Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Machine learning paradigms : advances in data analytics
Tsihrintzis G., Sotiropoulos D., Jain L., Springer International Publishing, New York, NY, 2019. 370 pp. Type: Book (978-3-319940-29-8)
Date Reviewed: Jan 28 2019

I have been in the eHealth field for several years, teaching and working with data mining and machine learning and especially topics related to ambient assisted living (AAL). We have developed various applications that can detect so-called activities of daily living (ADLs) like drinking, tumbling, washing, phoning, and so on. Surprisingly enough, this can be done quite well with models produced by data mining algorithms, especially deep learning networks. This book, edited by Tsihrintzis et al., grabbed my attention because it contains interesting work on machine learning in the medical domain. I mention this now because I mainly focused on those contributions when reading the book (and thus writing this review).

Chapter 2, by Dessi et al., describes a recommender system that analyzes patient records and helps doctors find similar reports for a given patient. It discusses the variables problems associated with medical documents (unstructured and so on) and then uses natural language processing (NLP)-based technologies to analyze the records. This is a classical recommender problem, identifying similar records--very similar to Amazon recommendations, but based on medical record content. Its core is the content analysis module that converts the unstructured input into a structured representation. The authors apply hierarchical and k-means clustering using the silhouette width measure based on the structured documents. Not surprisingly, the quality of the recommendation depends on the quality of the NLP part of the system, as the authors state--following the old wisdom: garbage in, garbage out. Hierarchical clustering outperforms k-means clustering. Finally, the authors conclude that recommender systems are useful in the medical domain, but their usefulness depends on the (open-source) availability of medical records.

Another chapter (4) that I found quite interesting deals with data mining for predicting heart disease. The early detection of heart disease symptoms could help save many lives (about 26 million people are affected by heart diseases worldwide). This has drawn interest from companies, for example, the Apple Watch added support for heart anomalies. The authors give a concise review of the current literature in the area and especially in sequential mining. They discuss some extensions to sequential mining and the application of those rules. However, the chapter is missing guidelines on when to use each strategy.

The book contains several other interesting examples of machine learning in social science, network analysis, and file forgery detection. Overall, it is an interesting collection of machine learning applications across multiple domains. It may be of interest to readers working in one of the discussed areas.

Reviewer:  K. Waldhör Review #: CR146403 (1904-0108)
Bookmark and Share
 
Learning (I.2.6 )
 
 
Content Analysis And Indexing (H.3.1 )
 
Would you recommend this review?
yes
no
Other reviews under "Learning": Date
Learning in parallel networks: simulating learning in a probabilistic system
Hinton G. (ed) BYTE 10(4): 265-273, 1985. Type: Article
Nov 1 1985
Macro-operators: a weak method for learning
Korf R. Artificial Intelligence 26(1): 35-77, 1985. Type: Article
Feb 1 1986
Inferring (mal) rules from pupils’ protocols
Sleeman D.  Progress in artificial intelligence (, Orsay, France,391985. Type: Proceedings
Dec 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy