Machine Learning Methods for Disease Prediction and Analysis

Abstract

Many people in today's fast-paced society choose not to seek medical care for their earlystage
illnesses for a variety of reasons, such as a lack of time or a dislike of going to
hospitals. This may cause the disease to advance to a more serious state. We suggest a
method that uses machine learning algorithms to forecast diseases based on patient
symptoms and individual characteristics, such as age and weight, in order to address this
problem. Additionally, the system uses a content-based approach to suggest suitable
medications and medical professionals depending on the anticipated ailment.The system
employs a thorough methodology, beginning with data preparation to transform category
and textual data into numerical form appropriate for model training. Then, 80 percent of
the dataset is used for training and 20 percent is used for evaluation. The illness
prediction models are trained and tested using six distinct machine learning methods,
including SVC, Logistic Regression, Naive Bayes, Decision Tree, Random Forest, and
XGBoost. Utilising accuracy ratings and cross-validation with a fold of five, performance
evaluation is carried out.The Nave Bayes model, which has the best accuracy of 96%
according to the findings, is chosen for disease prediction. Cosine similarity is used in a
content-based recommendation system that uses disease features to suggest doctors and
medications. A score between 0 and 1, where 0 denotes no resemblance and 1 denotes
total similarity, is produced by the cosine similarity method, which analyses the
similarities between diseases based on their characteristics.