Data-Driven Early Diagnosis of Chronic Kidney Disease: Development and Evaluation of an Explainable AI Model

Aim:

      The primary aim of this research is to design, build, and rigorously evaluate an interpretable AI model for the early diagnosis of Chronic Kidney Disease (CKD) by leveraging a diverse dataset of patient information.

Abstract:

       Chronic Kidney Disease (CKD) is currently experiencing a growing worldwide incidence and can lead to premature mortality if diagnosed late, resulting in rising costs to healthcare systems. Artificial Intelligence (AI) and Machine Learning (ML) offer the possibility of an early diagnosis of CKD that could revert further kidney damage. However, clinicians may be hesitant to adopt AI models if the reasoning behind the predictions is not understandable. Since explainable AI (XAI) addresses the clinicians’ requirement of understanding AI models’ output, this work presents the development and evaluation of an explainable CKD prediction model that provides information about how different patient’s clinical features contribute to CKD early diagnosis. The model was developed using an optimization framework that balances classification accuracy and explainability. The main contribution of the paper lies in an explainable data-driven approach to offer quantitative information about the contribution of certain clinical features in the early diagnosis of CKD. As a result, the optimal explainable prediction model implements with a high accuracy. In addition, an explainability analysis shows that hemoglobin is the most relevant feature that influences the prediction, followed by specific gravity and hypertension. These features are importance for results in a reduced cost of the early diagnosis of CKD implying a promising solution for developing countries.

Existing System:

           The existing system of diagnosis is based on the examination of urine with the help of serum creatinine level. Many medical methods are used for this purpose such as screening, ultrasound method. The datasets are preprocessing after that collected. Then the ordinal encoding, nominal encoding methods are implemented to the datasets. The features selected are denoted by their type (numerical, nominal, and ordinal) as well as the selection method (i.e. ANOVA, Chi-squared, Mutual Information, or Recursive Feature Elimination).

           Using the mutual information technique, the features selected are hemo, htn, and sg. Random Forest and Extra Trees (both bagging ensemble trees) achieved similar results. This method is old and slow. We want to detect the kidney disease as soon as possible. So we moved to the proposed system.

 Proposed System:

         Our Aim is to predict the chronic kidney disease using the machine learning algorithm. Chronic kidney disease (CKD) means your kidneys are damaged and can’t filter blood the way they should. The disease is called “chronic” because the damage to your kidneys happens slowly over a long period of time. This damage can cause wastes to build up in your body. CKD can also cause other health problems.10% of the population worldwide is affected by chronic kidney disease (CKD), and millions die each year because the doctors are unable diagnose the disease.

           Since the aim of this research is to achieve the most balanced CKD prediction model in terms of classification performance and explainability. The system is automation for predicting the CKD. We proposed Random Forest, Logistic Regression, Gradient Boost, XGBoost and SVM machine learning technique for kidney disease prediction of significant features. ML process starts from a pre-processing data phase followed by feature selection based on data cleaning, classification of modeling, performance evaluation, and the results with accuracy. In this process Gradient Boost achieves the results with high accuracy and low rate of error.