Health Monitoring System For Diabetes
Diabetes is a disease where blood sugar (glucose) is not metabolized in the body. Due to this reason blood sugar in the body increases to a very high level. This process is known as hyperglycemia. This is the condition where body is unable to produce much insulin that body needs t
2025-06-28 16:32:51 - Adil Khan
Health Monitoring System For Diabetes
Project Area of Specialization Wearables and ImplantableProject SummaryDiabetes is a disease where blood sugar (glucose) is not metabolized in the body. Due to this reason
blood sugar in the body increases to a very high level. This process is known as hyperglycemia.
This is the condition where body is unable to produce much insulin that body needs to control
blood sugar. There can also be a possibility where body cannot respond to the insulin that is
being produced in the body. As diabetes is incurable, it must be controlled. If it is not taken
seriously there can be complications like nerve damage, heart attack, kidney failure, and strokes
in the diabetic patient. According to statistics in 2019, an estimated 9.3% of global population
has diabetes. It is expected that the ratio will be increased up to 11% by the end of year 2045.
In our present work, we analyze Health Monitoring System (HMS) applying different machine
learning techniques for different diabetic patient by utilizing BLE-based sensors and real time
data processing on datasets, with help of real-time data processing, the vast amount of
continuous data (e.g., BG, heart rate, blood pressure, weight, and other personal data) from the
BLE-based sensor devices can be handled in real-time. This will allow the patients to monitor
and check their glucose level through a mobile application which will be connected to firebase
(cloud). Then mobile app will tell the patient what is your diabetes level and also it will suggest
what physical exercise a patient must do and it will also suggest insulin level to maintain his
diabetes level in a normal condition. In this way a patient will always remain in touch with his
diabetic level suggested by mobile application and the patient will not have to go to hospital
whenever he needs to have a check his glucose level.
The main aim of this project is to build a tool for diabetic people to help them to controls their diabetes by suggesting them the physical exercises, diets and insulin based on the data received from the wearing sensors.
Objective
- It will monitor diabetes level based on values of blood glucose and blood pressure from wearable sensors.
- It will predict whether patient should take insulin or not. If yes then how much amount of insulin.
- It will also suggest diet plan and exercises according to diabetes level.
- Android Studio
- Freestyle Libre (Continuous Glucose monitoring Sensor) 26000/-
- Blood Pressure Sensor 13000/-
- Neural compute 2 stick 26000/-
- Firebase (as a cloud server)
- Spyder (Machine Learning Algorithm)
SOFTWARE SIMULATIONS 1: Software Implementations and Simulations
We are using two platforms for software part of our system. These two important parts are:
• Machine Learning Based Implementation and Algorithms
• Android User Interface for End Users
Our project regarding diabetes classification that utilized clinical datasets such as the PIMA
Indian dataset. The dataset consists of 768 patients, of which 268 patients are diabetic and 500
patients are normal. There exist different techniques but we are using three techniques named
as, linear model, polynomial, random forest. The proposed model was compared to existing
techniques such as SVM and Naïve Bayes (NB). The result showed that the performance of a
random forest regression model is higher compared to other methods.
Moreover, we are using 3 steps for overall finalizing our algorithm selection process. In first
step, we have tested all algorithms and based on accuracy rate we selected our algorithm which
is random forest regression model. After successful classification of our dataset we predict
whether patient has diabetes or not. This is our second step after algorithm selection. In third
step we are using Decision Tree regression model for predicting whether patient need insulin
or not. If he needs then algorithm will predict how much amount of insulin a patient should
take. It is all about how we are training our system and them testing patients based on those
trainings.
Basically, this part is not included in this semester work but it will be included in our future
work. In this part our end users(patients) will get their prediction results like predictions about
diabetes and insulin. We will use firebase database for storing our data and processing it online.
When our sensors will measure data from patient’s body, they will directly send data to mobile
application using Bluetooth module. Mobile application will send inputs to firebase database
where our machine learning algorithm will work to predict required output based on training
dataset.
In hardware implementations, we only have BLE Sensors that we are using for measuring blood
pressure and blood glucose level for diabetes patients. Patients will wear these sensors and they
will continuously measure blood pressure and glucose level and will continuously send these
measurements into our mobile app. Then our mobile app will send that input to firebase for
processing.
No doubt Internet of thing (IoT) is improving the healthcare of many patients all around the world. As our project is working on Realtime data which is to get input from sensors using mobile application, it provides Realtime data related to health of the patient. Patient will just have to enter their health information on android application to better manage their health activity. In this way, a patient will be well aware of his health every time. Tracking health monitoring for a patient has become much easier with inter of things(IOT) because of expenditures of medical expenses.
So, our health monitoring system have following advantages
- Easier and better access to healthcare
- A patient can monitor his diabetes any he wants.
- a patient can have access to history of his diabetes level through database created by android App.
- Time saving for a patient
- Economical
- He will just have to input his BG level
- And App will suggest him physical exercise and sugar if he will have or not just in a matter of second.
We are developing a health monitoring system in which we predict diabetes for diabetic patients whether
they have or not. We predict based on our input features like blood pressure, glucose
level, skin thickness, age, weight and height. Based on these input features we have trained our
algorithm for a dataset of almost 750 samples. We are using Random forest regression model
for predicting diabetes. After, training our algorithm gives two output values. First value is
prediction for diabetes, it checks whether patient has diabetes or not. Second value depends on
first prediction, if patient has diabetes then system goes for prediction of insulin. It predicts
whether patient needed insulin or not, if yes then how much amount. Moreover, we have two
fundamental parts of our system. First part is our hardware system which is consist of sensors.
Sensor part is input part which is being used for measuring blood pressure and sugar level of a
patient. Sensors will send that input data continuously to mobile app which will send inputs to
our firebase server where our machine learning algorithm will be running for processing our
inputs that will be age, weight and height of patient. Our android app will be used as communicator between sensors and firebase. After overall processing, app will show output values to patients. Moreover, we will predict exercises and diets for patients who have diabetes. These exercise and diets will be based on training that we will give our system in future. These exercises and diets will help patients to reduce their diabetes level. In conclusion, we can say that our system will predict diabetes and will suggest insulin, exercises and diets to patients so that they can easily tackle their disease. It will be a
solution which they can keep with them in their home for regular testing and improvements.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 70000 | |||
| Freestyle Libre (Continuous Glucose monitoring Sensor) | Equipment | 1 | 26000 | 26000 |
| Blood Pressure Sensor | Equipment | 1 | 13000 | 13000 |
| Neural compute | Equipment | 2 | 13000 | 26000 |
| Travelling and lodging | Miscellaneous | 1 | 5000 | 5000 |