Paper Info
Title
The Structured Development Method Senn For Feedforward Neural Networks And The Application To The Multimedia Patient Record In The Bavarian Health Net
Abstract
Within the 'Bavarian Health Net' innovative applications will be combined using new communication techniques. This will lead to a standardized quality of medical services in all parts of Bavaria and can significantly reduce distance-induced costs and problems. It provides a basis for modern concepts such as shared care, distributed multi-media medical records, and protocol-based decision support.Within the 'Intensive Medicine Framework' where high level medical computer science projects are founded, knowledge based documentation systems are integrated into the 'Bavarian Health Net'. Beside protocols and other techniques like Causal Probabilistic Networks (CPNs) also artificial neural networks (ANNs) are used to implement knowledge bases for complex decision tasks. In the medical domain there is often only a small number of examples available for the setup of a knowledge base and much care has to be taken of quality assurance. To meet these requirements the structured development method for supervised feed forward ANNs (SENN System Engineering for Neural Networks) that have to solve approximation or classification tasks, was developed.The SENN method deals with the development of glass box systems which can be analysed and understood by an expert of the application domain. The development of adequate representations and encodings of the information, as well as the set-up, training, verification and validation of the ANN in a quality assured process are covered by the SENN method and are supported by the related tools.Quality and efficiency of results obtained from evaluation studies in Diabetes care are promising for generating the ANN applicable for decision support, Further applications in the ByMediks ophthalmology subproject will show the usefulness of the approach, its possibilities and necessary enhancements on the way to a commonly tool.
Year
DOI
Venue
1996
10.1007/978-0-387-34979-4_19
ADVANCED IT TOOLS
Keywords
Field
DocType
Medical Records, data analysis, neural networks, SENN method, development tool
Feedforward neural network,Computer science,Decision support system,Medical record,Shared care,Artificial neural network,Multimedia
Conference
Citations 
PageRank 
References 
0
0.34
2
Authors
4
Name
Order
Citations
PageRank
Rolf Engelbrecht100.34
Thomas Waschulzik263.27
Wilfried Brauer3969299.36
Matthias Scherf4294.08