JMIR Publications

ABSTRACT

Background:

In the EU, approximately 165 million people suffer from mental health disorders, namely anxiety, mood disorders, and substance abuse. These mental health disorders have resulted in both direct and indirect global economic costs estimated at US $2.5 trillion in 2010, with the indirect costs (USD $ 1.7 trillion) being significantly higher than the direct costs (USD $ 0.8 trillion), contrasting the trends of other key diseases, such as cardiovascular disease and cancer [1]. Concurrently, problematic alcohol consumption is considered the causal factor of more than 200 diseases, leading to 5.1% of all global diseases and damages being attributed to alcohol abuse. In addition to the risks posed to health, the social and economic damages involved not only for individuals suffering from alcohol abuse, but for society in general, must be taken into consideration [2]. Monitoring mental health outcomes has traditionally been based on heuristic decisions often based on scarce, subjective evidence, making the clinical decisions made by professionals, as well as the monitoring of these diseases, subject to flaws [3]. However, the "digital phenotype", which refers to the analysis of data collected by measuring human behavior with mobile sensors and smart bracelets, is a promising tool for filling this gap in current clinical practice, offering objective evidence in an otherwise subjective field of diagnosis, namely through the utilization of digital biomarkers [4]. Biomarkers are physiological, pathologic, or anatomic characteristics measured objectively and used to evaluate a patient’s health in the status quo by comparing their current biomarkers to ideal, healthy biomarkers. In the wake of advancing technology, “digital biomarkers” are emerging as a new form of tracking patient’s health. Digital biomarkers are considered digital as they utilize sensors and computational tools to collect data. These measurements can be taken outside of a traditional clinical environment using sensors such as mobile device sensors and wearable sensors [5]. Numerous studies have shown that information collected from behavioral and physiological sensors can be used to improve conventional evaluations of patients with symptoms of anxiety. In addition, patients who use these platforms show higher levels of consistency with their treatment plans than those only involved in conventional practice [6]. There are various examples of the effectiveness of digital phenotype data-collection in analyzing anxiety disorders: There exist significant patterns between the time spent in certain places and depression and anxiety, although these relationships were not consistent [7]. It has been shown that sensors integrated into smart bracelets measuring conductance and skin temperature can be utilized to reach a 78.3% (148/189) accuracy in classifying students into groups of high or low stress levels. Furthermore, the sensors provided a 79% (37/47) accuracy rate for classifying students’ state of mental health [8]. Furthermore, analysis from sensors tracking GPS location and incoming and outgoing text messages and calls collected from 54 college students over a two week period indicated that levels of social anxiety can be predicted with an accuracy of up to 85% by tracking these variables [9]. Alongside digital tracking of anxiety symptoms, there also exist various applications designed to facilitate the monitoring of alcohol consumption in patients with alcohol abuse disorder [10,11,12]. Although more research is needed, these applications have been shown to be effective in increasing patients’ ability to manage their condition [10], leading to reductions in days of consumption risk [11]. In addition, these tools allow for patients’ usual professional caretaker to monitor their condition, allowing for more informed care [10]. Therefore with REMOTE study we aim to provide valuable insights and knowledge about digital biomarkers in alcohol use disorders and anxiety disorders.

Objective:

Because of the emerging research on digital biomarkers and in the accuracy of passive monitoring of data in patients with mental health disorders, we pose a study in a group of 60 participants split into two subgroups of 30 participants. The primary objective of this study is to analyze the digital physiological patterns of two groups of participants, one group with symptoms of anxiety disorder and alcohol abuse disorder and one healthy control group without anxiety and alcohol abuse disorder, using data collected from a mobile (smartphone) and wearable (FitBit) sensor. The data collected from the sensors trough humanITcare platform will then be compared with the symptoms validated in clinical questionnaires, which participants will take four times over the course of the study, in order to determine whether sensor data is an efficient and accurate means of presenting objective diagnoses rather than the subjective diagnosis of the status quo. The secondary objective of this study is to analyze usability and patient satisfaction with the data collection service provided by the application.

