Feasibility of a Secure Wireless Sensing Smartwatch Application for the Self-Management of Pediatric Asthma

Clinical efficacy and satisfaction of a digital wheeze detector in a multicentre randomised controlled trial: the WheezeScan study

Introduction

Wheezing is common in preschool children and its clinical assessment often challenging for caretakers. This study aims to evaluate the impact of a novel digital wheeze detector (WheezeScan™) on disease control in a home care setting.

Methods

A multicentre randomised open-label controlled trial was conducted in Berlin, Istanbul and London. Participants aged 4-84 months with a doctor's diagnosis of recurrent wheezing in the past 12 months were included. While the control group followed usual care, the intervention group received the WheezeScan™ for at-home use for 120 days. Parents completed questionnaires regarding their child's respiratory symptoms, disease-related and parental quality of life, and caretaker self-efficacy at baseline (T0), 90 days (T1) and 4 months (T2).

Results

A total of 167 children, with a mean±sd age of 3.2±1.6 years, were enrolled in the study (intervention group n=87; control group n=80). There was no statistically significant difference in wheeze control assessed by TRACK (mean difference 3.8, 95% CI -2.3-9.9; p=0.2) at T1 between treatment groups (primary outcome). Children's and parental quality of life and parental self-efficacy were comparable between both groups at T1. The evaluation of device usability and perception showed that parents found it useful.

Conclusion

In the current study population, the wheeze detector did not show significant impact on the home management of preschool wheezing. Hence, further research is needed to better understand how the perception and usage behaviour may influence the clinical impact of a digital support.

Digital Technologies for Children and Parents Sharing Self-Management in Childhood Chronic or Long-Term Conditions: A Scoping Review

Worldwide, the prevalence of chronic (or long-term) conditions in children and young people from birth to 18 years (children) is increasing. Promoting competent and effective self-management skills early in the trajectory is important to improve adherence to treatment and optimise quality of life. Successful self-management, therefore, requires parents and children who are developmentally able to develop a range of complex skills, including the use of digital technologies. This scoping review aimed to identify primary research investigating digital technologies for children and parents sharing self-management in childhood chronic illnesses. A comprehensive search of electronic databases was conducted. Nineteen papers were included, assessed for quality and methodological rigour using the Hawker tool and thematically analysed. Three themes were identified: (i) the feasibility and acceptability of using technology, (ii) the usability of technologies and (iii) the effect of technologies on adherence and self-management skills. The results indicate that technologies such as mobile apps and websites can assist the management of long-term conditions, are an acceptable method of delivering information and can promote the development of effective self-management skills by parents and children. However, future technology design must include children and parents in all stages of development.

Semantic Data Mining in Ubiquitous Sensing: A Survey

Mining ubiquitous sensing data is important but also challenging, due to many factors, such as heterogeneous large-scale data that is often at various levels of abstraction. This also relates particularly to the important aspects of the explainability and interpretability of the applied models and their results, and thus ultimately to the outcome of the data mining process. With this, in general, the inclusion of domain knowledge leading towards semantic data mining approaches is an emerging and important research direction. This article aims to survey relevant works in these areas, focusing on semantic data mining approaches and methods, but also on selected applications of ubiquitous sensing in some of the most prominent current application areas. Here, we consider in particular: (1) environmental sensing; (2) ubiquitous sensing in industrial applications of artificial intelligence; and (3) social sensing relating to human interactions and the respective individual and collective behaviors. We discuss these in detail and conclude with a summary of this emerging field of research. In addition, we provide an outlook on future directions for semantic data mining in ubiquitous sensing contexts.

Machine Learning Within Studies of Early-Life Environmental Exposures and Child Health: Review of the Current Literature and Discussion of Next Steps

PURPOSE OF REVIEW

The goal of this article is to review the use of machine learning (ML) within studies of environmental exposures and children's health, identify common themes across studies, and provide recommendations to advance their use in research and practice.

