-cells due to insulin resistance in adipose tissue, liver, skeletal
muscle, and liver. Prediabetes occurs before hyperglycaemia, a high-risk situation that predisposes
individuals to T2DM. Prediabetes can be determined by elevated glycated haemoglobin A1c (HbA1c)
levels, impaired fasting glucose (IFG) levels, and impaired glucose tolerance (IGT). T2DM is heritable,
with higher risk among siblings of a T2DM patient and higher risk when the mother has the disease.
Genetic studies have identified single-nucleotide polymorphisms in T2DM cases, but these only elevate
the risk by 10-20%. T2DM control is complex due to various pathophysiological disorders and ‘ABCDE’
conditions.
An Intelligent Approach for Accurate Prediction of Chronic Diseases.
Chronic diseases pose a significant threat to healthcare communities worldwide, causing millions
of deaths. Early prediction of these diseases can help protect patients’ lives. This study uses a hybrid
optimization algorithm called the Hybrid Gravitational Search algorithm and Particle Swarm
Optimization algorithm (HGSAPSO) to improve the detection of chronic diseases. The algorithm
optimizes the parameters of six classifiers, including Artificial Neural Network (ANN), Support Vector
Machines (SVM), K-Nearest Neighbor (Knn), and Decision Tree (DT). The results show that the
proposed HGSAPSO algorithm achieves better accuracy on the ANN-HGSAPSO classifier compared to
other classifiers. The study evaluates the model on five benchmark datasets, showing that the hybrid
optimization algorithm performs finer than individual or ensemble classifiers in classifying diseases.
Machine learning techniques like Artificial Neural Network, Decision Tree, Support Vector Machine,
Naïve Bayes, and k-nearest Neighbor are used to detect various diseases. Deep learning models have been
used to predict diabetes, while Recurrent Neural Networks and Long-Term Short Memory have been used
to analyze and predict multiple diseases. Alzubi et al. used a boosted weighted optimized neural network
ensemble classification algorithm for improved classification performance.
23
Emerging misunderstood presentations of cardiovascular disease in young women
The document is a review that discusses the misunderstood presentations of cardiovascular
disease in young women (under 55). It emphasizes the high mortality rate of cardiovascular Artery
disease CVD in young women and the importance of raising awareness and identifying at-risk patients.
The review specifically focuses on certain categories of myocardial infarction and myocardial ischemia
with non-obstructive coronary arteries.
And According to the (Bullock‐Palmer et al.,2019), the main factors that contribute to the high mortality
rate of cardiovascular artery disease (CVD) in young women are:
a. Lack of awareness: There is a significant void in awareness of CVD in women, both among the
public and the medical profession. Many women are not aware that CVD is the leading cause of
death. Additionally, a survey showed that only a small percentage of primary care providers and
cardiologists feel well prepared to assess the CVD risk in their female patients.
b. Misunderstood presentations: There are emerging misunderstood presentations of CVD in young
women, such as myocardial infarction (MI) with non-obstructive coronary arteries (MINOCA)
and spontaneous coronary artery dissection (SCAD). These conditions may be misdiagnosed or
missed, leading to mismanagement and poorer clinical outcomes.
c. Inadequate risk assessment tools: Many traditional CVD risk estimate tools fail to identify the “at
risk” female, especially in young women. There needs to be a shift in focus from looking for
vulnerable plaques to identifying the “at risk” patient.
d. Higher rehospitalization rate: Even after treatment of their initial cardiac event, the
rehospitalization rate for women with CVD, particularly after acute (myocardial infarction) MI,
which is commonly known as a heart attack. is higher than that in men.
Early Prediction of Coronary Artery Disease (CAD) by Machine Learning Method – A
Comparative Study
This study presents a comparative on the early prediction of coronary artery disease (CAD) using
machine learning methods. Where (Zong Chen & P, 2021) explored various techniques and evaluate their
effectiveness in predicting CAD.
discusses several machine learning methods used in the study, including regression type of classification,
Naïve Bayes classifier, decision tree method, KNN method, neural network methodology, and genetic
algorithm.
The authors evaluate the effectiveness of the different techniques by comparing the existing classification
methods to predict CAD earlier for a higher accurate value. They also use a noisy type of database in their
research article for better clarity about identifying the classifier. where (Zong Chen & P, 2021) discuss the
performance of the proposed algorithm and compare it with other methods. They concluded Naïve Bayes
classifier is the suitable
24
Naïve Bayes method classifier is a suitable method to predict heart disease with a minimum number of
the dataset. It is a simple and efficient performance feature classification method based on prior
probability in the class attribute. This is a statistical regression analysis method to predict the diagnosis
with an approximation of dependent attributes.
