This paper presents the design, technical implementation, and feasibility evaluation of mCardia – a context-aware, mobile mCardia collection system for longitudinal arrhythmia screening under free-living conditions. Along with ECG, mCardia also records active and passive contextual data, including patient-reported symptoms and physical activity. This contextual data can provide a more accurate understanding of what happens before, during, and after an arrhythmia event, thereby providing additional information in the diagnosis of arrhythmia. By using a plugin-based architecture for ECG and contextual sensing, mCardia is device-agnostic and can integrate with various wireless ECG devices, and supports cross-platform deployment. We deployed the mCardia system in a feasibility study involving 24 patients who used the system over a two-week period. During the study, we observed high patient acceptance and compliance with a satisfactory yield of collected ECG and contextual data. The results demonstrate the high usability and feasibility of mCardia for longitudinal ambulatory monitoring under free-living conditions. The paper also reports from two clinical cases, which demonstrate how a cardiologist can utilize the collected contextual data to improve the accuracy of arrhythmia analysis. Finally, the paper discusses the lessons learned and the challenges found in the mCardia design and the feasibility study.

Deliverable D32:
Detailed set of privacy guidelines and schemata: Summarization of the outcome of the development of necessary data privacy and security schemata to (1) protect sensed data; (2) ascertain computational anonymity; (3) ensure privileged intervention access (associated with task T.7.5).

Abstract:
This deliverable report examines the outcomes of the REACH research pro-ject with regard to data privacy and data security, associated with Task T7.5. This document gives an overview of our analyses involving ethics and privacy concerns in terms of the individual touchpoints and shows how these findings guided the project towards the determination of the medical purpose and intended use – cornerstones to on the path to market entry. In addition, we provide a brief overview of how the guide-lines regarding data protection and encryption influenced the technical design and im-plementation of project components. Furthermore, we provide an update on the man-agement of legal implications (and the implications resulting from this for system re-quirements and business strategy) of the use of machine learning and artificial intelli-gence in the context of REACH solutions, incorporating an external expert opinion. Finally, this deliverable report contains a summary of our approach towards risk gov-ernance and standardization in this regard: our work in REACH on privacy and security schemata, culminated in a CEN Workshop Agreement (guideline) that generalizes REACH outcomes and makes them accessible and usable beyond the REACH con-sortium.

(PDF) Detailed set of privacy guidelines and schemata: Summarization of the outcome of the development of necessary data privacy and security schemata to (1) protect sensed data; (2) ascertain computational anonymity; (3) ensure privileged intervention access.

Digital tools have been developed to assess human cognitive functioning. It is unknown to what degree users’ cognitive test performance is correlated with their perceived usability and cognitive load induced by interaction with a tool. Moreover, the similarity between user groups in terms of their subjective usability and cognitive load has not been explored adequately despite its potential importance in designing digital cognitive assessment tools for people from diverse background. This paper presents a study of two smartwatch-based cognitive tests to assess participants’ attention and working memory. NASA Task Load Index (NASA-TLX) and Mobile App Rating Scale (MARS) questionnaires were used for cognitive load and usability evaluations, respectively. Aesthetics, functionality, and information quality and quantity were the metrics we selected for usability evaluations. Pearson’s correlation analysis was performed to investigate the associations and Ward’s clustering method was applied for data visualization. Our results showed that participants who received higher scores and longer scoring streak rated functionality of the cognitive tests better. Moreover, information quality and quantity of the tests were rated better by the participants who received longer scoring streak indicating the significant role of test instructions in gaining higher scores. In addition, participants with lower temporal demand received higher scores and faster mean response times. The key findings from the clusters visualized in this paper are: (i) Female and male participants rated their perceived usability and cognitive load completely differently; (ii) A discrepancy was found between participants’ perceived performance and their actual scores; (iii) Participants from diverse background rated their perceived usability and cognitive load different from each other.

Mobile cognitive tests have been emerged to first, bring the assessments outside the clinics and second, frequently measure individuals’ cognitive performance in their free-living environment. Patients with Bipolar Disorder (BD) suffer from cognitive impairments and poor sleep quality negatively affects their cognitive performance. Wearables are capable of unobtrusively collecting multivariate data including activity and sleep features. In this study, we analyzed daily attention, working memory, and executive functions of patients with BD and healthy controls by using a smartwatch-based tool called UbiCAT to 1) investigate its concurrent validity and feasibility, 2) identify digital phenotypes of mental health using cognitive and mobile sensor data, and 3) classify patients and healthy controls on the basis of their daily cognitive and mobile data. Our findings demonstrated that UbiCAT is feasible with valid measures for in-the-wild cognitive assessments. The analysis showed that the patients responded more slowly during the attention task than the healthy controls, which could indicate a lower alertness of this group. Furthermore, sleep duration correlated positively with participants’ working memory performance the next day. Statistical analysis showed that features including cognitive measures of attention and executive functions, sleep duration, time in bed, awakening frequency and duration, and step counts are the digital phenotypes of mental health diagnosis. Supervised learning models was used to classify individuals’ mental health diagnosis using their daily observations. Overall, we achieved accuracy of approximately 74% using K-Nearest Neighbour (KNN) method.

