1|Page

Survey on Predictive Medical Data Analysis using Pattern Recognition Algorithm

Dr. T. Senthil Kumar, Professor, Dept. of Computer Science, Amrita School of Engg., Coimbatore, Tamil Nadu, India. drtsk111 @gmail.com Kirti K. Pai Dept. of Computer Science, Amrita School of Engg., Coimbatore, Tamil Nadu, India

Adarsh Suresh, Student, Ajith A. V.,Student, Dept. of Computer Science, Dept. of Computer Science, Amrita School of Engg., Amrita School of Engg., Coimbatore, Tamil Nadu, India. Coimbatore, Tamil Nadu, India. [email protected] Nivedha Samiappan, Priyanka C., Dept. of Computer Science, Dept. of Computer Science, Amrita School of Engg., Amrita School of Engg., Coimbatore, Tamil Nadu, India. Coimbatore, Tamil Nadu, India

AbstractThe medical field is one of the most important industries that makes use of the many advantages of Cloud Computing. The system we propose enables doctors, patients, pharmacists, pharmaceutical companies to find hidden patterns in medical data (EMRs) stored on a cloud, using pattern recognition algorithms. The system is supported by a knowledge base that helps makes the working of the algorithm more accurate. The dataset consists of the attributes namely: age, gender, region, climate, time period, diseases and respective diagnosis. A user of the system can find patterns of diseases using one of the categories : age, gender, region and time period. The proposed system explores the application of the fuzzy k-means algorithm to cluster the data and tests for any modification of the algorithm that maybe required for producing more efficient and accurate results.

I. INTRODUCTION:
. the twenty-first century, infectious diseases Even in continue to emerge at a rapid pace, in turn causing economic and social disruptions. Methods for enhanced detection, verification, and response capabilities uniquely positions three industrieslife sciences, food and agriculture, and health care to mitigate the impact of emerging infectious diseases on society. Thus, being able to find patterns in the occurrences of diseases with respect to factors like gender, age, change of climate, etc. will help people to be better equipped to take preventive and precautionary measures. This paper undertakes the same purpose and tries to propose a system for the same. The aim of the paper is to mine interesting and hidden pattern of diseases and disorders that dwell in the community and to classify them depending on gender, age, region and time period with the help of data mining concepts. A knowledge base consisting of medical data, that is, names of frequent diseases and the causes and preventive measures for them are built. With the help of the developed knowledge base, patterns from Electronic Medical Records (EMR) is mined using different data mining methods. A graphical representation of the mined pattern is also created and the diseases based on the criteria of time period, gender, region, age are classified.

2|Page Keywords Cloud Computing; Data mining; results relevant to the users choice of criteria for the search and allowing the user to see the graphical plot are returned. The stakeholders of this system are the pharmaceutical companies, doctors and users or patients in general. To implement our proposed model and evaluate its performance, we use a dataset from a single tertiary hospital. The following sections talk about the significance of the component/ technique/technology involved. 1.Cloud Computing With the advent of cloud computing, many applications make use of its various benefits such as mobility of data, ease for storage and heavy computing etc. Cloud Computing is germinating its benefit to industrial sectors especially in medical scenarios. In Cloud Computing, IT-related capabilities and resources are provided as services, via the distributed computing on demand [2]. Cloud is necessary for this project because large no of EMRs are required to get precision and accuracy in our output. This makes cloud a necessity as its easier to store and retrieve large amounts of data in cloud. Mining of large data is a tedious job and it includes a lot of computational work which is easier and faster in cloud compared to other methods. 1. IaaS (Information as a Service): The data set can be stored on the cloud using IaaS. The data set would be stored in a cloud data center. The data set involves medical records. 2. SaaS (Software as a Service): Access of data by the users such as pharmaceutical companies and other stake holders is cloud acting as SaaS. Here the program we develop to find the pattern acts as the software. Software as a service[2] commonly referred to as on demand software service, which is hosted centrally. It is software delivery model, in which software and 2 its associated data are hosted (typically in the internet cloud) that are being characteristically accessed by users, normally using a web browser on the user interface layer over the Internet. SaaS has become a most necessarily needed application for the majority commerce applications, including accounting, collaboration, customer relationship management(CRM), enterprise resource planning (ERP), invoicing, human resource management (HRM), content management(CM) and service desk management. 2. Knowledge Base: A knowledge base is used in our system to store all necessary information such as known diagnosis for the disease, symptoms, etc. to help produce accurate results. The information is consulted with during the phase of mining and extracting patterns. As elaborated in the work of Jae-Kwon Kimet.al. [1], a knowledge base will contain domain knowledge and important supportive information (rules generated from a fuzzy membership function in the work here) along with medical guidelines, all which are verified by medical experts. In the above work, a decision-tree rule induction technique creates mining-based rules that are subjected to validation by medical experts. As the rules may not be medically suitable, the experts add rules that have been verified and delete inappropriate rules. 3. Role of Data mining Mining technique is necessary for our project is because: Extracting information on how and what diseases affect different patients from different with the help of EMRs.

