%0 Journal Article
%T Customized Prediction of Short Length of Stay Following Elective Cardiac Surgery in Elderly Patients Using a Genetic Algorithm
%A Joon Lee
%A Sapna Govindan
%A Leo A. Celi
%A Kamal R. Khabbaz
%A Balachundhar Subramaniam
%J World Journal of Cardiovascular Surgery
%P 163-170
%@ 2164-3210
%D 2013
%I Scientific Research Publishing
%R 10.4236/wjcs.2013.35034
%X <b><span>Objective:</span></b><span> To develop a customized short LOS (&lt;6 days) prediction model for geriatric patients receiving cardiac sur</span><span>gery, using local data and a computational feature selection algorithm.</span><span> </span><b><span>Design:</span></b><span> Utilization of a machine learning algo</span><span>rithm in a prospectively collected STS database consisting of patients who received cardiac surgery between January 2002 and June 2011.</span><span> </span><b><span>Setting: </span></b><span>Urban tertiary-care center.</span><span> </span><b><span>Participants: </span></b><span>Geriatric patients aged 70 years or older at the time of cardiac surgery.</span><span> </span><b><span>Interventions: </span></b><span>None.</span><span> </span><b><span>Measurements and Main Results: </span></b><span>Predefined morbidity and mortality events were collected from the STS database. 23 clinically relevant predictors were investigated for short LOS predic</span><span>tion with a genetic algorithm (GenAlg) in 1426 patients. Due to the absence of an STS model for their particular sur</span><span>gery type, STS risk scores were unavailable for 771 patients. STS prediction achieved an AUC of 0.629 while the GenAlg achieved AUCs of 0.573 (in those with STS scores) and 0.691</span><span> </span><span>(in those without STS scores). Among the pa</span><span>tients with STS scores, the GenAlg features significantly associated with shorter LOS were absence of c<span>ongestive heart failure (</span>CHF) (OR</span><span> </span><span>=</span><span> </span><span>0.59, p</span><span> </span><span>=</span><span> </span><span>0.04), aortic valve procedure (OR</span><span> </span><span>=</span><span> </span><span>1.54, p</span><span> </span><span>=</span><span> </span><span>0.04), and shorter cross clamp time (OR</span><span> </span><span>=</span><span> </span><span>0.99, p</span><span> </span><span>=</span><span> </span><span>0.004). In those without STS prediction, <span>short LOS was significantly correlated with younger age </span>(OR</span><span> </span><span>=</span><span> </span><span>0.93, p</span><span> </span><span>&lt;</span><span> </span><span>0.001)<span>, absence of CHF </span>(OR</span><span> </span><span>=</span><span> </span><span>0.53, p</span><span> </span><span>=</span><span> </span><span>0.007)<span>, no preoperative use of beta blockers </span>(OR</span><span> </span><span>=</span><span> </span><span>0.66, p</span><span> </span><span>=</span><span> </span><span>0.03)<span>, and shorter cross clamp time </span>(OR</span><span>
%K Cardiac Surgery
%K Elderly
%K Length of Stay
%K Risk Prediction
%K Genetic Algorithm
%U http://www.scirp.org/journal/PaperInformation.aspx?PaperID=37146