Review Paper
Content
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY) Abstract : Data mining is the process of selecting ,exploring and modeling a large database in order to discover patterns and models that are unknown. Unorganized and enormous data in various organizations leads to delay in monitoring ,improper planning, inaccuracy in decision making. This paper focusses on various models and techniques used in data mining in health care and industrial area.It provides recommendations for future research in the application of Data Mining in health care database and industrial area. Key Words:-Data Mining Tools,HL7-health level 7,Primary Health Care System, INTRODUCTION: Data Mining is the blend of concepts and algorithms from machine learning ,statistics ,artificial intelligence and data management. with the emergence of data mining, researchers and practitioners began applying this technology on data from different areas such as banking, finance, retail marketing, insurance, fraud detection, science, engineering, health care centers to discover any hidden relationship or patterns. Data mining is rapidly expanding field with growing interest and patterns and manufacturing is an application area where it can provide significant competitive advantage. The use of data mining techniques in manufacturing began in 1990s and it has gradually progressed by receiving attention from the production community. Primary health care center where health related details of each individual are collected region wise .The PHC information system is a subset of the National Health Management System .It consists of health maps ,house numbering ,home-based records, Family Master ,the wall chart, health facility, pictorial and telly sheets ,Medical forms and Health Facility Registers, etc. In this paper the different relevant data mining tools are used for PHC’s are reviewed and proposes a data model for monitoring individuals information for population based health care management system. II.RELATED WORK: 1.Data mining methods with logistic regression in childhood obesity prediction. 2.Data mining warehouse. techniques in Healthcare data
3.High performance data mining using the nearest neighbor join. 4.Describing pattern in health database. 5.Clustering algorithm and decision support. 6.Hub and Spoke using artificial intelligence. Kinds of data:”An EHR is an electronic version” of a patient’s medical history ,that is maintained by the health care provider over time and includes all of the key administrative clinical data relevant to that person’s care under a particular provider ,including demographics, progress notes, problems, medication, vital signs ,past medical history, immunizations, laboratory data, medical images and radiology reports. HL7-It is known as health level 7.It was developed to improve health informatics interoperability. It is working in the 7th layer, application layer, of the open systems interconnection model. EMR AND EHR: Electronic medical record contains patient record that is stored and retrieved locally in a standalone system used by a provider. Electronic health record contains patient information that is stored and retrieved in systems used by all
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY) providers who care about the patient. Examples of applying data mining on EMR/EHR: Ludwick and Doucette study is the adaptation of EMR in primary care. Cerrito works on EMR from an emergency department. Buczak et al works on disease surveillance on EMR. HEALTH LEVEL SEVEN (HL7):HL7 provides standards for interoperability that improve care delivery ,optimize workflow, reduce ambiguity and enhance knowledge transfer among health care providers. Example1:Patient referral data can vary extensively between cases because structure of patient referrals is upto general practitioner who refers the patient[Persson09]. Example2:Catley et al. use neural networks to predict preterm birth on a heterogeneous maternal population[Catley06]. Example3:”Traditional clinical based prognosis models were discovered to contain some restrictions to address the heterogeneity of the breast cancer”[Ahmad09]. Data from heterogeneous sources present challenges [KwiatKowska07]: 1. Sampling Bias:”clinical studies use diverse collecting methods, inclusion criteria and sampling methods”. 2. Refferal Bias: Data represent the preselected group with high prevelance of disease”. 3. Selection bias: clinical data sets include patients with different demographics”. 4. Method bias:”Predictors have varied specifications, granularities and precisions.” 5. Clinical spectrum bias:”Patient record represent varied severity of a disease and co-occurrence of other medical problems”. 6.Transcription and manipulation of patient records often results in high volume of noise and a high portion of missing values. IV. PROPOSED FRAMEWORK: we propose to develop an appropriate data model to detect fraud and abuse ,customer relationship management decisions, to identify best treatment and best practices in health care resource management. we can apply CRISP-DM model to health care industry to improve overall customer satisfaction, usage
O’ Sullivan et al propose to incorporate formalized external expert knowledge in building a prediction model for asthma exacerbation severity for pediatrics patients in the emergency department[O’Sullivan08]. The secondary knowledge source identified as relevant for our retrospective asthma data Preschool Respiratory Assessment Measure(PRAM)asthma index. III. PROBLEMS STATED: Data Sets from various data sources[Stolba06].