Heart disease Prediction system (HDPS)
The research discusses the development of a Heart Disease Prediction System (HDPS) using
machine learning algorithms (Naive Bayes algorithm is). The system uses the medical profile of a patient
to predict the likelihood of them getting a heart disease. The system performs binary classification to
determine the presence of a likelihood of disease and then multiclass classification to classify the disease
among different stages. The system is implemented in Python and can be used as a training tool for nurses
and medical students.
The Heart Disease Prediction System (HDPS) can be used as a training tool for nurses and medical
students. By inputting the medical profile of a patient, the system can predict the likelihood of the patient
getting a heart disease. The system performs binary classification to determine the presence of a
likelihood of disease and then multiclass classification to classify the disease among different stages. The
system can recommend and prescribe a suitable medication to the patient and immediately provide the
necessary information. This allows nurses and medical students to learn how to diagnose patients with
heart disease and make informed decisions based on the predictions and recommendations provided by
the system.
The (Jayasree et al.,2019) determined the advantage of using the Naive Bayes algorithm is that it is easy
to build and highly scalable. It can handle large datasets and can be trained quickly compared to other
classification methods. Additionally, where (Jayasree et al., 2019) assumes Naive Bayes feature
independence, which simplifies the modeling process and allows for efficient computation of
probabilities. It is known to be one of the highly sophisticated classification methods and has been shown
to have better accuracy rates compared to other algorithms, such as Bayesian filters.
Sex Inequalities in Medical Research: A Systematic Scoping Review of the Literature
The document discusses the gender gap and misogyny in medical research. It highlights the
historical exclusion of female participants in medical studies and the subsequent generalization of
research data from males to females. This gender gap in medical research leads to real-life disadvantages
for female patients, such as longer waiting times for diagnosis and treatment, misdiagnosis, and
inadequate management.
Based on the (Merone et al., 2022) review, there are two fundamental issues in the field of medical
research:
25
1. Underrepresentation of females in biomedical research: Many studies have shown that women are
significantly underrepresented in medical research. This lack of representation can lead to gender
bias in the findings and recommendations derived from research studies. It has been observed that
diseases thought to affect men more than women tend to dominate the research picture, while
diseases that affect women may receive less attention and funding.
2. Poor reporting and analysis of sex and gender: The document highlights that sex and gender are
poorly reported and analyzed in contemporary biomedical research. Many studies fail to include
sex as a variable in their analysis, and there is underreporting of the influence of sex and gender
on study outcomes.
The results of the study revealed several important results. First, females remain underrepresented in
biomedical research, suggesting a gender gap in the literature. Second, the analysis showed that sex and
gender are poorly analyzed and reported in research, indicating a lack of attention to these important
factors. Finally, the study identified several contemporary research articles that presented ideas that could
be interpreted as misogynistic, highlighting the presence of misogyny in medical literature.
Artificial Intelligence Solutions for Health 4.0: Overcoming Challenges and Surveying
Applications
Health 4.0 is a concept that focuses on integrating advanced artificial intelligence technologies
into the healthcare industry .and its supports health service technologies by efficiently using existing
resources in health institutions. It contributes to personalized treatment and drug development by
establishing a centralized patient management system. Moreover, it contributes to reducing medical errors
by making the proper diagnosis by people expertly trained in these techniques. Health institutions should
encourage diagnostic procedures while supporting reducing digital health costs. Governments aim to
develop the level of medical care in hospitals and clinics to ensure the provision of healthcare benefits at
low costs and increase patient satisfaction.
AI can be used to improve patient monitoring and diagnosis in several ways. For example, modern
applications help older adults and people with disabilities to access health services faster, even if they are
in remote geographical locations. Also, it reduces the workload of healthcare workers, supports doctors,
makes appropriate clinical decisions, and provides early treatment for rapid diagnosis. In line with
advances in imaging techniques, visualization of lesions that are difficult to see with the naked eye and
detection of potentially overlooked images also give a positive direction for treatment. Therefore, it is
necessary to use artificial intelligence applications to develop hospitals and medical clinics. AI can also
be used for personalized medicine, which allows the development of personalized treatment plans based
on an individual’s genetic makeup, medical history, and other relevant factors. Through artificial
intelligence, it is possible to identify each patient’s requirements, increase the effectiveness of treatment,
and reduce harmful effects. Remote monitoring is another application of AI, which involves wearable
devices and sensors that support the Internet of Things, which track the patient, collect data, and send it to
doctors in real-time. These devices allow healthcare workers to intervene immediately and reach patients
quickly.
