Project

Disease diagnoser-a doctor's reference

The diagnosis of a disease is the most critical and vital job in medicine and it mostly depends on a doctor’s intuition based on past experiences. The unfortunate case of 
 recognizing the incorrect symptoms results in a misdiagnosis. To avoid such medical 
 misdiagnoses, my project shows it is beneficial to utilize large datasets collected by 
 healthcare industries to automate the diagnosis of diseases. Such a tool can assist doctors 
 to avoid the unwanted biases in diagnosis. In addition, an automated medical diagnosing 
 system would be useful if the symptoms are ambiguous because including large datasets to 
 the currently known symptoms may illuminate the case.
 This project aims to develop machine-learning models to automate the diagnosis 
 of disease. Apart from existing base algorithms and hybrid algorithm (Naïve Bayes with 
 decision tree) in literature, my project implements a new hybrid algorithm with Naïve 
 Bayes and random forest for classification of diseases. Hybrid algorithms are used to 
 overcome the limitations of basic algorithms thereby producing a better machine learning 
 model, which improves accuracy of classification. 
 The tool allows the user to select the category of disease and enter the symptoms 
 of the selected category. Among all the machine-learning models developed using the 
 preprocessed datasets, the model that achieves highest accuracy with the test dataset is used 
 to analyze the symptoms entered by the user. The result of the diagnosis along with the 
 probability of its occurrence and accuracy of the developed model are displayed to the user.

Project (M.S., Computer Science)--California State University, Sacramento, 2018.

The diagnosis of a disease is the most critical and vital job in medicine and it mostly depends on a doctor’s intuition based on past experiences. The unfortunate case of recognizing the incorrect symptoms results in a misdiagnosis. To avoid such medical misdiagnoses, my project shows it is beneficial to utilize large datasets collected by healthcare industries to automate the diagnosis of diseases. Such a tool can assist doctors to avoid the unwanted biases in diagnosis. In addition, an automated medical diagnosing system would be useful if the symptoms are ambiguous because including large datasets to the currently known symptoms may illuminate the case. This project aims to develop machine-learning models to automate the diagnosis of disease. Apart from existing base algorithms and hybrid algorithm (Naïve Bayes with decision tree) in literature, my project implements a new hybrid algorithm with Naïve Bayes and random forest for classification of diseases. Hybrid algorithms are used to overcome the limitations of basic algorithms thereby producing a better machine learning model, which improves accuracy of classification. The tool allows the user to select the category of disease and enter the symptoms of the selected category. Among all the machine-learning models developed using the preprocessed datasets, the model that achieves highest accuracy with the test dataset is used to analyze the symptoms entered by the user. The result of the diagnosis along with the probability of its occurrence and accuracy of the developed model are displayed to the user.

Relationships

Items