Metode umetne inteligence za modeliranje mehanizmov tremorjev

GROZNIK, VIDA (author), Bratko, Ivan (mentor)

Tremorji so ena najpogostejših motenj gibanja, ki jih povezujemo predvsem z različnimi boleznimi živčevja. Ker obstaja več kot 20 različnih tipov tremorjev, je z vidika pravilnega zdravljenja pomembno, da jih znamo med seboj razločevati. V disertaciji smo se osredotočili na razločevanje med parkinsonskim, esencialnim in mešanim tremorjem, saj se ti najpogosteje pojavljajo. V disertaciji smo se najprej lotili gradnje diagnostičnega modela za razločevanje med parkinsonskim, esencialnim in mešanim tremorjem na podlagi podatkov kliničnega pregleda, družinske anamneze in digitalne spirografije. Sam proces gradnje diagnostičnega modela je potekal z uporabo argumentiranega strojnega učenja. Ta nam je omogočil, da smo skozi proces elicitacije ekspertovega znanja (v našem primeru nevrologa) zgradili model, ki zajema trinajst pravil, ki so z medicinskega vidika smiselna. Sam postopek elicitacije znanja pa je pripomogel tudi k višji klasifikacijski točnosti končnega diagnostičnega modela v primerjavi z začetnim. V končnem diagnostičnem modelu so spirografski atributi nastopili v več kot polovici pravil. To nas je motiviralo, da smo izdelali model, ki temelji zgolj na podatkih digitalne spirografije. Za potrebe gradnje razumljivega modela smo najprej zgradili atribute. Namen atributov je bil zajem medicinskega znanja, ki se nanaša tako na spirografijo kot tudi na samo domeno. Na podlagi več kot 500 različnih atributov smo z uporabo logistične regresije zgradili končni model, ki z 90% klasifikacijsko točnostjo razločuje med preiskovanci s tremorjem in tistimi, ki tremorja niso imeli. Med samo gradnjo atributov smo želeli ugotoviti, kaj naši atributi zaznajo. Tako smo razvili metodo za vizualizacijo atributov na vrstah. Ta nam je pomagala pri gradnji atributov, poleg tega pa je uporabna tudi za vizualno razlago odločitev diagnostičnega modela. Samo vizualizacijo in posledično diagnostični model smo evalvirali s pomočjo treh neodvisnih ekspertov. Izkazalo se je, da sta tako sam diagnostični model kot tudi vizualizacija smiselna in zajameta medicinsko znanje. Končni diagnostični model je vgrajen v prosto dostopni mobilni aplikaciji ParkinsonCheck.

• Language: Slovenian
• Keywords: umetna inteligenca, vizualizacija atributov, tremor, digitalna spirografija, parkinsonova bolezen, esencialni tremor
• Work type: Doctoral dissertation
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2018
• PID: 20.500.12556/RUL-102233
• COBISS.SI-ID: 1537809091
• Publication date in RUL: 26.07.2018

Secondary language

• Language: English
• Title: Artificial Intelligence Methods for Modelling Tremor Mechanisms

Abstract

Tremors are one of the most common movement disorders primarily associated with various neurological diseases. Since there are more than 20 different types of tremors, differentiation between them is important from the treatment point of view. In the thesis, we focus on differentiation between three of the most common tremors: Parkinsonian, essential and mixed type of tremor. Our first goal was to build a diagnostic model for distinguishing between Parkinsonian, essential and mixed type of tremors, based on clinical examination data, family history and digital spirography. The process of building a model was carried out using argument-based machine learning which enabled us to build a decision model through the process of knowledge elicitation from the domain expert (in our case from a neurologist). The obtained model consists of thirteen rules that are medically sensible. The process of knowledge elicitation itself contributed to the higher classification accuracy of the final model in comparison with the initial one. In the final diagnostic model, attributes derived from the spirography were included in more than half of the rules. This motivated us to build a model based solely on the digital spirography data. For the needs of constructing an understandable model, we first built several attributes which represented domain medical knowledge. We have built more than 500 different attributes which were used in a logistic regression to construct the final diagnostic model. The model is able to distinguish subjects with tremors from those without tremors with 90% classification accuracy. During the process of attribute construction, we wanted to know what our attributes were detecting. Thus, we have developed a method for attribute visualisation on series. The method not only helped us with attribute construction, but it is also useful for visual interpretation of the diagnostic model's decisions. The visualisation method and consequently the decision model were evaluated with the help of three independent neurology experts. The results show that both the diagnostic model and the visualisation are meaningful and cover medical knowledge of the domain. The final diagnostic model is built into the freely available ParkinsonCheck mobile application.

• Keywords: artificial intelligence, attribute visualisation, tremor, digital spirography, Parkinson's disease, essential tremor