A Bio Inspired Approach for Cardiotocogram Data Classification

Introduction

Medical data mining has high potential for exploring the hidden patterns in the data sets of the medical domain. These patterns can be utilized for clinical diagnosis for widely distributed in raw medical data which is heterogeneous in nature and voluminous. This collected data can be integrated to form information system for a hospital.

Cardiotocography is a composed word that contains three words: cardio that means fetal heartbeats, toco which means uterine contractions, and graphy that means to the recording technique. Cardiotocography referrers to the technique of recording baby’s fetal heartbeat and uterine contractions for mothers during pregnancy. The machine used for that called Electronic Fetal Monitor (EFM), this machine contains two separate transducers, one to record fetal heartbeats, and the second for recording of uterine contractions. Data recorded are called Cardiotocogram data, which referrers to a set of attributes that help doctors for survey during pregnancy. Figure 1 shows an example of the recorded data

Figure 1.

Example of Cardiotocogram data

Increase and decrease of heart fetal rate is affected by uterine contractions. There are three different mechanisms that are considered as first cause of decreasing in fetal heart rate by uterine contractions: fetal head, umbilical cord, and uterine myometrial vessels. These mechanisms are seen as compression, where when the uterine myometrial veins are compressed, blood flow from the mother to the baby stopped temporary. In this period, medicines say that the mother and baby are physiologically separated from each other. Then after the veins are re-opened again, they allow blood flow from mother to baby, carrying the oxygen and nutrients, also blood carries fetal waste products to flow from the baby to the mother.

Nowadays, technology becomes a necessity to do a lot of tasks in life. Medical data managing also a domain that have been affected by technology, with its development and growing, medical data need more developed systems to manage, where data mining appears in medical systems to assist doctors in their work. Data mining is the core stage of the knowledge discovery process that is aimed at the extraction of interesting nontrivial, implicit, previously unknown and potentially useful information from data in large databases. Machine learning is a part of data mining which it focuses on prediction, based on known properties learned from the training data. Searching for more solutions in data mining and artificial intelligence leads to search for more resources, here, nature appeared as a solution with its different phenomenon and its amazing power to solve hard problems to become a complete field of study and finding new algorithms called Bio-Inspired algorithms. It is the translation of a natural phenomenon to a mathematic algorithm used usually for optimization, bio-inspired algorithms now are used in many fields either software such as data mining, information retrieval, or hardware such as mechanisms for robots.

The following paper reviewed about Cardiotocogram data classification using a bio-inspired supervised approach, based on mechanism of bee worker in hive. The paper is divided to 4 main sections, beginning with related works in both classification of medical data using various data mining techniques and bio-inspired algorithms, then we present our approach in natural life and its adaptation, finally we will discuss obtained results and we finished by major conclusions.