Motor activity in physical and psychological stress exposure has been studied almost exclusively with self-assessment questionnaires and from reports that derive from human observer, such as verbal rating and simple descriptive scales. However, these methods are limited in objectively quantifying typical behaviour of stress. We propose to use accelerometer data from smartphones to objectively quantify stress levels. Used data was collected in a real-world setting, from 29 employees in two different organisations over 5 weeks. To improve classification performance we propose to use intermediate models. These intermediate models represent the mood state of a person which is used to build the final stress prediction model. In particular, we obtained an accuracy of 78.2% to classify stress levels.

Objective
Stress at work is a significant occupational health concern. Recent studies have used various sensing modalities to model stress behaviour based on non-obtrusive data obtained from smartphones. However, when the data for a subject is scarce it becomes a challenge to obtain a good model.

Methods
We propose an approach based on a combination of techniques: semi-supervised learning, ensemble methods and transfer learning to build a model of a subject with scarce data. Our approach is based on the comparison of decision trees to select the closest subject for knowledge transfer.

Results
We present a real-life, unconstrained study carried out with 30 employees within two organisations. The results show that using information (instances or model) from similar subjects can improve the accuracy of the subjects with scarce data. However, using transfer learning from dissimilar subjects can have a detrimental effect on the accuracy. Our proposed ensemble approach increased the accuracy by ≈ 10% to 71.58% compared to not using any transfer learning technique.

Conclusions
In contrast to high precision but highly obtrusive sensors, using smartphone sensors for measuring daily behaviours allowed us to quantify behaviour changes, relevant to occupational stress. Furthermore, we have shown that use of transfer learning to select data from close models is a useful approach to improve accuracy in presence of scarce data.

There is growing amount of scientific evidence that motor activity is the most consistent indicator of bipolar disorder. Motor activity includes several areas such as body movement, motor response time, level of psychomotor activity, and speech related motor activity. Studies of motor activity in bipolar disorder have typically used self-reported questionnaires with clinical observer-rated scales, which are therefore subjective and have often limited effectiveness. Motor activity information can be used to classify episode type in bipolar patients, which is highly relevant, since severe depression and manic states can result in mortality. This paper introduces a system able to classify the state of patients suffering from bipolar disorder using sensed information from smartphones. We collected audio, accelerometer and self-assessment data from five patients over a time-period of 12 weeks during their real-life activities. In this research we evaluated the performance of several classifiers, different sets of features and the role of the questionnaires for classifying bipolar disorder episodes. In particular, we have shown that it is possible to classify with high confidence (≈ 85%) the course of mood episodes or relapse in bipolar patients. To our knowledge, no research to date has focused on naturalistic observation of day-to-day phone conversation to classify impaired life functioning in individuals with bipolar disorder.

Stress at work is a significant occupational health concern nowadays. Thus, researchers are
looking to find comprehensive approaches for improving wellness interventions relevant to
stress. Recent studies have been conducted for inferring stress in labour settings; they
model stress behaviour based on non-obtrusive data obtained from smartphones. However,
if the data for a subject is scarce, a good model cannot be obtained. We propose an
approach based on transfer learning for building a model of a subject with scarce data. It is
based on the comparison of decision trees to select the closest subject for knowledge
transfer. We present an study carried out on 30 employees within two organisations. The
results show that the in the case of identifying a “similar” subject, the classification accuracy
is improved via transfer learning.

INTRODUZIONE

Per il 53% dei lavoratori intervistati in Europa, lo stress risulta il più importante rischio percepito durante il lavoro con una crescita significativa negli ultimi anni. Il 27% dei lavoratori ha riportato di soffrire di ‘stress, depressione, ansia” causata o peggiorata dal lavoro nei 12 mesi precedenti [1].

Secondo l’OMS, le patologie neuropsichiatriche rappresentano la seconda causa di morte prematura dopo le patologie cardiovascolari 1 .

Il 79% dei manager europei riconosce la presenza di stress nei loro ambienti di lavoro e il 40 % ritiene che i rischi psicosociali siano più difficili da affrontare rispetto ai rischi lavorativi tradizionali 2 . Per contro solo il 30% delle organizzazioni europee ha adottato procedure per gestirli 3 .

Attraverso l’utilizzo delle nuove tecnologie è oggi possibile monitorare importanti fattori relativi al contesto e contenuto del lavoro (e.g., carico di lavoro) e realizzare interventi preventivi dello stress e del burnout che non interferiscono con le attività lavorative quotidiane. Tale approccio consente inoltre di soddisfare in modo più efficace e personalizzato quanto previsto dalla legislazione specifica in tema di Salute e Sicurezza negli ambienti di lavoro (e.g., D.L.vo 81/2008).

