International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 11 Issue: 06 | Jun 2024
www.irjet.net
p-ISSN: 2395-0072
PSO Based Feature Optimization for Diabetes Detection using Machine Learning Shivani Nai1, Mr.Nagendra Kumar2, Mr. Prateek Gupta3 1Reseacrh Scholar, Department of CSE, Shri ram Institute of Science and Technology, Jabalpur, M.P. 2Professor, Departmet of CSE, Shri Ram Institute of Science and Technology, Jabalpur, M.P. 3Professor, Departmet of CSE, Shri Ram Institute of Science and Technology, Jabalpur, M.P.
---------------------------------------------------------------------***-------------------------------------------------------------------diagnosed during pregnancy. While insulin secretion does Abstract – Diabetes is a chronic disease that occurs when
not work in type 1 diabetes due to the destruction of pancreatic beta cells, there is a disorder in insulin secretion and function in type 2 diabetes. Gdm is a glucose intolerance first diagnosed during pregnancy; it may be mild, but it is also associated with high blood sugar and high insulin levels during pregnancy. All of these types can cause a lack of blood sugar in the body, which can lead to serious diseases in the body. In other words, when blood sugar rises above normal, this condition is called hyperglycemia. On the other hand, when it decreases and falls below normal, the condition is called hypoglycemia [3]-[5]. For example, hyperglycemia can cause chronic problems and lead to kidney disease, retinopathy, heart disease and other tissue damage, while hypoglycemia can cause temporary short-circuits. It can cause kidney disease, retinopathy, heart and heart disease, and other tissue damage can lead to diabetic coma [1] , [3] , [4] . Diabetes has become an important health problem in today's world due to its prevalence in children and adults. According to [6] , [7] , approximately 8.8% of adults worldwide had diabetes in 2015, and this number was approximately 415 million and is expected to reach approximately 642 million by 2040. During this time, more than 500,000 children were killed and approximately 5 million people died. On the other hand, the global economic burden of diabetes was estimated to be approximately 673 billion dollars in 2015, and is expected to reach 802 billion dollars in 2040. Self-monitoring of blood glucose (smbg) using fingertip blood glucose meters is a diabetes treatment method introduced three years ago [8], [9]. In this way, diabetics invasively measure their blood sugar levels three to four times a day by using a thumb glucometer to prick the skin of their fingers. The idea is to provide this: to increase insulin resistance. However, in addition to being laborious and laborious, this method can only be understood if insulin estimation is obtained from small smbg samples. In other words, this may cause the blood sugar in the blood to be higher than normal. To overcome this problem, continuous blood glucose monitoring (cgm) has been introduced, which can provide maximum information about changes in blood sugar within a few days, allowing a good treatment decision to be made for people with diabetes. In this way, blood sugar concentration is constantly monitored thanks to small devices/systems that monitor the glucose level in the blood environment. These systems can be invasive, minimally invasive or noninvasive. Moreover, cgm systems can be
the body cannot use insulin properly or the pancreas cannot produce enough hormones to control blood sugar. High blood sugar is a symptom of diabetes, a group of metabolic diseases. The two most common types of diabetes are type 1 and type 2, but there are other types, such as gestational diabetes, which occur during pregnancy. There is an increase in the number of type 1 diabetic patients. The genetic disease called type 1 diabetes has a long incubation period and usually manifests itself early in life. People with type 2 diabetes have cells that do not respond to insulin. It changes over time and mostly depends on people's lifestyle. According to the 2022 report of the international diabetes federation, approximately 382 million people worldwide are currently living with diabetes. This number is expected to increase to 592 million in 2035. One of the most common causes of tissue and organ damage and dysfunction, including blindness, kidney failure, heart failure, and stroke, is diabetes. Therefore, early diagnosis of diabetes is important. This project focuses on the use of machine learning such as logistic regression, knn, svc, ann, and random forest to predict diabetes. Each algorithm is calculated to determine the accuracy of the model. Additionally, the highest accuracy of 96.00% was achieved in logistic regression using the pso-based optimization technique.
Keywords: Diabetes Detection, Blood Glucose, Feature Selection, Machine Learning, deep Learning, PSO, Accuracy.
1.
INTRODUCTION
Diabetes is a non-communicable disease that affects the control of blood sugar levels in the body. Blood glucose concentration is normally regulated by insulin and glucagon, two hormones secreted by the beta (β) and alpha (α) cells of the pancreas, respectively. The normal release of the two hormones regulates blood sugar levels in the body between 70 - 180 mg/dl (4.0 - 7.8 mmol/l). Insulin lowers blood sugar, while glucagon increases blood sugar. However, abnormality of these hormones can lead to diabetes. There are many types of diabetes, many different types, but the most common types are type 1 diabetes, type 2 diabetes and gestational diabetes (gdm). Type 1 diabetes is more common in children; while type 2 diabetes is more common in adults and the elderly, gdm is more common in women and is
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 637