Prediction of Neurological Disorders among Children Using Machine Learning Techniques

Author Name(s): G. Reshma, Dr. P.V.S Lakshmi
Author Email: reshma.gorripati@gmail.com

Abstract

Diagnosis of neurological problems in children at the earliest helps the medical associates to improve the children’s health conditions. Hence, there is an important need to diagnose neurological disorders that occur which may lead to critical problems if care is not taken in advance. Machine learning Techniques helps for analyzing medical data and the problem can be diagnosed in an effective way. This research has identified machine learning techniques on different measures of accuracy in diagnosing three basic neurological disorders. A Neurological data set is collected from Neuro clinic centre to evaluate the performance of Machine Learning Techniques. Among all the attributes collected some attributes were identified as vital for diagnosing the problem. It is evident from the results that the chosen ML techniques produced more accurate results and there is only a slight difference between their performances.

Introduction

Neurological disorders among children rather than adults are increasing around the world and it is estimated that one in five people in the world will be affected by neurological disorders. These are accompanied by a range of symptoms that can be structural and functional like pain, confusion, muscle weakness, unconsciousness, Depressive disorders etc. Neurological disorders diagnosis involves many steps because disorders like Epilepsy, Dementia, and Schizophrenia etc are the most compelling and unable to elucidate the mechanism of the path physiology. Actually, the diagnosis involves analyzing the symptoms, reviewing past medical history and conducting physical tests, especially in case of Neurological disorders, the test report of EEG is recorded to estimate normal disorder or abnormal disorder. Various psychological tests namely clinic neuropsychiatric observation, audio assessment, assessment of Intellectual coefficient are also conducted which are caused by the neurological disorders. In some of the cases, the cooperation of the patient is very much essential in Diagnosing children with these types of disorders. Machine learning (ML) is the learning behaviour done by machines. Researchers have developed numerous techniques that help machines to accept the behaviour of human and tried to make inferences. Methods have also been found in machine learning techniques to deal with uncertain problems. This research has identified three machine learning techniques that are applied on neurological disorders over collected data set. A comparison is made on applied machine-learning techniques and identified the most accurate one which can be utilized in diagnosing neurological problems. Five basic neurological disorders namely, Dementia, Epilepsy, Neurological disorders associated with mal nutrition, Attention Deficit Hyperactivity Disorder (ADHD) are considered. The symptoms that are observed are given as input to the learning techniques and the disorder can be diagnosed based on the symptoms, finally the disorder based on the symptoms is retrieved as their output.

Conclusion

Prediction of diseases accurately at the beginning stage of treatment can be done by expert systems. As a part of that there are various learning techniques which helps to construct expert systems, Medical Expert systems needs to identify the best technique and to be applied for better results even in critical conditions of the patients. In this paper three machine learning techniques are compared and identified from the results that the three techniques produce more accurate results. The data set for the neurological data is not available in large and in future, the research may be applied for a large data set with more number of attributes considering values other than 0,1,2,3 because some attributes might required reports like blood tests, EEG etc. these values are of numeric type . The machine learning techniques need to be more trained for the implementation in real time prediction for all possible set of values chosen as attributes.

974 total views, no views today

Download PDF File

About the author: admin