The Electronic Health Records are the systematized collection of patient and population electronically-stored health information in a digital format. Currently, an increasing number of hospital structures are upgrading their internal recording processes to an electronic based system, instead of the classical papery one. With these systems, it is built an internal database that guarantees the access at the patient's information with absolute reliability and speed from all the users (doctors, nurses and hospital staff). Despite their increasing diffusion, they represent a vast and underused data resource for biomedical research, mainly due to the lack in tools to process them adequately. Indeed, beside standard structured data in tabular format, Electronic Health Records contains unstructured or semi-structured data, mainly in text form. This work represents a pilot study focused on heart diseases and aimed to identify the most suitable Data Mining tools as well as assess the feasibility and the criticality in the use of these techniques on complex and semi-structured clinical data. The information coming from Electronic Health Records is also compared with classical Administrative Databases, collected from hospital structures and institutions mainly for governance purposes. These two data repositories, Electronic Health Records and Administrative Data, represent a complementary and non-overlapping source of information for research in disease treatment, preventive healthcare, adherence to prescriptions and hospital administrative governance.
La Cartella Clinica Elettronica rappresenta la collezione sistematica, in formato digitale, di informazioni sulla salute di un singolo individuo o di una popolazione. Attualmente un numero crescente di strutture ospedaliere sta convertendo i propri processi interni di registrazione cartacea in sistemi elettronici. Con lo sviluppo di questi sistemi viene costruito un database interno che garantisce l'accesso alle informazioni riguardanti il paziente con assoluta affidabilità e velocità da parte di tutti gli utenti (dottori, infermieri e staff dell'ospedale). Nonostante la sua esponenziale diffusione, la Cartella Clinica Elettronica rappresenta una vasta e sottoutilizzata risorsa per la ricerca biomedica, principalmente a causa della mancanza di strumenti adeguati per processare le informazioni in essa contenute. Infatti, accanto a dati strutturati in formato tabulare, la Cartella Clinica Elettronica contiene dati non strutturati o semi-strutturati, principalmente sotto forma di testo. Questo lavoro rappresenta uno studio pilota ristretto al caso di malattie cardiologiche, allo scopo di identificare gli strumenti di Data Mining più adatti in un contesto di dati clinici complessi e semistrutturati e studiarne la fattibilità e le criticità. L'informazione proveniente dalla Cartella Clinica Elettronica è poi confrontata con i dati amministrativi raccolti dalle strutture ospedaliere e dalle istituzioni, principalmente per scopi governativi. Questi due archivi, la Cartella Clinica Elettronica e i dati amministrativi, forniscono informazioni complementari e non sovrapponibili per la ricerca nell'ambito dell'efficacia delle terapie, della prevenzione, dell'aderenza alle prescrizioni e della gestione amministrativa delle strutture ospedaliere.
MinEHR : mining electronic health records. A proposal for analyzing medical records in heart diseases with data mining techniques
BROCCA, MARTINA
2018/2019
Abstract
The Electronic Health Records are the systematized collection of patient and population electronically-stored health information in a digital format. Currently, an increasing number of hospital structures are upgrading their internal recording processes to an electronic based system, instead of the classical papery one. With these systems, it is built an internal database that guarantees the access at the patient's information with absolute reliability and speed from all the users (doctors, nurses and hospital staff). Despite their increasing diffusion, they represent a vast and underused data resource for biomedical research, mainly due to the lack in tools to process them adequately. Indeed, beside standard structured data in tabular format, Electronic Health Records contains unstructured or semi-structured data, mainly in text form. This work represents a pilot study focused on heart diseases and aimed to identify the most suitable Data Mining tools as well as assess the feasibility and the criticality in the use of these techniques on complex and semi-structured clinical data. The information coming from Electronic Health Records is also compared with classical Administrative Databases, collected from hospital structures and institutions mainly for governance purposes. These two data repositories, Electronic Health Records and Administrative Data, represent a complementary and non-overlapping source of information for research in disease treatment, preventive healthcare, adherence to prescriptions and hospital administrative governance.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi_MartinaBrocca.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
1.63 MB
Formato
Adobe PDF
|
1.63 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/152426