The aim of the study was to create own "models" of organism's response in selected life-threatening conditions (from the Emergency Medicine field) based on a set of simple biochemical and clinical parameters thus producing a diagnostic tool. Material consisted of 162 patients hospitalized in selected Clinics of University Hospital of Collegium Medicum of Jagiellonian University in Krakow from 1st May 2001 do 1st January 2005, belonging to the groups of: multiple injuries, severe acute pancreatitis, upper GI bleeding, acute myocardial infarction, subarachnoid hemorrhage and control.. Set of simple nineteen clinical and biochemical parameters was evaluated in following days of treatment (first, third and seventh) in every patient. Discrimination analysis and two types of decision trees were used for statistical analysis. Then discrimination functions were calculated allowing assignment of the new observation (new patient) to one of the created "models". In every studied day, selected parameters proved to discriminate groups properly, but for each day different parameter turned out to be the most effective. Discrimination analysis constituted the best statistical technique for proper classification of the patients with overall ratio of proper classifications equating 86,83%. Analysis of decision trees proved to be less effective method. Abovementioned results indicate that pre ; sented models can be successfully used in diagnostic process of the patients, especially in emergency cases, for the set of clinical and biochemical parameters used is simple and easy to obtain even in non-specialized centers.
Jul 18, 2022
Nov 21, 2012
12
0
http://dl.cm-uj.krakow.pl:8080/publication/1223
Edition name | Date |
---|---|
ZB-102158 | Jul 18, 2022 |
Budzyński, Piotr
Chmiel, Izabela
Cieszkowski, Jakub
Konarska-Bajda, Katarzyna
Sendur, Paweł
Sporek, Mateusz
Dembiński, Marcin
Zasada, Jakub Paweł