26
Prediction of heart diseases utilising support vector machine and artificial neural network
The heart, like a pump, is an organ about the size of a fist, mainly composed of muscle and
connective tissue that functions to distribute blood to tissues. The heart is located under the rib cage,
above the diaphragm between the lungs, slightly closer to the left. Sometimes a small, unexpected
problem with the veins or the valves that supply the heart affects a person’s life and can lead to death.
Early diagnosis is essential to predict diseases that affect the human heart and lead people to live another
period of life. In this context, the authors introduce two methods for early diagnosis of heart disease, the
support vector machine and artificial neural network (ANN). The medical data is taken from the
University of California Irvine (UCI) Machine Learning Repository database, and it contains reports of
170 people. The investigation results confirm that the optimal execution is the support vector machine
technique. It gives high-accuracy prediction results. As for the performance of the forward propagation
artificial neural networks technique is acceptable.
Researchers have undertaken several experiments on using artificial intelligence to predict and categorize
cardiac problems in recent years. This section presents five 2020-2021 articles that used machine learning
to analyze heart disease data. selected. In the beginning, four machine learning algorithms (k-nearest) are
studied. Heart disease prediction uses neighbours, naive bayes, random forest, and SVM. Enhancements
have been made to these methods to improve results extraction. The random forest approach yields the
most accurate findings, according to this research. above 88% accuracy.
Neural network Artificial neural networks, component of artificial intelligence, are widely employed and
offer significant promise for the future. Their name comes from mathematical models that mimic
biological neurons. Besides, Artificial neural networks (ANNs) can tackle the hardest problems.
ONCLUSION AND FUTURE DIRECTION
In this paper, two techniques are utilised in predicting whether or not individuals have heart disease.
Four diseases are predicted using 13 medical data of 170 individuals taken from the Cleveland database.
Data
of the 170 individuals are classified into 90 learning datasets and 80 testing datasets. The accuracy and the
Experimental findings indicate that support vector machine is the most accurate tool for predicting and
detecting cardiac problems, assisting clinicians in making informed decisions. in assessing the patient.
Other cardiac prediction methods will be used in the future. illnesses utilizing similar data
An Augmented Artificial Intelligence Approach for Chronic Diseases Prediction
Hospitals and online medical systems provide extensive data that academics might use to
construct improved models using AI. Disease diagnosis, stage prediction, medical wearables, hospital
stay, and death prediction are medical computing’s key study fields. Poor diet, lifestyle, and exercise
habits may be changed through early illness diagnosis and risk assessment. Patients benefit from early
27
chronic illness risk assessment and treatment (1). Medical illness diagnosis using AI is a trend. Physicians
use AI and algorithms to make decisions. Medical lab tests and experiments help doctors diagnose
diseases. It’s simpler to get insight from old data using AI.
A study offered a machine learning-based prediction model for diabetes, kidney, and heart disorders. The
most valuable traits are selected via adaptive probabilistic divergence. Optimizing the most important
illness detection features yielded the maximum accuracy (2). A feature selection-based machine learning
method predicts diabetes, heart attack, and cancer. The incremental feature selection strategy using CNN
predicts illness presence. A faster technique with 93% classification accuracy was presented (18). Critical
characteristics of chronic illnesses are studied in another research. We use Information Gain, Gain Ratio,
and correlation-based feature selection methods.
Proposed Model Architecture
The proposed model architecture is described in this section. Artificial intelligence algorithms are used to
construct the chronic disease detection model. Artificial intelligence is an advanced machine learning
approach to build detection models that can receive input data, train a model, and predict the output of
future data. These model-building techniques are divided into two main categories: supervised learning
and unsupervised learning.
Early diagnosis of chronic diseases is a major research challenge. Various artificial intelligence
techniques are used in literature for medical data classification and disease prediction. Such techniques
are often applicable for selected datasets to diagnose specific disease with a limited set of attributes. In
this paper, chronic diseases are predicted using an augmented artificial neural network-based approach.
The accuracy of the proposed model is improved using the particle swarm optimization (PSO) feature
selection algorithm.
Comparison of sexual functions in women with and without type 1 diabetes
This study aimed to investigate female sexual function in patients with type 1 diabetes by comparing
female sexual function index scores between women with and without type 1 diabetes.
A total of 62 women with type 1 diabetes and 69 age-matched women without diabetes but with similar
backgrounds were enrolled in the patient and control groups, respectively. All participants were sexually
active and had no systemic diseases other than diabetes in the patient group. The frequency of female
sexual dysfunction was significantly higher, and the mean female sexual function index score was
significantly lower in women with diabetes compared to the control group (p=0.01). There was a
significant relationship between sexual dysfunction and duration of diabetes, glycosylated hemoglobin
test, and body mass index (p
