Design of two combined health recommender systems for tailoring messages in a smoking cessation app

📝 Abstract

In this article, we describe the design of two recommender systems (RS) designed to support the smoking cessation process through a mobile application. We plan to use a hybrid RS (content-based, utility-based, and demographic filtering) to tailor health recommendation messages, and a content-based RS to schedule a timely delivery of the message. We also define metrics that we will use to assess their performance, helping people quit smoking when we run the pilot.

💡 Analysis

In this article, we describe the design of two recommender systems (RS) designed to support the smoking cessation process through a mobile application. We plan to use a hybrid RS (content-based, utility-based, and demographic filtering) to tailor health recommendation messages, and a content-based RS to schedule a timely delivery of the message. We also define metrics that we will use to assess their performance, helping people quit smoking when we run the pilot.

📄 Content

Design of two combined health recommender systems for tailoring messages in a smoking cessation app Santiago Hors- Fraile University of Seville ETSII Universidad de Sevilla, Avda. Reina Mercedes S/N 41012 Seville, Spain (+34) 954556817 sanhorfra@gmail.com Francisco J Núñez Benjumea Virgen del Rocío University Hospital Technological Innovation Group Av Manuel Siurot S/N (+34) 955013616 francisco.nunez.exts @juntadeandalucia.es Laura Carrasco Hernández Virgen del Rocío University Hospital Smoking Cessation Unit Av Manuel Siurot S/N (+34) 635581010 lauracarrascohdez@g mail.com Francisco Ortega Ruiz Virgen del Rocío University Hospital Smoking Cessation Unit Av Manuel Siurot S/N (+34) 955013163 francisco.ortega.sspa @juntadeandalucia.es
Luis Fernandez- Luque Salumedia Tecnologías Sevilla, Spain (+34) 656930901 luis@salumedia.com

ABSTRACT In this article, we describe the design of two recommender systems (RS) designed to support the smoking cessation process through a mobile application. We plan to use a hybrid RS (content-based, utility-based, and demographic filtering) to tailor health recommendation messages, and a content-based RS to schedule a timely delivery of the message. We also define metrics that we will use to assess their performance, helping people quit smoking when we run the pilot.
CCS Concepts • Human-centered computing, Collaborative and social computing, Collaborative filtering; 300, • Human-centered computing – Mobile phones – Ubiquitous and mobile devices; 300, • Applied computing– Life and medical sciences– Consumer health; 300. Keywords smoking, cigarettes, motivation, recommender systems, mobile apps, tailoring, personalization, messages

1. INTRODUCTION Smoking is responsible for severe conditions such as exacerbation of asthma, pulmonary fibrosis, chronic obstructive pulmonary disease (COPD), and lung cancer, among others [1]. People who try to quit suffer from nicotine abstinence syndrome. They usually experience cravings, headaches, intestinal disorders, weight gain, insomnia, dullness and irritability among other withdrawal symptoms [2]. Although tobacco prevalence differs substantially among countries [3], statistics show how people still try to quit [4] and that the success rate has room for improvement.
   The high penetration of smartphones in everyday lives lets us reach people in a way that we could not before [5]. Mobile devices, and their associated mobile application software – referred from now on as apps – have brought a new channel of communication. Since they are ubiquitous, we can reach patients almost anywhere anytime. Besides, smartphones let us have a real-time feedback loop to understand what, how, and when patient support is most effective and well-received. In order to help people who want to quit smoking, many authors have shown that apps are a suitable tool [6]. Despite their results, there is still a need to find out how patients can receive more effective support. Several authors have proved that personalized and tailored apps may increase the effectivity of the lifestyle recommendations that patients can get via their smartphones [7]. In this context, one of the things the SmokeFreeBrain (SFB) project – funded by the EU within the Horizon 2020 program – aims to explore is how effective an app can be throughout the smoking cessation process.
   Among other features, the app prompts users motivational messages after the quitting date. The message frequency is based on the trans-theoretical Behavior Model [8], as some studies have already proposed [9]. In order to maximize the effectiveness of the system, the following questions should be addressed: What type of topic motivates each user more? When does each user prefer to receive the message so that the message is more helpful and not intrusive? To solve these questions, we included two RSs in a server that worked in coordination with the app. One of them was a hybrid RS to select the messages of those topics more relevant for the user, that actually help them change their behavior. The other RS was to tune the time in which those messages were more effective. Each night a scheduled task runs to calculate when and which message has to be sent for each user. We will measure the impact of the RS outcomes both with the in-app metric statistics and a satisfaction questionnaire at the end of the pilot.
2. METHODS 2.1 Pilot protocol overview This is a randomized open-label parallel-group trial. For eight months, all patients attending the SCU of the VRUH will be offered to join this study. The inclusion criteria are: Smoking population attending to the SCU of VRUH, subjects older than 18 who want to give up smoking with Android-based smartphones and the ability to interact with the smartphone; subjects who can also sign an Informed Consent Form. Subjects who have adverse effects related to the pharmacolog

This content is AI-processed based on ArXiv data.