RECENT FINDINGS

We identified 42 articles reporting upon the use of ML within studies of environmental exposures and children's health between 2017 and 2019. The common themes among the articles were analysis of mixture data, exposure prediction, disease prediction and forecasting, analysis of complex data, and causal inference. With the increasing complexity of environmental health data, we anticipate greater use of ML to address the challenges that cannot be handled by traditional analytics. In order for these methods to beneficially impact public health, the ML techniques we use need to be appropriate for our study questions, rigorously evaluated and reported in a way that can be critically assessed by the scientific community.

Artificial Intelligence and Machine Learning in Chronic Airway Diseases: Focus on Asthma and Chronic Obstructive Pulmonary Disease

Chronic airway diseases are characterized by airway inflammation, obstruction, and remodeling and show high prevalence, especially in developing countries. Among them, asthma and chronic obstructive pulmonary disease (COPD) show the highest morbidity and socioeconomic burden worldwide. Although there are extensive guidelines for the prevention, early diagnosis, and rational treatment of these lifelong diseases, their value in precision medicine is very limited. Artificial intelligence (AI) and machine learning (ML) techniques have emerged as effective methods for mining and integrating large-scale, heterogeneous medical data for clinical practice, and several AI and ML methods have recently been applied to asthma and COPD. However, very few methods have significantly contributed to clinical practice. Here, we review four aspects of AI and ML implementation in asthma and COPD to summarize existing knowledge and indicate future steps required for the safe and effective application of AI and ML tools by clinicians.

Artificial intelligence techniques in asthma: a systematic review and critical appraisal of the existing literature

Artificial intelligence (AI) when coupled with large amounts of well characterised data can yield models that are expected to facilitate clinical practice and contribute to the delivery of better care, especially in chronic diseases such as asthma.The purpose of this paper is to review the utilisation of AI techniques in all aspects of asthma research, i.e. from asthma screening and diagnosis, to patient classification and the overall asthma management and treatment, in order to identify trends, draw conclusions and discover potential gaps in the literature.We conducted a systematic review of the literature using PubMed and DBLP from 1988 up to 2019, yielding 425 articles; after removing duplicate and irrelevant articles, 98 were further selected for detailed review.The resulting articles were organised in four categories, and subsequently compared based on a set of qualitative and quantitative factors. Overall, we observed an increasing adoption of AI techniques for asthma research, especially within the last decade.AI is a scientific field that is in the spotlight, especially the last decade. In asthma there are already numerous studies; however, there are certain unmet needs that need to be further elucidated.

Biosensors in Occupational Safety and Health Management: A Narrative Review

A sensor is a device used to gather information registered by some biological, physical or chemical change, and then convert the information into a measurable signal. The first biosensor prototype was conceived more than a century ago, in 1906, but a properly defined biosensor was only developed later in 1956. Some of them have reached the commercial stage and are routinely used in environmental and agricultural applications, and especially, in clinical laboratory and industrial analysis, mostly because it is an economical, simple and efficient instrument for the in situ detection of the bioavailability of a broad range of environmental pollutants. We propose a narrative review, that found 32 papers and aims to discuss the possible uses of biosensors, focusing on their use in the area of occupational safety and health (OSH).

Domain Adaptation in Children Activity Recognition

Among the major challenges in training predictive models in wireless health, is adapting them to new individuals or groups of people. This is not trivial largely due to possible differences in the distribution of data between a new individual in a real-world deployment and the training data used for building the model. In this study, we aim to tackle this problem by employing recent advancements in deep Domain Adaptation which tries to transfer a model trained on a labeled dataset to a new unlabeled one that follows a different distribution as well. To show the benefits of our approach, we transfer an activity recognition model, trained on a popular adult dataset to children. We show that direct use of the adult model on children loses 25.2% in F1-score against a supervised baseline, while our proposed transfer approach reduces this to 9%.

