A distributed environment for the integration of multiple high-performance decision support systems into clinical workflow

Article type: Research Article

Authors: Androulidakis, Aggelos^{a; *} | Nielsen, Anders Dencker^b | Prentza, Andriana^a | Koutsouris, Dimitris^a

Affiliations: [a] Biomedical Engineering Laboratory, School of Electrical and Computer Engineering, National Technical University of Athens (NTUA), GR-157 73 Zografou, Athens, Greece | [b] Center for Model-based Medical Decision Support, Aalborg University, Denmark

Correspondence: [*] Corresponding author. Tel.: +30 210 7722286; Fax: +30 210 7722431; E-mail: androu@biomed.ntua.gr.

Abstract: Current studies conclude that clinical decision support systems can help reduce serious medical errors. The importance of Causal Probabilistic Networks (CPNs) for constructing such systems is already well-known. However, the computational complexity of probabilistic inference, which results in unacceptably high response times, can hinder acceptance and integration into clinician workflow. This paper investigates the optimization and parallelization potential of complex CPN-based medical decision support systems and evaluates the results of implementing a parallel, high performance version of an existing decision support system concerning proper antibiotic treatment therapy. Furthermore, it discusses distributed computing techniques for making multiple high performance decision support systems available at the time and location of decision making, by exploiting computing resources residing inside, as well as outside the hospital walls optimally.

Keywords: Decision support systems, parallel algorithms, causal probabilistic networks, distributed computing

DOI: 10.3233/THC-2006-14303

Journal: Technology and Health Care, vol. 14, no. 3, pp. 157-170, 2006