Patients Classification by Risk Using Cluster Analysis and Genetic Algorithms

Knowing a patient’s risk at the moment of admission to a medical unit is important for both clinical and administrative decision making: it is fundamental to carry out a health technology assessment. In this paper, we propose a non-supervised learning method based on cluster analysis and genetic algorithms to classify patients according to their admission risk. This proposal includes an innovative way to incorporate the information contained in the diagnostic hypotheses into the classification system. To assess this method, we used retrospective data of 294 patients (50 dead) admitted to two Adult Intensive Care Units (ICU) in the city of Santiago, Chile. An area calculation under the ROC curve was used to verify the accuracy of this classification. The results show that, with the proposed methodology, it is possible to obtain an ROC curve with a 0.946 area, whereas with the APACHE II system it is possible to obtain only a 0.786 area.

[1]  L. Iezzoni.  An introduction to risk adjustment. , 1996, American journal of medical quality : the official journal of the American College of Medical Quality.

[2]  L I Iezzoni,et al.  The importance of comorbidities in explaining differences in patient costs. , 1996, Medical care.

[3]  Moshe Sipper,et al.  Evolutionary computation in medicine: an overview , 2000, Artif. Intell. Medicine.

[4]  M. Aldenderfer.  Cluster Analysis , 1984 .

[5]  Pedro Larrañaga,et al.  Predicting survival in malignant skin melanoma using Bayesian networks automatically induced by genetic algorithms. An empirical comparison between different approaches , 1998, Artif. Intell. Medicine.

[6]  J. Zimmerman,et al.  ICU scoring systems allow prediction of patient outcomes and comparison of ICU performance. , 1996, Critical care clinics.

[7]  Robert E. Smith,et al.  Optimal clustering of power networks using genetic algorithms , 1994 .

[8]  D. E. Lawrence,et al.  APACHE—acute physiology and chronic health evaluation: a physiologically based classification system , 1981, Critical care medicine.

[9]  S. Lemeshow,et al.  A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. , 1993, JAMA.

[10]  T. Caliński,et al.  A dendrite method for cluster analysis , 1974 .

[11]  R. Dybowski,et al.  Prediction of outcome in critically ill patients using artificial neural network synthesised by genetic algorithm , 1996, The Lancet.

[12]  Jonathan E. Rowe,et al.  An evolutionary approach to constructing prognostic models , 1999, Artif. Intell. Medicine.

[13]  R. Evans.  Health care technology and the inevitability of resource allocation and rationing decisions. Part II. , 1983 .

[14]  M. Weinstein,et al.  Clinical Decision Analysis , 1980 .

[15]  J. Horbar,et al.  Predicting mortality risk for infants weighing 501 to 1500 grams at birth: A National Institutes of Health Neonatal Research Network report , 1993, Critical care medicine.

[16]  W. Knaus,et al.  APACHE II: a severity of disease classification system. , 1985 .

[17]  D. E. Goldberg,et al.  Genetic Algorithms in Search , 1989 .

[18]  D. E. Goldberg,et al.  Genetic Algorithms in Search, Optimization & Machine Learning , 1989 .

[19]  Emanuel Falkenauer,et al.  Genetic Algorithms and Grouping Problems , 1998 .