The evaluation of Medical applications is difficult for common users because they usually have poor specialized medical knowledge. Hence providing these common users with the possibility to understand and to judge the robustness of an app in matter of reliability of its contents becomes crucial. The keyword “pharma” has been used for the search of medical apps on iMedicalApp.com and, thanks to this search, 48 apps have been found within the appstores iTunes, GooglePlay, BlackBerry and Windows Phone. A result set of 60 descriptions came out. We searched and characterized the medical terms contained in the descriptions. These terms have been compared to the medical words from the Consumer Health Vocabulary (CHV), which is a medical dictionary consisting of common medical terms. In addition, the CHV has been connected to specialist medical concepts explained through the Unified Medical Language System (UMLS). The application we developed is composed of a database to collect and manage the apps data and a user interface to interact with the application. The analysis of the percentage of medical terms out of the total words got in the apps descriptions allowed us to create a characterization index. Five classes compose this index. For the result set “pharma” apps, the lower class (2,82-10,84%) holds the 13% of descriptions, the other ones: 10,84-18,86% the 43,3%; 18,86-26,89% the 33,3%; 26,89-34,91% the 8,3% and 34,91-42,93%, the 1,6%. We can gather that 90% of the total apps are in the first three characterization classes (less than 26.89% of medical terms). From this, it is reasonable to suppose that the most of “pharma” apps has poor medical contents. This project is the first approach for the assessment of medical apps starting from their descriptions. It is based on an objective assessment of applications provided with the execution of computational and automatic processes and it is not depending on any subjective thoughts of operators.
Le apps di carattere medico sono di difficile valutazione per il cittadino comune poichè un utente medio possiede conoscenze mediche specialistiche poco approfondite. È fondamentale mettere il cittadino in condizioni di comprendere e valutare la caratterizzazione di una apps in termini di affidabilità dei contenuti. In questo progetto di tesi si usa un’analisi lessicale del linguaggio delle descrizioni delle apps mediche per valutarne la caratterizzazione dei contenuti. La parola chiave “pharma” è stata ricercata nel sito iMedicalApp.com e da questa ricerca sono state recuperate 48 apps. Queste sono state ricercate negli appmarket iTunes, GooglePlay, BlackBerry e Windows Phone producendo un totale di 60 descrizioni. Nelle descrizioni vengono ricercate i termini medici proveniente da un dizionario in cui sono presenti i termini medici di uso familiare provenienti dal dizionario Consumer Health Vocabulary (CHV) collegati univocamente ai concetti medici specialistici espressi tramite lo Unified Medical Language System (UMLS). L’applicativo sviluppato consiste in una base di dati per la raccolta e la gestione delle informazioni relative alle apps e in un’interfaccia grafica per permettere agli utenti di interagire con l’archivio. L’analisi delle percentuali di termini medici rispetto alle parole totali contenute in ogni descrizione ha permesso di costruire un indice di caratterizzazione suddiviso in 5 classi delimitate da valori percentuali di termini medici. Per le apps “pharma”, la classe più bassa (2,82-10,84%) è popolata dal 13% di apps, quelle successive: 10,84-18,86% dal 43,3%, 18,86-26,89% dal 33,3%, 26,89-34,91% dal 8,3% e 34,91-42,93% dal 1,6%. Dai risultati si evince che il 90% delle apps appartengono alle prime 3 classi di caratterizzazione (meno del 26,89%) quindi si può ipotizzare che la maggior parte delle apps di dominio “pharma” abbia di contenuti medici poco specialistici. Questo progetto è stato il primo approccio per la valutazione delle apps mediche a partire dalle descrizioni delle apps che comporta una valutazione oggettiva delle applicazioni, basata sull’esecuzione di processi computazionali automatici e non influenzata da opinioni soggettive.
MedApp-Onto : un prototipo software per una analisi del lessico delle descrizioni delle apps di ambito medico
BOLCHINI, VALENTINA MARIA
2013/2014
Abstract
The evaluation of Medical applications is difficult for common users because they usually have poor specialized medical knowledge. Hence providing these common users with the possibility to understand and to judge the robustness of an app in matter of reliability of its contents becomes crucial. The keyword “pharma” has been used for the search of medical apps on iMedicalApp.com and, thanks to this search, 48 apps have been found within the appstores iTunes, GooglePlay, BlackBerry and Windows Phone. A result set of 60 descriptions came out. We searched and characterized the medical terms contained in the descriptions. These terms have been compared to the medical words from the Consumer Health Vocabulary (CHV), which is a medical dictionary consisting of common medical terms. In addition, the CHV has been connected to specialist medical concepts explained through the Unified Medical Language System (UMLS). The application we developed is composed of a database to collect and manage the apps data and a user interface to interact with the application. The analysis of the percentage of medical terms out of the total words got in the apps descriptions allowed us to create a characterization index. Five classes compose this index. For the result set “pharma” apps, the lower class (2,82-10,84%) holds the 13% of descriptions, the other ones: 10,84-18,86% the 43,3%; 18,86-26,89% the 33,3%; 26,89-34,91% the 8,3% and 34,91-42,93%, the 1,6%. We can gather that 90% of the total apps are in the first three characterization classes (less than 26.89% of medical terms). From this, it is reasonable to suppose that the most of “pharma” apps has poor medical contents. This project is the first approach for the assessment of medical apps starting from their descriptions. It is based on an objective assessment of applications provided with the execution of computational and automatic processes and it is not depending on any subjective thoughts of operators.File | Dimensione | Formato | |
---|---|---|---|
2014_10_Bolchini.pdf
accessibile in internet per tutti
Descrizione: Manoscritto di tesi
Dimensione
3.38 MB
Formato
Adobe PDF
|
3.38 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/10589/96504