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY) behavior can help to provide proactive initiatives to reduce overall cost and increase customer satisfaction. we may use different data mining algorithm to predict overweight and obese children in early age with the help of decision trees, association rules and Bayesian networks and SVM to predict obesity and overweight from database. we can use data mining techniques in a healthcare data warehouse. It combines online-analytical processing tool(OLAP)with proprietary data mining techniques. Algorithms of knowledge discovery in databases such as k-means,k-medoid clustering,nearest neighbor classification,data cleansing,post processing of sampling based data mining and k-join operation.Another use is clustering algorithm and decision support. Use of k-means clustering algorithm to analyze cervical cancer patients.use of hub and spoke model using artificial intelligence,dynamic load transfer,balancing and planning deployment of manpower and equipment. Application of CRISP-DM to primary health care system: 1.Define the problem(Business understanding). 2.Collecting the data. 3.Understanding the data. 4.creating database for data mining(data preparation). 5.Modeling of data. 6.creating the target dataset. 7.Data cleaning and preprocessing. 8.Selection for creating a data mining model. 9.Building a data mining model. 10.Evaluating the data mining model. V.FUTURE DIRECTIONS: 1. Data Storage and Access. 2. Data Collection and analysis. 3. Integration of models from other domains. 4. Theoretical and applied research. 5. Knowledge Extraction by integrating data from various sources and formats. 11. Deploy the data mining model. PROPOSED MODEL
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY)
VI.CONCLUSIONS: This paper proposes different data mining methods which are used in PHC.The proposed framework which works by combining CRISP-DM model with primary health care database management system.It will help in reducing the complexity of database and increasing its efficiency. REFERENCES: 1.WIKIPEDIA. 2.Implementing data mining in primary health care.By S. Pusphpalatha and Dr jagdesh Pandya. 3.Data mining for health care management.By jaideep shrivastava,kuo-wei hsu(national chengchi university ,Taiwan),Prasanna Desikan(center for healthcare innovation,Allina hospitals and Clinics USA). 4 A Decision-theoretic Approachto data mining.By yuval Elovici and Dan Bran. 5.Data mining industrial Applications.By waldemar wojcik and konrad gromaszek(Lublin university of technology ,Poland). 6.Musical data mining for distribution.(francois westermann,Damien Laigre). electronic music pachet,gert
7.promise and pitfalls of data mining:ethical issues.By William Seltzer(Fordham university). 8.From data mining to knowledge discovery in databases.By USSAMA FAYYAD, GREGORY PIATETSKYSHAPIRO, AND PADHARIC SMYTH. 9.A data mining and knowledge discovery process model.By Oscar marban(spain).
3.High performance data mining using the nearest neighbor join. 4.Describing pattern in health database. 5.Clustering algorithm and decision support. 6.Hub and Spoke using artificial intelligence. Kinds of data:”An EHR is an electronic version” of a patient’s medical history ,that is maintained by the health care provider over time and includes all of the key administrative clinical data relevant to that person’s care under a particular provider ,including demographics, progress notes, problems, medication, vital signs ,past medical history, immunizations, laboratory data, medical images and radiology reports. HL7-It is known as health level 7.It was developed to improve health informatics interoperability. It is working in the 7th layer, application layer, of the open systems interconnection model. EMR AND EHR: Electronic medical record contains patient record that is stored and retrieved locally in a standalone system used by a provider. Electronic health record contains patient information that is stored and retrieved in systems used by all
