Title:Computational Analysis of Diabetes Kidney Diseases Using Machine Learning
Volume: 18
Author(s): Ganesh Chandra, Namita Tiwari, Urmila Mahor, Parashu Ram Pal, Vikash Yadav*Deepak Kumar Mishra
Affiliation:
- Government Polytechnic Bighapur Unnao, Board of Technical Education,
Uttar Pradesh, India
Keywords:
Random forest classifier, gradient boosting classifier, machine learning, support vector classifier, multilayer perceptron, gradient boosting classifier.
Abstract:
The increasing complexity of healthcare, coupled with an ageing population, poses significant
challenges for decision-making in healthcare delivery. Implementing smart decision support
systems can indeed alleviate some of these challenges by providing clinicians with timely and
personalized insights. These systems can leverage vast amounts of patient data, advanced analytics,
and predictive modeling to offer clinicians a comprehensive view of individual patient needs
and potential outcomes.
Currently, researchers and doctors need a faster solution for various diseases in health care. So
they started to use the Machine Learning (ML) algorithms for better solution. ML is a sub field of
Artificial Intelligence (AI) that provides a useful tool for data analysis, automatic process and others
for healthcare system. The use of ML is increasing continuously in healthcare system due to its
learning power.
In this paper the following algorithms are used for the diagnosis of Diabetes and Kidney Disease
such as: Gradient Boosting Classifier (GBC), Random Forest Classifier (RFC), Extra Trees Classifier
(ETC), Support Vector Classifier (SVC) and Multilayer Perceptron (MNP) Neural Network,
In our model, Gradient Boosting Classifier is used with repeated cross validation to develop our system
for better results. The experiment analysis performed for both unbalanced and balanced dataset. The
accuracy achieved in case of unbalanced and balanced datasets for GBC, ETC, RFC SVC, MLP &
DTC are 75.7 & 92.2, 75.7 & 90.1, 74.4 & 80.0, 62.5 & 66.4, 58.3 & 63.0 and 59.4 & 74.5 respectively.
On comparing these results, we found that GBC results are better than other algorithms.
