Diagnose Kidney Cancers Using Transcriptome Profiles of Patients Using Classification Algorithm

Abstract

Kidney cancer is one of the dangerous diagnosed cancers globally, resulting in the high number of deaths and suffering every year. The WHO states that it is the it is common in men and the fourth in women, with 338,000 new cases and 144,000 deaths a year. The detection of a disease at the earliest stage is the key to the successful treatment and a better prognosis, but the manual diagnosis based on clinical data and imaging is a big problem for healthcare professionals. Therefore, this crucial problem can be solved by advanced technologies like AI and image processing which are highly effective. Using the strength of AI, for instance, machine learning algorithms, scientists have turned to datasets like The Cancer Genome Atlas (TCGA) Kidney Cancers Dataset, which contain the detailed transcriptome profiles of patients. Through the study of these profiles, which reveal the nature of gene activity, we are working to create accurate and reliable diagnostic tools for kidney cancer. In our project, a multi-faceted approach was adopted, by using Maximum Relevance Minimum Redundancy (MRMR) feature selection techniques and regression models. The main objective was to use the elaborate patterns of Transcriptome profiles for the machine learning models to be trained accurately. Especially, the regression model proved to be very good, as the accuracy rate was 97%. The importance of these discoveries is the possibility of transforming the clinical decision-making process in the diagnosis of kidney cancer. Through the combination of Transcriptome profiles with machine learning algorithms, clinicians can use improved accuracy and efficiency in the detection of kidney cancer cases. This not only aids in early detection but also provides the patients with the appropriate treatment at the right time, thus the overall results improve and even some lives are saved.