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY) providers who care about the patient. Examples of applying data mining on EMR/EHR: Ludwick and Doucette study is the adaptation of EMR in primary care. Cerrito works on EMR from an emergency department. Buczak et al works on disease surveillance on EMR. HEALTH LEVEL SEVEN (HL7):HL7 provides standards for interoperability that improve care delivery ,optimize workflow, reduce ambiguity and enhance knowledge transfer among health care providers. Example1:Patient referral data can vary extensively between cases because structure of patient referrals is upto general practitioner who refers the patient[Persson09]. Example2:Catley et al. use neural networks to predict preterm birth on a heterogeneous maternal population[Catley06]. Example3:”Traditional clinical based prognosis models were discovered to contain some restrictions to address the heterogeneity of the breast cancer”[Ahmad09]. Data from heterogeneous sources present challenges [KwiatKowska07]: 1. Sampling Bias:”clinical studies use diverse collecting methods, inclusion criteria and sampling methods”. 2. Refferal Bias: Data represent the preselected group with high prevelance of disease”. 3. Selection bias: clinical data sets include patients with different demographics”. 4. Method bias:”Predictors have varied specifications, granularities and precisions.” 5. Clinical spectrum bias:”Patient record represent varied severity of a disease and co-occurrence of other medical problems”. 6.Transcription and manipulation of patient records often results in high volume of noise and a high portion of missing values. IV. PROPOSED FRAMEWORK: we propose to develop an appropriate data model to detect fraud and abuse ,customer relationship management decisions, to identify best treatment and best practices in health care resource management. we can apply CRISP-DM model to health care industry to improve overall customer satisfaction, usage
O’ Sullivan et al propose to incorporate formalized external expert knowledge in building a prediction model for asthma exacerbation severity for pediatrics patients in the emergency department[O’Sullivan08]. The secondary knowledge source identified as relevant for our retrospective asthma data Preschool Respiratory Assessment Measure(PRAM)asthma index. III. PROBLEMS STATED: Data Sets from various data sources[Stolba06].
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY) behavior can help to provide proactive initiatives to reduce overall cost and increase customer satisfaction. we may use different data mining algorithm to predict overweight and obese children in early age with the help of decision trees, association rules and Bayesian networks and SVM to predict obesity and overweight from database. we can use data mining techniques in a healthcare data warehouse. It combines online-analytical processing tool(OLAP)with proprietary data mining techniques. Algorithms of knowledge discovery in databases such as k-means,k-medoid clustering,nearest neighbor classification,data cleansing,post processing of sampling based data mining and k-join operation.Another use is clustering algorithm and decision support. Use of k-means clustering algorithm to analyze cervical cancer patients.use of hub and spoke model using artificial intelligence,dynamic load transfer,balancing and planning deployment of manpower and equipment. Application of CRISP-DM to primary health care system: 1.Define the problem(Business understanding). 2.Collecting the data. 3.Understanding the data. 4.creating database for data mining(data preparation). 5.Modeling of data. 6.creating the target dataset. 7.Data cleaning and preprocessing. 8.Selection for creating a data mining model. 9.Building a data mining model. 10.Evaluating the data mining model. V.FUTURE DIRECTIONS: 1. Data Storage and Access. 2. Data Collection and analysis. 3. Integration of models from other domains. 4. Theoretical and applied research. 5. Knowledge Extraction by integrating data from various sources and formats. 11. Deploy the data mining model. PROPOSED MODEL
USE OF DATA MINING IN PRIMARY HEALTH CARE CENTERS By BARKHA BHADANA (MVN UNIVERSITY)
VI.CONCLUSIONS: This paper proposes different data mining methods which are used in PHC.The proposed framework which works by combining CRISP-DM model with primary health care database management system.It will help in reducing the complexity of database and increasing its efficiency. REFERENCES: 1.WIKIPEDIA. 2.Implementing data mining in primary health care.By S. Pusphpalatha and Dr jagdesh Pandya. 3.Data mining for health care management.By jaideep shrivastava,kuo-wei hsu(national chengchi university ,Taiwan),Prasanna Desikan(center for healthcare innovation,Allina hospitals and Clinics USA). 4 A Decision-theoretic Approachto data mining.By yuval Elovici and Dan Bran. 5.Data mining industrial Applications.By waldemar wojcik and konrad gromaszek(Lublin university of technology ,Poland). 6.Musical data mining for distribution.(francois westermann,Damien Laigre). electronic music pachet,gert
7.promise and pitfalls of data mining:ethical issues.By William Seltzer(Fordham university). 8.From data mining to knowledge discovery in databases.By USSAMA FAYYAD, GREGORY PIATETSKYSHAPIRO, AND PADHARIC SMYTH. 9.A data mining and knowledge discovery process model.By Oscar marban(spain).
