Detection of Cases of Noncompliance to Drug Treatment in Patient Forum Posts: Topic Model Approach (Preprint)

• DOI: 10.2196/preprints.9222
• Publication Date: 2017-10-22T19:57:04Z
• Authors: Redhouane Abdellaoui; Pierre Foulquié; Nathalie Texier; Carole Faviez; Anita Burgun; Stéphane Schück
• Keywords: Aripiprazole; Escitalopram; Microblogging

<sec> <title>BACKGROUND</title> Medication nonadherence is a major impediment to the management of many health conditions. A better understanding factors underlying noncompliance treatment may help professionals address it. Patients use peer-to-peer virtual communities and social media share their experiences regarding treatments diseases. Using topic models makes it possible model themes present in collection posts, thus identify cases noncompliance. </sec> <title>OBJECTIVE</title> The aim this study was detect messages describing patients’ noncompliant behaviors associated with drug interest. Thus, objective clustering posts featuring homogeneous vocabulary related nonadherent attitudes. <title>METHODS</title> We focused on escitalopram aripiprazole used treat depression psychotic conditions, respectively. implemented probabilistic topics that occurred corpus mentioning these drugs, posted from 2004 2013 three most popular French forums. Data were collected using Web crawler designed by Kappa Santé as part Detec’t project analyze for safety. Several treatment. <title>RESULTS</title> Starting 3650 an antidepressant (escitalopram) 2164 antipsychotic (aripiprazole), latent Dirichlet allocation allowed us several themes, including interruptions changes dosage. approach detected recall 98.5% (272/276) precision 32.6% (272/844). <title>CONCLUSIONS</title> Topic enabled explore discussions community websites behaviors. After manual review topics, we found 6.17% (276/4469) posts.