Methods:

DESING: This study is a case-controlled, prospective study consisting of two groups: one group of healthy control individuals who do not have symptoms of Alcohol Use Disorders or Anxiety symptoms and one experimental group that meets the selection criteria. Participants in both groups will be matched for age and sex, since both can be confounding factors regarding usage patterns of electronic devices. Both groups will receive mobile and wearable sensors that will track their physiological and behavioral activity over the course of one month, and both groups will take the State-Trait Anxiety Inventory (STAI) [13], Beck Depression Inventory-II (BDI-II) [14], and AUDIT [15] once per week. The data will be monitored by the research team through an online, encrypted compilation of participants’ data. Study Population and Setting: The study will be conducted at the Hospital Clínic de Barcelona. It is a single-center, national study. Experimental participants will be drawn from the outpatient clinic at the Addictions Unit of the Hospital Clínic de Barcelona and control group participants will be drawn from random volunteers recruited through fliers and social media. The research team members at the Hospital Clínic de Barcelona Addictions Unit will be tasked with finding eligible subjects for the experimental group. The inclusion criteria for the experimental group are that the participants must be between the ages of 18-65, have knowledge and daily use of new technologies, have alcohol use disorder (based on the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition [DSM-5]), have significant anxiety symptoms (State-Trait Anxiety Inventory [STAI] > 33 percentile), and have signed informed about the study and data processing consent. Potential experimental participants are not eligible if they meet any of the following criteria: their mobile device is not compatible with Android mobile operating system, have a diagnosis of affective disorder (based on DSM-5), have cognitive deficits that prevent proper participation, actively consume other substances (except nicotine). Recruitment The recruitment of the 30 experimental patients, conducted by psychiatrists and clinical psychologists, will take place in the outpatient clinic and day hospital of the Addictions Unit of the Hospital Clínic de Barcelona. The usual healthcare professional for patients with anxiety symptoms will refer the selected patients to the research team who will assess whether the selected patients meet the eligibility criteria. 30 healthy controls will be recruited using brochures distributed by the faculty of medicine at the University of Barcelona as well as through social media. There will be a total of 60 participants split between the experimental and control subgroups, and patients will be sorted into subgroups according to whether they fit the eligibility criteria (enter experimental subgroup) or are healthy volunteers (control subgroup). Patients deemed eligible for the study will be contacted by the research team and will have the study explained to them. They will be given a sheet with instructions for proper usage of the FitBit device (how to load it, how to connect to the application, etc.). After a full explanation of the trial, signing for informed consent will be explained (including the trial’s use of data assets and potential liabilities). Study Procedures Participants will complete the study phase as follows: Initial in-person visit (1st visit): Once deemed eligible by the research team, participants who meet the requirements for the experimental group as well as the volunteer control participants will be called to the Hospital Clínic de Barcelona. The first visit will last approximately one hour and will include an explanation of the project to the participant with time for them to ask questions and clarify any doubts, explanation of the participant’s rights and signature of informed consent by the participants in the presence of the research team, and the measuring of the following variables: age, gender, education level, and marital status, symptoms of anxiety (measured by STAI), symptoms of depression (measured by BDI-II), evaluation of alcohol use disorder (measured by AUDIT), use of medication, medical co-morbidities, and substance use. Once the variables are profiled, the participant will be instructed to download the app “U-shine" of humanITcare platform and receive their data encryption code and password. The “U-shine” app will consolidate and send the data collected by the mobile phone sensors to the research team for evaluation. A FitBit device will also be provided and the participant will be informed of their obligation to return the device once the study is completed. For proper operation of the FitBit, participants will be instructed to access the application “FitBit” through personal email, and specify their age, sex, weight and height, date of birth. This application complies with data protection laws. Physiological and behavioral data from the participants’ smartphone devices will be compiled in the U-Shine servers and will be transmitted, along with the data collected in the FitBit servers, to a separate server available to the research team which assigns each participant a unique, randomized character code that functions as a profile name (ex: “mt1e45al”). The data collected through the U-shine app and servers as well as the Fitbit servers for each participant will be represented this 8 digit code in order to protect the participants’ privacy. Weekly assessment of depression and anxiety symptoms(4x): Patients will receive a notification from the U-Shine application reminding them to respond to the STAI, BDI-II , and AUDIT [15] questionnaires available on the U-Shine application once per week for a total of 4 evaluations throughout the duration of the study period. Non-contact visits: Research assistants will perform supervision of participants’ data and will monitor whether participants are answering the three weekly questionnaires through the U-shine application. They will make reminder calls to the participants in the case that the participants are not responding to the questionnaires, in which case the factors that have influenced non-response will be explored. All this data will be recorded. Final in-person visit (2nd visit): The investigator shall summon each participant to the Hospital Clínic de Barcelona to return the borrowed FitBit device. In the case of the experimental group, data will be collected regarding user experience with the application through an adaptation of the System Usability Scale (SUS) [18] and the Post-Study Usability Questionnaire (PSSUQ) [17]. Once the study has been completed, data analysis and data processing will follow. After the statistical analysis is complete, we will proceed to the discussion and interpretation of the results and the writing of a scientific article that outlines the findings. HumanITcare (Ushine App) HumanITcare has develop an Internet of Things (IoT) platform for the collection and analysis of the daily data of the participants. The App UShine is an Android and IOS application developed by humanITcare with the intent of collecting mobile phone sensor data for clinical usage. The app continuously collects and stores users’ sensor data within the UShine servers, automatically uploading the data collected by the sensors to the humanITcare servers when connected to WiFi. This allows for constant gathering of sensor data regardless of the quality of network service available at the time, and also ensures that mobile data fees aren’t charged to users. HumanITcare platform, ensures anonymity for users by utilizing the SHA-256 hashing algorithm to encrypt the data collected by users’ mobile phone sensors. Furthermore, all of the data collected for each participant is only recognizable by a randomized 8-character code (ex. “mt1e45al”) assigned to each participant, ensuring that the data is anonymous. The data collected by UShine was evaluated in tandem with data collected by a FitBit wearable device, however the UShine application offered a wider breadth of sensors The data collected by the UShine app and the FitBit wearable device.

