Decision Support System for Diagnosing Cardiac Diseases Being Deployed on Private Cloud
The fourth industrial revolution brought exceptional development in daily life. Automated, Predictive and Decision Support Systems backed with Artificial Intelligence rise have become one of the major outcomes of this fourth industrial revolution. The recent development in Artificial Intelligence op
2025-06-28 16:31:05 - Adil Khan
Decision Support System for Diagnosing Cardiac Diseases Being Deployed on Private Cloud
Project Area of Specialization Artificial IntelligenceProject SummaryThe fourth industrial revolution brought exceptional development in daily life. Automated, Predictive and Decision Support Systems backed with Artificial Intelligence rise have become one of the major outcomes of this fourth industrial revolution. The recent development in Artificial Intelligence opened many doors to predict information and take measures for almost every industry. The medical industry is one of those industries on which Artificial Intelligence will be an excellent fit.
Our Final Year Project is a Decision Support System to diagnose possible disease using machine learning. Initially, we are working on the data-sets to diagnose Heart Disease and their types using case-based reasoning with machine learning. To determine whether a particular person has heart disease or not we are using BNG Cleveland's data-set which has over a million instances of patients. Binary classification will be used to detect the possibility of heart disease after that multi-classification algorithms will be used to find the type of heart disease a patient may have. Each time on successful prediction and authentication by the doctor, new patient data would be added in the dataset and in this way very valuable data for the future would be generated for further research and performance. The project consists of a Web Application, Mobile application and in the end a research paper that will be written on the analysis and results of the data-sets. The web application will have five different modules for each user type. The system admin will be responsible to add the hospitals where we are providing this software as a service. Each registered hospital using this software will have a Hospital admin-end through which details of doctors, patients, and their reports will be added. The concept of data management will be similar to EMR( Electronic Medical Record). The doctor end of the web application will be able to see a patient's history, reports and perform algorithms to predict possible cardiac disease. The patient can see it's details, appointment timings, and test results.
The mobile application would be for the general public where different trained models of several diseases such as cardiac diseases, cancer, liver diseases will be kept. The users can add symptoms and will be able to know the possibility of a particular disease. The mobile application will also provide information about the latest research on fatal diseases.
The software as a whole would be deployed on a private cloud to make sure Electronic Medical Record (EMR) concept is achieved by providing software as a service for the hospitals.
- Gathering valuable medical-related dataset.
- Mining and analysis of each disease's data-set.
- Classification of Heart Diseases.
- Speeding up diagnosing and treatment stages.
- Electronic Medical Record Management of Hospitals, Doctors Patients.
- Automated Test Report referrals from the doctor to the laboratories.
- Single unique record for a patient throughout the system.
- Mobile Application for the community.
- Deploying Software on a private cloud for providing Software as a service.
- Research paper on the gathered data.
Machine Learning:
Diagnosing the disease is the core functionality of this project and for that, we need to have all kinds of data-sets related to cardiac diseases. The BNG Cleveland dataset has a record of 1 million patients. The dataset contains symptoms and such as chest pain type, ECG, Blood Sugar levels, etc. We are using Python's Apache Spark to train and test our data with different machine learning models and optimization methods. The models we are applying are SVM, CNN, Random Forest, Gradient Boosting. In the end, we are applying the voting technique to get the best accuracy of the prediction. The algorithms will first predict disease's existence with binary classification and then disease's type with multi-classification. The next part will be image classification of MRI scans given the availability of the dataset. In this phase, we will be using Keras with TensorFlow backend to perform classification.
Web Application:
We are using Python's Django framework for developing the front and the backend of our Web Application. Django uses the MVT (Model View Template) pattern to make the app functional. In our case, it is the right choice to use it as our machine learning models are also in the same language and both together will speed up the process.
Mobile Application:
Mobile Application is been developed using React Native. It will be a cross-platform mobile application. Firebase will be used in the backend for sessions, authentication, storage, and security. Application program interfaces will be used to get the information from the web and view it in the app.
Cloud Computing:
We will be using the OpenStack cloud computing platform to create and configure our private cloud. The software will be deployed on the cloud. The whole software would be given as software as a service and it will provide different resources such as virtual servers to the user of the application.
Research Paper:
Based on performed techniques on the available dataset we will be writing a research paper. The research paper will describe the pre-processing and analysis of the dataset. It will also include the best techniques used in the development phase for model training, accuracy, and prediction.
According to statistics one out of three heart attack cases is misdiagnosed. It can be because of any reason such as inexperience, disease complexity, emergency, etc. Understandably, a doctor's memory cannot keep in mind all the previous cases to decide and diagnose the current one.
The software will assist the doctor to predict, diagnose and treat the cardiac disease. Each case will be analyzed and compared with millions of previous cases and their results, based on those facts and results system predicts the outcome. This system would be extremely beneficial in an emergency where a single detail might get neglected due to a shortage of time. In such situations, a system like this would be perfect to analyze every single detail and make a prediction in seconds. The data predicted and authenticated by the doctors will be so helpful for researchers in the health industry. Researchers can get insights from the data from each hospital about the disease and perform further research. The system will make sure to eliminate Data redundancy, a patient's record and its history would be the same in any hospital, this will save a lot of resources such as time, storage and processing. The Electronic Medical Record Management will be very useful for the government to have data on one single platform with no redundancy.
The mobile application will help the general public to predict a particular disease based on their symptoms. The result will provide necessary information about a particular disease if any, and suggest the nearby hospital which treats it. The mobile application will also be informative for ongoing research in the medical field as it will be fetching news from the internet with the help of API.
Deployment of Software on cloud infrastructure will make the software more scalable, accessible stable and secure.
In the end, the paper written on the heart disease prediction will help in finding new techniques to predict and diagnose the disease and it will open doors for further research in the medical field under the umbrella of Artificial Intelligence
The Final Deliverable will include a fully functional Decision Support System to diagnose cardiac disease. The web application will store information of Hospitals, Staff, Doctors, Patients and their complete medical history with reports files. In the reports, machine learning algorithms will be applied to predict and diagnose the disease. After the doctor's confirmation each patient's instance will be added in the training dataset to generate more data for the future and improve the accuracy. The doctor can also refer the patient to the lab for more test and after the result, the same decision support cycle will be continued throughout the treatment. Each report will be tested on the machine learning models to help the doctor decide the treatment. The data management system will be EMR based with no data redundancy. Each registered patient will have a single record throughout the whole software with deployment on the cloud the system will also make sure data consistency.
The Mobile Application will contain different trained models of several diseases. The user can add symptoms of a particular disease and can predict its existence and type. The mobile application will also show disease's complete detail including treatment, cautions, warning, etc. The mobile application will be connected with the latest medical research news on the internet with the help of an API. A registered user can save the results of predictions, compare and search about it.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 80000 | |||
| Documentation Printing (SDS, SRS, Research Paper) | Miscellaneous | 7 | 1000 | 7000 |
| Wire Frames Printing and Documentation | Miscellaneous | 3 | 1000 | 3000 |
| 16GB DDR4 SODIMM Memory( For Machine Learning Models Training) | Equipment | 1 | 28000 | 28000 |
| SSD (Data-set Storage and Processing) | Equipment | 1 | 22000 | 22000 |
| GPU (Machine Learning Images Model Training) | Equipment | 1 | 20000 | 20000 |