The content of this thesis, which is the result of a 6-month work at the Artificial Intelligence Laboratory of the University of Ljubljana (Slovenia), is based on the identification and analysis of clinical predictors for the study of Parkinson's disease (PD) progression in patients with a diagnosis not older than two years. In the Thesis, we will present the disease by describing the signs that characterize it, how it is diagnosed and the treatment currently available. We will continue with an in-depth analysis of the content of Parkinson's Progression Markers Initiative dataset. Since PD is primarily a motor disease, it is important to understand in advance the evolution and severity of these signs to prepare patients for a long coexistence with the disease (and possibly family members for the care). One of the main features of motor symptoms in PD is asymmetry. The severity of motor symptoms tends to be different on the right and left side of the body. As a part of this Thesis, we analyzed the relationship between asymmetry and disease progression. The results indicated that asymmetry is not an informative indicator for the progression of PD (although it is a reliable clinical marker to differentiate PD from other types of parkinsonisms). Once we excluded the asymmetry as a marker, we focused on the construction of some classifiers that, using other predictors that are recorded in routine visits, allow us to identify patients who will develop serious motor problems 3 years in advance. The obtained model (logistic regression classifier), is able, with only 6 clinical predictors, to recognize the 'Early Progressor' patients.
Analisi di predittori clinici per lo studio della progressione della malattia in pazienti affetti da Parkinson. Il contenuto della presente tesi, frutto di un lavoro di 6 mesi presso il Laboratorio di Intelligenza Artificiale dell'Università di Ljubljana (Slovenia), si basa sull'individuazione e analisi di predittori clinici per lo studio della progressione della malattia di parkinson in pazienti con una diagnosi non anteriore a due anni. Nella tesi presenteremo la malattia descrivendone i segni che la contradistinguono, come viene effettuata la diagnosi e le cure attualmente disponibili. Proseguiremo analizzando approfonditamente il contenuto del data set Parkinson’s Progression Markers Initiative. Essendo il Parkinson una malattia che presenta principalmente segni motori, è importante capire in anticipo l'evoluzione e la gravità di questi al fine di preparare i pazienti ad una lunga convivenza con la malattia (ed eventualmente i familiari nell'assistenza). Una delle principali caratteristiche nei sintomi motori dovuti al Parkinson è l'asimmetria. Infatti, l'entità dei sintomi motori tendenzialmente è diversa nel lato destro e sinistro deil corpo. La tesi prosegue analizzando la relazione tra l'asimmetria e la progressione della malattia. Il risultato ottenuto però ci ha mostrato come in realtà l'asimmetria non è un indicatore informativo per la progressione della malattia ( pur restando un importante marker per differenziare diversi tipi di patologie parkinsoniane). Una volta esclusa l'asimmetria come marker, ci siamo concentrati sulla costruzione di alcuni classificatori che, utilizzando altri predittori clinici che vengono rilevati in visite di routine, permettono di individuare con 3 anni di anticipo, pazienti che svilupperanno gravi problemi motori. Il modello ottenuto regressione logistica) è in grado di identficare, con solo 6 predittori clinici, i pazienti 'Early Progressor'.
Analysis of clinical predictors to study disease progression in De Novo Parkinson's patients
COTOGNI, MARCO
2019/2020
Abstract
The content of this thesis, which is the result of a 6-month work at the Artificial Intelligence Laboratory of the University of Ljubljana (Slovenia), is based on the identification and analysis of clinical predictors for the study of Parkinson's disease (PD) progression in patients with a diagnosis not older than two years. In the Thesis, we will present the disease by describing the signs that characterize it, how it is diagnosed and the treatment currently available. We will continue with an in-depth analysis of the content of Parkinson's Progression Markers Initiative dataset. Since PD is primarily a motor disease, it is important to understand in advance the evolution and severity of these signs to prepare patients for a long coexistence with the disease (and possibly family members for the care). One of the main features of motor symptoms in PD is asymmetry. The severity of motor symptoms tends to be different on the right and left side of the body. As a part of this Thesis, we analyzed the relationship between asymmetry and disease progression. The results indicated that asymmetry is not an informative indicator for the progression of PD (although it is a reliable clinical marker to differentiate PD from other types of parkinsonisms). Once we excluded the asymmetry as a marker, we focused on the construction of some classifiers that, using other predictors that are recorded in routine visits, allow us to identify patients who will develop serious motor problems 3 years in advance. The obtained model (logistic regression classifier), is able, with only 6 clinical predictors, to recognize the 'Early Progressor' patients.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/12017