- Title
- Select and Test (ST) algorithm for medical diagnostic reasoning
- Creator
- Fernando, D. A. Irosh P.; Henskens, Frans A.
- Relation
- 7th IEEE/ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing (SNPD 2016). Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing [presented in Studies in Computational Intelligence, Vol. 653] (Shanghai, China 30 May - 1 June, 2016) p. 75-84
- Publisher Link
- http://dx.doi.org/10.1007/978-3-319-33810-1_6
- Publisher
- Springer
- Resource Type
- conference paper
- Date
- 2016
- Description
- This paper presents an enhanced version of the ST Algorithm, which has been published previously (Fernando and Henskens , Polibits 48:23–29, 2013 [17]). The enhancements include improved presentation of the knowledgebase, a special bipartite graph in which the relations between a clinical feature (e.g. a symptom) and a diagnosis represent two posterior probabilities (probability of the diagnosis given the symptom, and the probability of the symptom given the diagnosis). Also, the inference step, induction, which estimates the likelihood (i.e. how likely each diagnosis is) has been improved using an orthogonal vector projection method for calculating similarities. The algorithm has been described in a more mathematical form (e.g. using sets rather than the linked lists that were used in its earlier version) mostly as a manipulation of sets by adding and removing elements that are in the bipartite graphs. The algorithm was implemented in Java, and a small knowledge base has been used in this paper as an example for illustration purpose only. The focus of this paper is on the algorithm, which is intended to give a theoretical proof that medical expert systems are achievable; the design and implementation of a knowledgebase that can be practically useful for clinical work, was not within the scope of this work.
- Subject
- ST algorithm; medical expert systems; medical diagnosis
- Identifier
- http://hdl.handle.net/1959.13/1343364
- Identifier
- uon:29145
- Identifier
- ISBN:9783319338095
- Language
- eng
- Reviewed
- Hits: 830
- Visitors: 812
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|