Classification of Various Diseases Using Machine Learning And Deep Learning Algorithms
DOI:
https://doi.org/10.46243/jst.2021.v6.i4.pp25-31Keywords:
Disease Classification, Machine Learning, Deep Learning, Random Forest ClassificationAbstract
The latest trending market has many medical helps available, but there is inadequacy of an application or a website where we can have several machine learning paradigms implemented to predict diseases. This approach will have your disease predicted on based of several existing datasets using many of the machine learning algorithms. It can further also be improved with additional of speech modules. The datasets can be .csv or .xlsx or database files. It has a symptom inputting module where the user can enter the information of how he is suffering. The input is parsed and on basis of the keywords found, another panel related to those keywords will appear and take the health update in a more precise format. After submitting it, the disease is predicted and a possibility will be recommended. The project will be a combination of lung pneumonia detection system, chronic heart disease detection, diabetes risk prediction, lung pneumonia, brain tumor, malaria and other detections in a detailed manner to use more parameters thereby increasing the accuracy. Various ML algorithms like Convolutional Neural Networks, Random Forest Classification, Decision Tree and Support Vector Machines, SVM have been used to generate highest possible accuracy. CNN was used to classify Chest X-Ray images and gave 97.03% of accuracy. The pre-existing VGG-16 was used by add-up of the brain tumor prediction dataset and it was combined with the Canny Edge Detection Algorithm to generate an accuracy of 96.32%. Later, a hybrid ML algorithm was designed to classify heart and diabetic risk. It was developed as a Stacking Hybrid Classifier that has SVM on Level-0 and RFC on Level-1 of the stack. It gave cross-validated (boosting) accuracies after a10-fold CV as 91.66% for diabetes risk prediction and around 100% for heart risk prediction.