Design and Implementation of a Clinical Data Classifier: A Supervised Learning Approach

作者：Gracia Jacob^*; Geetha Ramani R

来源：Research Journal of Biotechnology, 2013, 8(2): 16-26.

• Artificial intelligence
• Clinical Data Classifier
• Evaluation measures
• Machine Learning
• Supervised Learning

Research in the sphere of artificial intelligence and its application in medicine have witnessed a remarkable advancement in the recent past initiating a renaissance in the field of bio-informatics, medicine and computer science. In this paper, our aim is to present a new methodology to formulate and evaluate a rule-based Clinical Data Classifier (CDC) that is designed to diagnose diverse ailments and classify DNA sequences. The proposed methodology utilizes computational approaches to prepare and process clinical patient records followed by the application of supervised machine learning techniques to generate classification rules. We record the performance and evaluation results of twenty supervised machine learning techniques on diverse clinical datasets comprising of more than 11, 000 patient records. The design is implemented as a Java-based clinical classifier with a user-friendly interface. The classification accuracy obtained with designed methodology in the training phase is 100% while the implemented CDC reports an accuracy ranging from 55% to 98% spanning varied clinical ailments. Our results prove that the C4.5 Decision Tree Algorithm and the Random Tree Classification Algorithm are the most efficient in classifying clinical patient records of varied nature.

• 出版日期2013-2