Results:

Outcomes Primary Outcome: Sensor Data Accuracy and Efficiency The symptoms of anxiety and alcohol disorder presented to the research team derived from the algorithm calculated from data collected by the humanITcare platform (UShine app) and the FitBit devices will be compared with the diagnoses provided by the status-quo “golden standard” questionnaires the State-Trait Anxiety Inventory (STAI [13]), and The Diagnostic and Statistical Manual of Mental Health Disorders 5th (DSM-5 [16]) to determine the feasibility and efficiency of using objective sensor data in place of the current subjective methods of diagnosing mental health disorders. This will be done in a two-step process: first by determining the relationship between the individual types of data tracked by sensors (eg. distance traveled, sleep schedule) and levels of anxiety/alcoholism symptoms as presented by the questionnaires; second, by using a mathematical model to estimate questionnaire scores using solely the data collected by sensors. The primary outcome will be the efficiency of the developed algorithm’s ability to predict disorder symptoms Secondary Outcome: System Usability Satisfaction with and usability of the data collection applications will be evaluated after the trial period in order to determine the practicality of the user interface. This will be done to determine the practical feasibility of widespread usage of similar applications. 1) Satisfaction with the application will be scaled using the Post-Study Usability Questionnaire (PSSUQ) [17]- Scores range from 0-100, 0 being least satisfactory and 100 being the most satisfactory. Participants will be asked to take the PSSUQ after the data collection period is over. 2) Usability of the mobile application will be scaled using the System Usability Scale (SUS) [18]. Scores range from 0-100, with a score of ≥68 being considered above average. Participants will be asked to take the SUS after the data collection period is over. Data Collection, Management, Security and Ethics: Treatment, communication and transfer of personal data of all participants will be adjusted to compliance with the EU Regulation 2016/679 of the European Parliament and of the Council of 27, April 2016 on the protection of individuals with regard to the processing of personal data and the free movement of data. the legal basis that justifies the processing of data is the consent hereby given, pursuant to the provisions of Article 9 of EU Regulation 2016/679. The data collected will be collected from each participant is identified only by a randomized 8-digit code, so it will not include any information that could identify participants. Only the research team members with the right of access to the source data (medical history) may relate the data collected in the study with the clinical history of the patient. The identity of participants will not be available to any other person except for a medical emergency or legal requirement. The Ethics Committee for Research, health authorities, and the personnel authorized by the study sponsor will have access to identifiable personal information when necessary to check data and study procedures, but will always maintain confidentiality in accordance with current legislation. Moreover, the project is under the Declaration of Helsinki (2013). RESULTS; DATA ANALYSIS; Quantitative i. Sample size and statistical power; At the moment we have recruited at 60 participants, of those 55 participants have already finished, the 5 patients missing will finish by the first week of November 2019. ii. Statistical methods: The statistical analysis will be carried out as follows. First, it will be select the sample of patients by groups to do the 75% of patients in stratified form will be reserved for the training; with 10% the model will be validated and with the remaining 15% it will be demonstrated its effectiveness.  

Conclusions:

This is a study protocol, final results will be publish before january 2020. Clinical Trial: https://clinicaltrials.gov/ct2/show/NCT03991650