Clinical Data Mining for Physician Decision Making and Investigating Health Outcomes: Methods for Prediction and Analysis

Citation

Bacao, F. (2011), "Clinical Data Mining for Physician Decision Making and Investigating Health Outcomes: Methods for Prediction and Analysis", Online Information Review, Vol. 35 No. 4, pp. 685-686. https://doi.org/10.1108/14684521111162025

Leveraging data mining methods and techniques to turn data into insight has gained quite a momentum in recent years. This notion is quite powerful and promises to change numerous aspects of our lives, and especially the way organisations are managed and how they learn from their experience. The idea that organisations can tap into massive databases and learn from their past experience is quite appealing and has potential to change the way business is done in many industries. Healthcare is an obvious example. Considering the money that most developed countries now spend in healthcare, the challenges that lie ahead (such as widening the access to medical care, an aging population, the need to maintain balanced healthcare budgets), the wealth of data available and the new tools provided by data mining, the healthcare industry surely will change.

This collection was planned around two major assumptions: the reader is interested in exploring the healthcare data available in the USA, and the reader has access to SAS Enterprise Guide (preferably also to SAS Enterprise Miner). If one does not qualify on both counts, this book can be ignored: it is based entirely on the exploration and analysis of data from the Medical Expenditure Panel Survey (MEPS), the National Inpatient Sample, the Medpar data, and the Centers for Medicare and Medicaid, based on SAS Enterprise Guide. The examples are well explained, and the reader is walked through the pre‐processing and analysis tasks in great detail.

The book contains 17 chapters, which can be divided into two major parts. The first part contains eight chapters and focuses on basic pre‐processing tasks. As in the rest of the book, the pre‐processing procedures are closely related to the datasets used throughout the book, and include tasks as extracting subsamples of the datasets, the computation of basic statistics and simple database operations such as the creation of new observational units (e.g. join all the prescriptions for each patient).

The second part (Chapters 9 through 17) deals with analysis tools and provides different examples of how knowledge can be extracted from the datasets used. Here the emphasis is not on the tools themselves but rather on the type of problem that can be addressed with the available datasets. While most books on data mining highlight the tools and provide examples of their application, in this book the starting point tends to be a specific healthcare issue, such as patient compliance or patient severity indices, and then the authors propose a specific methodology to deal with it.

If you are a researcher or healthcare professional interested in exploring MEPS, the National Inpatient Sample or Medpar databases, have access to SAS Enterprise Guide and Enterprise Miner, and are interested in learning about how data mining tools can be applied to extract knowledge from these specific databases, then you should consider buying this book. On the other hand, if you are interested in cutting edge research on data mining applications in healthcare or in learning about data mining tools and techniques to apply to your professional/research problems, you should probably look elsewhere. This is essentially a practical book, and its strict adherence to the databases and problems proposed by the authors makes it difficult for the inexperienced reader to generalise the knowledge acquired to other settings and problems.