3|Page

Analyzing EMRs through the several methods bounded along with mining such as Automated discovery of previously unknown pattern. On basis of performance, mining is better in gathering and analyzing data faster through parallel processing systems.

Algorithm followed as per Mahout framework: Initialize k clusters Until converged o Compute the probability of a point belong to a cluster for every <point,cluster> pair

As applying several methods Association rule learning, we are able to identify the pattern within the EMRs and extract interrelationship among the dataset. From IV.LITERATURE that we can SURVEY: predict the future pattern. Data mining techniques have recently been attracting attention as a means of enhancing data processing capacity and solving complex problems using computers [1]. Expert systems, which use data mining techniques, help to deal with complex and specialized decision-making issues . 4. Pattern Recognition Algorithm suggested Clustering techniques are unsupervised learning methods for grouping similar data, hence, popularly used for various data mining and pattern recognition purposes. Fuzzy K-means (also called Fuzzy C-Means or Soft clustering) is an extension of K-means (or Hard Clustering) is the algorithm chosen for the proposed system. K-means algorithm discovers clusters wherein a point belongs to only one cluster whereas Fuzzy Kmeans discovers soft clusters, that is, a particular point can belong to more than one cluster with certain probability. This is more apt for finding patterns and trends.

Recompute the cluster centers using above probability membership values of points to clusters. II. RELATED WORK The literature chosen to support our work is basically derived from two research papers. We aim to imitate the use of the knowledge base as the knowledge-based Clinical Decision Support System built in the work by Jae-Kwon Kimet.al.[1] executes. This paper proposes the Fuzzy Rule-based Adaptive Coronary Heart Disease Prediction Support Model (FbACHD_PSM), which gives content recommendation to coronary heart disease patients. The proposed model uses a mining technique validated by medical experts to provide recommendations. The aim of the work by N. Karthikeyan .is to provide palm vein pattern recognition based on a medical record retrieval system, using cloud computing. This system was made in consideration for mentally affected, differently abled and unconscious patients who cant communicate about their medical history to the medical practitioners during an emergency. The paper reveals an efficient means to view, edit or transfer the DICOM images instantly which was a challenging task for medical practitioners , thus making use of the efficiency of cloud.

4|Page

Fig. 1. The Proposed System

III. LITERATURE SURVEY

PATTERN RECOGNITION PAPERS Methodology/Approach Author/Year

Adaptive mining prediction model for content recommendation to coronary heart - disease patients; Jae-Kwon Kim; Jong-Sik Lee ; Dong-Kyun Park ; et. al This paper proposes the Fuzzy Rulebased Adaptive Coronary Heart Disease Prediction Support Model (FbACHD_PSM), which gives content recommendation to coronary heart disease patients. The proposed model uses a mining technique validated by medical experts to provide recommendations.

