System architecture for support to medical doctor in decision making

Diplomsko delo opisuje arhitekturo sistema, ki pomaga zdravstvenim uslužbencem bolje razumeti različne bolezni in vpliv meritev nanje. Sistem deluje na podlagi znanj, ki jih črpa iz velikih količin kliničnih podatkov. V delu se uporabljajo podatki, pridobljeni iz FHIR strežnika in so umetno generirani. Glavno jedro arhitekture je spletna aplikacija, do katere lahko dostopa uporabnik. Njen osrednji namen je zagotovitev primernega okolja, kjer se lahko izvaja statistična analiza podatkov. Analiza se izvede z uporabo statističnih testov za testiranje domnev, kjer se ugotavlja statistična različnost posameznih meritev glede na določeno bolezen. Meritve so razvrščene glede na izračunane p-vrednosti. Dodatno se pri skalarnih spremenljivkah izvede še ROC analiza, s katero ugotavljamo napovedano ustreznost skalarnih meritev. Sistem je osnovan za potrebe zdravstvenih uslužbencev, ki skrbijo za diagnosticiranje bolnikov in si želijo boljši vpogled v delovanje bolezni. Arhitektura je sestavljena iz dveh ključnih procesov, in sicer vnos novih podatkov ter pridobitve rezultatov analize.The thesis describes the architecture of the system, which helps health professionals better understand the various conditions and an impact of observations on them. The system works on the basis of the knowledge from large amounts of clinical data. In this thesis, the data is being derived from FHIR servers and it is program generated. In the core of the architecture is a web application that can be accessed by the user. Its purpose is to provide an appropriate environment where statistical analysis of data can be performed. The analysis is performed by an algorithm that calculates the importance factors for individual measurements of patients and returns them according to their value. With the help of this system, the user has a better insight into various conditions. The system is designed for the needs of health care professionals who are responsible for diagnosing patients and appreciate a better overview of conditions. The architecture consists of two key processes, namely the input of new data and the computing of the analysis results