Development of a mobile health app for the management of hypertension, including treatment adherence assessment,using image detection technology - inspirers-HTN

• Autor(a) principal: Nogueira-Silva, Luís
• Data de Publicação: 2021
• Outros Autores: Viera-Marques, Pedro, Valente, José, Holtkötter, Jannis, Amaral, Rita, Jácome, Cristina, Ferreira, Ana, Almeida, Rute, Almeida Fonseca, João
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.22/18310
• topic: Mobile Health APP; Treatment Adherence Assessment

We aim to develop a mHealth smartphone app with a novel strategy to support the management of hypertension, including the measurement of adherence to treatment, taking advantage of the widespread use of smartphones and using only their embedded sensors. We have designed and developed a cross-platform, multi-language app which allows to register a pharmacological treatment and to customize alerts for the patient to take his/her medication and to measure his/her blood pressure (BP). Moreover, the app will be able to identify the number of pills in a blister and to capture the BP values from the screen of BP measuring devices, using the smartphone's camera. Thus, the app will quantify adherence to therapy and generate automatic BP reports. Blister photos and BP values collected by the users are enhanced using standard image processing methods for contrast increase, gap filling and relevant elements location. Classification strategies allow to count the pills present in the blisters, while the Google MLKit text mining API is employed for BP values recognition. Health Level Seven International (HL7) Fast Health Interoperability Resources (FHIR) standard is used for health care data modeling and exchange, promoting interoperability while guaranteeing data quality and security. Evaluation of the app's performance by real users will be presented, regarding its usability and the offline validity of the data acquisition. The user interface concept of the app has been defined and the mockups have been produced. Regarding pill counting, blisters with diverse materials and textures were considered. Different processing strategies are used depending on blisters’ characteristics, which has allowed an accuracy above 95% for most of the tested blisters. Extracting the BP measures from smartphone acquired images using the MLKIT app seems to be feasible. Further improvements and evaluations are ongoing. We will show preliminary results regarding usability and offline validity. We propose a new smartphone app that will support the management of hypertension, including an innovative image detection tool that will allow to objectively measure adherence to therapy and will facilitate the capture of BP values.