Merits
To increase its prediction accuracy, knowledge and mining-based rules were generated, and the decision tree technique was used.

Limitations
Not scalable to a large dataset.

5|Page

Legend Extraction From ELearning Video Streams: Ji-Wen Chio, Shu-Yuan Chen

The proposed method separates the foreground from video streams Shot boundaries are detected to resample a video stream in a nonredundancy training set during the training phase background map together with a set of colors used frequently is then constructed based on the training set, subsequently used for illustration extraction a foreground map is generated according to the background map and frequent color set for each frame of the video stream.

CLOUD COMPUTING PAPERS Author/Year

Cloud Based Emergency Health Care Information Service in India; N. Karthikeyan & R. Sukanesh

Methodology/Approach
With software as a service (SaaS) by means of Cloud computing, it aims to bring emergency health care sector in an umbrella with physical secured patient records. Also giving a ubiquitous access to appropriate records using Palm vein pattern recognition

Merits
It reveals an efficient means to view, edit or transfer the DICOM images. Merits of cloud computing well supported.

6|Page

DATA MINING PAPERS Author/Year

Mining and Integrating Reliable Decision Rules for Imbalanced Cancer Gene Expression Data Sets; Huang Jun Ni2, Yuanyuan Dan3, Sen Xu4

Methodology/Approach
This paper presents a skewed gene selection algorithm that introduces a weighted metric into the gene selection procedure.

Merits
Classification performance is improved and over fitting is avoided by combining multiple feature gene pairs (decision rules) into an ensemble learning framework.

Limitations
Mining of the decision rules is quite time consuming

Rajaram, M. A novel approach for mining Association rules on sports data using principle component analysis;
UmaMaheswari.PComput. Sci. & Eng., Sona Coll. of Technol. Anna Univ., Salem Rajaram, M. Year: 2009

Association using principal component analysis

The efficiency of mining algorithm is improved provided that Principal Component Analysis generates frequent patterns.

Experiment outcomes may not be fully accurate

7|Page

V.REFERENCES AND ACKNOWLEDGEMENTS 1. Jae-Kwon Kim Jong-Sik Lee Dong-Kyun Park Yong-Soo Lim Young-Ho Lee Eun-Young Jung ; Adaptive mining prediction model for content recommendation to coronary heart disease patients ;Cluster Comput. ; DOI 10.1007/s10586-013-0308-1;; Received: 30 April 2013 / Revised: 7 August 2013 / Accepted: 28 August 2013. 2. N. Karthikeyan& R. Sukanesh; Cloud Based Emergency Health Care Information Service in India; Received: 4 May 2012 / Accepted: 27 July 2012 / Published online: 3 August 2012. 3. Chen Xu; Biomimetic pattern recognition;Topological pattern recognition; Accepted:July 20. 4. Ji-Wen Chio , Shu-Yuan Chen ;Legend Extraction From E-Learning Video Streams;pattern recognition; Accepted: August 2011; 5. Jackson, K.R. , Ramakrishnan, L. , Muriki , K. , Canon,S. , Cholia.S. , Shalf.J. ,Wasserman and Harvey J.;Performance Analysis of High Performance Computing Applications on the Amazon Web Services Cloud; Accepted: 2010; 6. Gaizhen Yang ,Zemin Zhu and Fen Zhuo; The Application of SaaS-Based Cloud Computing in the University Research and Teaching Platform; Accepted: 20 August 2011 7. UmaMaheswari.PComput. Sci. & Eng., Sona Coll. of Technol. Anna Univ., Salem Rajaram, M. A novel approach for mining Association rules on sports data using principle component analysis Accepted: 2009 8. Saraee, M.H. Isfahan Ehghaghi. Z., Meamarzadeh, H. ;Applying Data Mining in medical data with focus on mortality related to accident in children. Accepted: 2008