The management of critical information in the healthcare sector has over the years proved to be rather tedious exercise. The sector is allocated a significant proportion of federal budget and if efforts to make service delivery more efficient are not made, the cost of providing quality healthcare to the population will only skyrocket. Healthcare providers believe that integration of information technology in managing healthcare will go a long way in enhancing competitiveness of the sector (Kraft, 2002). This is because organizations will be able to keep a clear record of individual patients and thus making it easy to track progress in different settings, assess the quality of service provided by monitoring results, and to get timely supply of information for effective financial planning. The paper explores two data mining techniques used in healthcare management. It will assess the benefits, limitations, and risks of each technique. Practical example under which each technique would be used effectively by healthcare organizations will also be discussed.
Data Mining Techniques
Artificial Neural Networks (ANN)
The healthcare sector in most countries has been struggling to address the challenge of reducing the cost of service provision without compromising quality (Kraft, 2002). One of the most promising responses to these pressures is use of healthcare management information systems for making critical decisions and enhancing the management of knowledge. Virtually all healthcare facilities have loads of data accumulated from patients that have ever visited these facilities. A purposeful analysis of the large amounts of data on specific health problems will yield important information that will facilitate evidence-based decision making (Kraft, 2002). Advanced analysis of the data will ultimately lead to the acquisition of useful knowledge.
To gather relevant information, healthcare organizations have embraced the use of data mining techniques. One of the techniques is the use of the artificial neural networks (ANNs). Generally, this technique adopts the way the brain operates when it comes to problem solving and applying the knowledge in information technology (Kraft, 2002). Although the computer can perform some tasks like computations faster than human beings, the human brain can still perform excellent in other complex situations like image recognition or processing, and speech. The study of the brain has revealed that when dealing with complex tasks, it breaks them into manageable components and other massive simultaneous processes. ANNs technique uses the two concepts in designing computer operations (Kraft, 2002). The pattern of operation of ANNs is the input-output model/network. The network has an intermediate hidden layer between the input and output. After processing, the output will send the information to the designated recipient, even to another information system.
ANNs can be of different types: prune and dynamic networks, multilayer perception, and the radial basis function network. They are useful in developing predictive models especially for length of stay (LOS) and predicting the possible out of treating people spinal cord injury. However, some types of ANNs are known to be too slow when being used causing delayed output. Computerization of available nursing data needs to be done using proper language standardization to minimize repeated cleaning data.
Decision Tree Analysis
The healthcare sector is from time to time faced with numerous challenges of making decisions before taking action. Decision Trees are useful data mining techniques used in healthcare management. This technique involves the description of data before any meaningful decision is made. In essence, data is first classified into various possibilities which in turn act as useful input upon which decisions may be made (Yokota, Desouza, & Androwich, 2004). Decision tree analysis uses a model that is designed to aid in predicting outcomes from a huge database with specified input variables. In healthcare management, trees refer to a set of mathematical and computing techniques. They can be used to describe, categorize, and even make generalizations from a given set of information.
There are two types of decision trees employed during data mining; classification tree analysis and regression tree analysis. Regression tree analysis refers to a situation where the expected outcome is an actual number like the number of days that patient will be hospitalized, or the price of a property (Yokota, et al., 2004). On the other hand, classification tree analysis refers to a case where the anticipated outcome falls in the category under which the data belongs. In some instances, the two types may be combined depending on the nature of the problem under consideration.
Decision tree technique is associated with a number of benefits. The technique is not too complex to understand and draw conclusions from the observations. Unlike ANNs, this technique does not require data cleaning and other standardization procedures. It accepts both numerical data (regression tree analysis) and categorical data (classification tree analysis). The model is highly reliable since it can be verified using statistical tests. However, the technique has some limitations. New users of the technique may come up with a complex tree that makes it even harder to make predictions from the datasets (Yokota, et al., 2004). Technical language is often used in this technique making it harder for learners. Many critics have also pointed out that information obtained can be highly biased depending with the number of levels.
The paper has briefly discussed the importance of effective management of information in the healthcare organizations. It has highlighted the need to introduce information technology managing data in hospitals. Two data mining techniques have been elaborated including some of their advantages and disadvantages. Given the large amount of data that can be found in a given hospital, therefore, healthcare management information systems through the use of data mining techniques can help in predicting outcomes of any treatment.
Kraft, M. R. (2002). Data mining in healthcare information systems: case study of veterans’ administration spinal cord injury population. Journal of Information System Sciences, 10 (4): 292-298
Yokota, F., Desouza, K. C. & Androwich, I. (2004). Value of information literature analysis: a review of applications in health risk management. Medical Decision Making, 24 (3): 287-298