Mobile computing is changing the landscape of clinical monitoring and self-monitoring. One
of the major impacts will be in healthcare, where increase in number of sensing modalities is
providing more and more information on the state of overall wellbeing, behaviour and
health. There are numerous applications of mobile computing that range from wellbeing
applications, such as physical fitness, stress or burnout up to applications that target mental
disorders including bipolar disorder. Use of information provided by mobile computing
devices can track the state of the subjects and also allow for experience sampling in order to
gather subjective information. This paper reports on the results obtained from a medical trial
with monitoring of bipolar disorder patients and how the episodes of the diseases correlate
to the analysis of the data sampled from mobile phone acting as a monitoring device.

In this paper, we present the concept of grouping individuals and detecting their proximity by emitting/receiving inaudible tones using their mobile phones. The inspiration stems from uniforms metaphor (of different colors) that groups subjects based on the roles, occupations or teams. The goal is to get an insight into the social context and social interaction patterns.

Social interactions play an important role in the overall well- being. Current practice of monitoring social interactions through ques- tionnaires and surveys is inadequate due to recall bias, memory depen- dence and high end-user effort. However, sensing capabilities of smart- phones can play a significant role in automatic detection of social in- teractions. In this paper, we describe our method of detecting interac- tions between people, specifically focusing on interactions that occur in synchrony, such as walking. Walking together between subjects is an important aspect of social activity and thus can be used to provide a better insight into social interaction patterns. For this work, we rely on sampling smartphone accelerometer and Wi-Fi sensors only. We analyse Wi-Fi and accelerometer data separately and combine them to detect walking in synchrony. The results show that from seven days of moni- toring using seven subjects in real-life setting, we achieve 99% accuracy, 77.2% precision and 90.2% recall detection rates when combining both modalities.

The emergence of genomically targeted cancer treatments has spurred the development of methods that correlate genomic information with treatments and outcomes. Because this information is usually pulled from published literature, such methods are limited to summarizing only the data generated through the slow and narrow publication pipeline. However, many thousands of patients are treated each year whose data does not make it into publications. Each of these is, in effect, a case study whose capture would add to our overall knowledge of cancer treatment, and could speed up the search for treatments and cures. Our work extends one such literature- based system, the Melanoma Targeted Therapy Advisor (TTA), by adding direct patient profiling. This extension of the TTA, or PTTA (Personalized or Patient TTA), both enriches the TTA knowledge base by collecting case reports directly from patients, and gives patients and/or physicians immediate feedback by ranking the best-performing treatments for genomic profiles of interest to them. The PTTA will permit patients to register their test and treatment results and then to obtain rankings for additional potentially useful tests and treatments. It will also provide a report with statistical and literature evidence that justifies the rankings. These functionalities can aid physicians in treating patients in the most effective manner.

The level of participation in social interactions has been shown to have an impact on various health outcomes, while it also reflects the overall wellbeing status. In health sciences the standard practice for measuring the amount of social activity relies on periodical self-reports that suffer from memory dependence, recall bias and the current mood. In this regard, the use of sensor-based detection of social interactions has the potential to overcome the limitations of self-reporting methods that have been used for decades in health related sciences. However, the current systems have mainly relied on external infrastructures, which are confined within specific location or on specialized devices typically not-available off the shelf. On the other hand, mobile phone based solutions are often limited in accuracy or in capturing social interactions that occur on small time and spatial scales. The work presented in this paper relies on widely available mobile sensing technologies, namely smart phones utilized for recognizing spatial settings between subjects and the accelerometer used for speech activity identification. We evaluate the two sensing modalities both separately and in fusion, demonstrating high accuracy in detecting social interactions on small spatio-temporal scale.

(PDF LINK or press the link at the bottom left)

A number of clinical studies investigated associations between mood states and environmental factors. However, they mostly rely on self-reporting methods to describe past activities which, due to recall difficulties, may not be reliable. In this pilot study, we attempted to measure the amount of social interaction at workplace in an objective way and to investigate correlations with mood states. The results show correlation between social interactions and mood states both in the beginning and at the end of monitored intervals.

Stress assessment is a complex issue and numerous studies have examined factors that influence stress in working environments. Research studies have shown that monitoring individuals’ behaviour parameters during daily life can also help assess stress levels. In this study, we examine assessment of work-related stress using features derived from sensors in smartphones. In particular, we use information from physical activity levels, location, social-interactions, social-activity and application usage during working days. Our study included 30 employees chosen from two different private companies, monitored over a period of 8 weeks in real work environments. The findings suggest that information from phone sensors shows an important correlation with employees’ perceived stress level. Secondly, we used machine learning methods to classify perceived stress levels based on the analysis of information provided by smartphones. We used decision trees obtaining 67.57% accuracy and 71.73% after applying a semi-supervised method. Our results show that stress levels can be monitored in unobtrusive manner, through analysis of smartphone data.

(PDF LINK or press the link at the bottom left)