Title :
Assessing the robustness of algorithms for detecting transient ischemic ST segment changes
Author :
Jager, F. ; Moody, GB ; Divjak, S. ; Mark, RG
Author_Institution :
Fac. of Electr. & Comput. Eng., Ljubljana, Slovakia
Abstract :
Proposes a protocol and techniques to assess the robustness of algorithms for detecting transient ischemic ST segment changes. An algorithm is robust if it is not critically dependent on the distribution or noise content of input signals, or the exact values of its analysis parameters. The protocol includes bootstrap evaluation of algorithm performance, a noise stress test and a sensitivity analysis of the most important algorithm parameters. The authors illustrate these methods by a case study in which they assessed the robustness of their Karhunen-Loeve Transform based ischemic ST change detection algorithm.<>
Keywords :
electrocardiography; sensitivity analysis; signal detection; transforms; Karhunen-Loeve Transform-based algorithm; algorithm performance; algorithms robustness assessment; analysis parameters; bootstrap evaluation; input signals; most important algorithm parameters; noise content; noise stress test; transient ischemic ST segment changes detection; Algorithm design and analysis; Change detection algorithms; Detection algorithms; Karhunen-Loeve transforms; Noise robustness; Protocols; Sensitivity analysis; Signal analysis; Stress; Testing;
Conference_Titel :
Computers in Cardiology 1994
Conference_Location :
Bethesda, MD, USA
Print_ISBN :
0-8186-6570-X
DOI :
10.1109/CIC.1994.470208
