China
Africa
Explore chapters and articles related to this topic
Data Analysis Tools
Published in Jim Goodell, Janet Kolodner, Learning Engineering Toolkit, 2023
Erin Czerwinski, Tanvi Domadia, Scotty D. Craig, Jim Goodell, Steve Ritter
The goal of mining is to discover features, patterns, correlations, or anomalies of a data set that are useful for decision-making or further analysis, for example, which features of an instructional design correlate to desired outcomes. (The word “discover” implies that mining is more of a method for science than engineering, but especially when combined with other techniques these methods can be a valuable part of making data sets useful for answering specific engineering questions.) Mining techniques include, but aren’t limited to, the following: Relationship Mining: The goal of relationship mining is to discover variables that are related to one another within sets of many variables. Relationship mining is also key to big data reduction, that is, reducing the number of variables and size of the data set while keeping the relevant information needed to answer the question.Correlation Mining is used to find substantial linear correlations between variables. Remember, though: Correlation doesn’t mean causation!Causal Mining is used to infer causality, as in x causes y.Association Rule Mining is used to find simple if / then rules in the data set. For example, if a learner does x, they’re likely do y. This is useful for finding unexpected connections and generating hypotheses.Sequential Pattern Mining is used for finding patterns over time. For example, if a person reads a book on learning engineering now, how likely is it that person will take a class on education data mining later?Network Analysis is used for finding connections and their relative strengths, such as in social networks.
A systematic perspective on the applications of big data analytics in healthcare management
Published in International Journal of Healthcare Management, 2019
Sachin S. Kamble, Angappa Gunasekaran, Milind Goswami, Jaswant Manda
Garg et al. [83] developed an application that targeted at efficiently managing scarce healthcare resources at the hospitals and develop effective healthcare management policies. Based on the concepts of sequential pattern mining this application this application helps the managers and the policy makers with the useful healthcare information in real-time. Phillips-Wren et al. [96] studied the lung cancer patient for assessment of the utilization of healthcare resources by them. The demographic characteristics, socioeconomic factors, ethnic backgrounds, medical histories and access to healthcare resources were used as the predictive parameters. The findings were used to guide medical decision making and public policy.
Target association rule mining to explore novel paediatric illness patterns in emergency settings
Published in Scandinavian Journal of Clinical and Laboratory Investigation, 2022
Pradeep Kumar Dabla, Kamal Upreti, Divakar Singh, Anju Singh, Jitender Sharma, Aashima Dabas, Damien Gruson, Bernard Gouget, Sergio Bernardini, Evgenija Homsak, Sanja Stankovic
Association rule mining (ARM) [10–12] was initially used as a brute-force strategy in which all potential rules are enumerated first, and those rules that do not match the specified conditions are subsequently removed. Equally as effective in reducing database scanning is the equivalency transversal method (Eclat) described by Zaki et al. [13] for ARM, which required shifting information from a horizontal to a vertical layout. For the data mining community, ARM is a currently active area of research technique [14–16] for extracting found patterns [17,18]. When it comes to medicine, understanding disease and its biomarkers requires delving into the intricate web of hidden interactions between qualities (symptoms and diseases) [19–21]. Borah and Nath [22] used ARM to mine several risk variables of cardiovascular disease, hepatitis, and breast cancer, while Noguchi et al. [23–24] used it to identify adverse outcomes caused by drug effects. Sequential pattern mining with a gap constraint was utilized by Pokharel et al. [25] to discover commonalities between patients, such as death prediction and detection of symptomatic patients. One study used segment-based constraints in ARM to help health professionals look for patterns in sick people with dyspepsia [26], and another used ARM to obtain drug-symptom pairs for concept/relation extraction, both of which lend credence to the idea that the ARM technique can be used to capture clinical symptoms and uncover novel trends in data sets. ARM was also used to discover symptom pattern and overall symptom rules in COVID-19 patients [27]. However, to the best of our knowledge, use of association rule mining (ARM) is very rare in pediatric critical illness context. Therefore, in this study, we employ an illustrative method for extracting symptom rules based on a set of very straightforward pattern mining algorithms called ARM.