|
|
This is only a preview of the paper
Click here to register and get the full text.
Existing members click here to login
|
|
|
Data mining in databases has been adopted in health care field because of the major characteristics of data mining, prediction and discovery of useful information in large scale of databases. However, some technological weaknesses such as imprecise predictions, improper usages of models, and some technical mistakes in processing of data reduce the appreciation of helpfulness of data mining. A survey conducted to the graduate students shows the attitudes toward employment of data mining in health care and several common practical errors in data processing. In conclusion, researchers can approach more optimal adoptions of data mining provided meliorated algorithms for the computerized system.
Introduction
Data mining in health care databases is becoming an important resource for health care management. The knowledge discovery system is used for various analyses and research projects to help improve patient care while looking for ways to decrease health care costs. Data mining in health care documents patient observations, treatments and services provided, and assists clinicians to address the needs of patients. The findings from data mining are used to improve treatment effectiveness and to evaluate the variability of health care and the associated effects on patient outcomes such as adverse effects of drug, diabetic mortality, and death rate in damage control. In short, data mining in health care facilitates researchers obtaining valuable information from large medical databases. However, users of data mining may encounter some technological weaknesses when employing algorithms as computerized assistants. First, the reliability of data mining is debatable because the association rules adopted to predict trends or discover useful patterns may not be accurate enough. Moreover, processing data using data mining in databases can also contain certain problems such as errors, missing data, and variant data. ... /~} attitudes to adopting data mining in health care and the technical limitations in processing data. The results show that they would consider data mining a perfect tool offering valuable references in health care only if its weaknesses are improved.
Literature Review
Data mining has been used in health care because its two major characteristics, prediction of trends and discovery of hidden patterns, greatly facilitate in development in disease treatments, biotechnology, nursing, and other fields in health care. Scholars have focused on the prediction ability of data mining in huge databases. ... studied some statistical methods and techniques of data mining to predict the mortality of damage control laparotomy (2000). ... continued this study by designing the prognostic models in order to make the best use of scarce medical resources and employing feature mining and machine learning techniques to predict the results of damage control of severely injured patients (2001). ... To broaden the domain of adoption of data mining in health care, scholars have been devoted to research pathways in clinics (Lin 2001). ... In conclusion, prediction in data mining has been studied in clinical respects such as damage control and pathways. ... Some scholars discussed the data of diabetic patients and identified the relationship between observations and mortalities by adopting knowledge discovery in databases (KDD) (Richards et al. ... For example, data envelopment analysis (DEA) and artificial neural networks (ANN) are technologically improved methods to check breast caner (Pendharkar et al. ... However, some data are not so concrete that they are only described as numbers or grouped in a certain category. ... It is one approach of data mining dealing with vague distributes. In conclusion, data mining in health care forecasts the trends of diseases or damages mostly based on applying a useful association rule.
Moreover, many scholars are interested in the applications of data mining to the study of genomes. The analysis of RNA may meet difficulties because the data for bioassay always are micro-represented. Zweiger focused on the assays of RNA and KDD (knowledge discovery in databases) applications on analyzing biological data (1999). ... specified their research on ERGL as an instance for EST data mining (2001). ...
There are still lots of technological applications of data mining in other aspects of health care. One important area is the use of data mining in health care management. Knowledge discovery is needed in a great and increasing database in health insurance (Delesie & Croes 2000). The computerized application technologically reduces the consumption of time and human resources so that medical providers will have more energy to offer better quality in health care. Furthermore, there is application software of data mining developed for biotechnological use. ... For example, hierarchical decision models, such as DEX and HINT, are employed as data mining methods in health care today (Bohanec, Zupan, & Rajkovic 2000). ... Thus, data mining is taught in nursing informatics education in order to raise the performance of treatments in hospitals (Kokol 1999). ... To summarize, the purpose of employing the applications, software, and models is to assist professional users in health care to provide better treatments to patients and make progress in medical research.
Data mining obtains its name as it searches for valuable information in a large database. It is really a beneficial tool supporting studies in numerous fields in health care. However, we are not able to obtain the reliable results unless the accuracy of transformation and the dependency of application of data mining models are trustworthy. Although data mining seemingly can assist decision makers in health care, doctors and experts should be cautiously seeking and developing insights on symptoms and causes to provide more proper treatments. In fact, there are still some vital and inevitable technological weaknesses in prediction of the trends, discovery of hidden patterns, and process of the data in the system.
Method
In order to examine possible weaknesses encountered when putting data mining into practice, I surveyed graduate students majoring in Information Systems. The reason for choosing the students is that they have certain backgrounds in data management and might become information systems experts or decision makers in the future. ...
My hypothesis in the research is that people would doubt the dependence of data mining employed into the practice of health care because of the existing technical problems. ... The questions are categorized according to possible weaknesses in 1) prediction of trends, 2) discovery of buried information, and 3) processing of data. The reason I chose a questionnaire as the research tool is that the results could reveal the weaknesses of the applications of data mining adopted in health care from employees~{! ... The data were easy to collect in a short amount of time and the results could be quickly tabulated so that the results can be achieved expeditiously for analysis. ... The students in the class have already taken certain prerequisite courses related to data management before. ... Rank of How Valuable Are Patterns Discovered by Data Mining in Health Care
Invaluable 1 2 3 4 5 6 Valuable
Percentage 0% 0% 0% 20% 20% 20% 30% 10%
The respondents have taken an average of 2.7 courses related to data management. ... Table 1 shows 80% of the respondents consider the patterns found by data mining valuable. ... Basically, most respondents admitted that data mining is useful in digging out the hidden treasure of information in databases. ... Rank of Agreement of a) KDD Has Sufficient Intelligence to Accurately Forecast in Health Care, and b) Classification of Diseases and Cures Is Complex. ... Eight of the respondents think classifying information, as a method of predictive function, is a highly complex process for data mining. ... /~} reactions to the two major characteristics of data mining, prediction of tendency and discovery of concealed valuable information. Although the respondents basically agree that patterns discovered by data mining are valuable, they do not think that KDD has sufficient intelligence to deal with the complex classification. Most likely, the respondents will positively appreciate the usefulness of data mining applied to health care only if the intelligence mechanism in the system is reliable to a certain extent.
Approximate Word count = 6284
Approximate Pages = 25.1
(250 words per page double spaced)
|
|
|
|
|
|