PROCESS MINING IN CANCER PREDICTION OF PROGNOSIS BY USING CLASSIFIERS BASED ON MIXED VARIABLES

Abstract

A healthy decision-making process for health of the individual is a major challenge in this day and age of abundance of information all over the place. Machine learning, data mining and computation statistics have become among the top research areas which enable individuals empowered to make important decisions that will improve the results of any field. A high demand for data handling is evident in the healthcare sector, since the increase in the number of patients is proportional the increase in population and lifestyle changes. Strategies for early diagnosis and prognosis predictions of diseases are of great importance at the present time to offer better healthcare for the entire human population. Data mining is an advantage in the development of high-quality and effective models for applications to predict health. Since cancer has been everywhere in recent years, information from the cancer registry are being used as medical information in this research. The principal goal of the thesis is to create an efficient and effective classification model for predictions of prognosis for cancer. . The majority of the current system is based on diagnosis prediction model from surveys or screening data since the data is readily available and simple to gather due to the insensitivity of the variables involved in these studies. For prognosis prediction, it requires specific information about the patients who are receiving treatment for a recognized disease. Hospitals and community registries managed by the state constitute the most important sources for data gathering. Electronic hospital records that are well-maintained that include histopathology data are not accessible in India to researchers. Thus, data on cancer obtained from an US accessible data centre was utilized in this study to conduct all experiments. This research study is a model that improves accuracy of predictions by utilizing the appropriate methods for data mining for each stage. Prognosis refers to the survival percentage of cancer patients generally, but is also a measure of how severe the cancer in the future timeline that the person. The two-fold goal of this study is to discover the key response variables that are a part of the prediction system used to determine the prognosis as well as enhance the predictive models. .