The Features of Mobile-Based Software in Self-Management of Patients with Asthma: A Review Article

BACKGROUND

The use of mobile-based software for the self-management of patients with asthma improves the quality of life, reduces healthcare costs, provides effective health care interventions in asthma, and supports the patients in self-management. The current study was performed to identify the features of mobile-based self-management software for patients with asthma (MSSPA).

MATERIALS AND METHODS

The present review study was performed in 2018. Four databases including PubMed, Scopus, Emerald, and Google Scholar were screened by the combination of selected keywords. Data were collected using a data extraction form. Data were analyzed using the content analysis method. Results were abstracted and reported based on the study objectives.

RESULTS

Of the 297 articles retrieved during the first round of search, 24 were selected; 15 of which were the original articles (62.5%). As the most important applications of MSSPA, it could be used as a tool to support patients in self-management, provide them with educational information, and self-observation. Also, 75% of the studies (n=18) emphasized the effectiveness of MSSPA. Identification of the required field of the software was the most important requirement in using MSSPA. Nevertheless, some of the studies reported the low quality and compatibility of some designed apps compared with those of the available information systems.

CONCLUSION

Identification of MSSPA features and considering them in new versions can promote the quality of MSSPA. However, according to the results of the study, in addition to identifying the software features, more attention should be paid to the users' needs in software design.

Children Activity Recognition: Challenges and Strategies

In this paper, we study the problem of children activity recognition using smartwatch devices. We introduce the need for a robust children activity model and challenges involved. To address the problem, we employ two deep neural network models, specifically, Bi-Directional LSTM model and a fully connected deep network and compare the results to commonly used models in the area. We demonstrate that our proposed deep models can significantly improve results compared to baseline models. We further show benefits of activity intensity level detection in health monitoring and verify high performance of our proposed models in this task.

A Systematic Review of Patient-Facing Visualizations of Personal Health Data

OBJECTIVES

As personal health data are being returned to patients with increasing frequency and volume, visualizations are garnering excitement for their potential to facilitate patient interpretation. Evaluating these visualizations is important to ensure that patients are able to understand and, when appropriate, act upon health data in a safe and effective manner. The objective of this systematic review was to review and evaluate the state of the science of patient-facing visualizations of personal health data.

METHODS

We searched five scholarly databases (PubMed, Embase, Scopus, ACM Digital Library [Association for Computing Machinery Digital Library], and IEEE Computational Index [Institute of Electrical and Electronics Engineers Computational Index]) through December 1, 2018 for relevant articles. We included English-language articles that developed or tested one or more patient-facing visualizations for personal health data. Three reviewers independently assessed quality of included articles using the Mixed methods Appraisal Tool. Characteristics of included articles and visualizations were extracted and synthesized.

RESULTS

In 39 articles included in the review, there was heterogeneity in the sample sizes and methods for evaluation but not sample demographics. Few articles measured health literacy, numeracy, or graph literacy. Line graphs were the most common visualization, especially for longitudinal data, but number lines were used more frequently in included articles over past 5 years. Article findings suggested more patients understand the number lines and bar graphs compared with line graphs, and that color is effective at communicating risk, improving comprehension, and increasing confidence in interpretation.

CONCLUSION

In this review, we summarize types and components of patient-facing visualizations and methodologies for development and evaluation in the reviewed articles. We also identify recommendations for future work relating to collecting and reporting data, examining clinically actionable boundaries for diverse data types, and leveraging data science. This work will be critically important as patient access of their personal health data through portals and mobile devices continues to rise.

Applying Multivariate Segmentation Methods to Human Activity Recognition From Wearable Sensors' Data

Background

Time-resolved quantification of physical activity can contribute to both personalized medicine and epidemiological research studies, for example, managing and identifying triggers of asthma exacerbations. A growing number of reportedly accurate machine learning algorithms for human activity recognition (HAR) have been developed using data from wearable devices (eg, smartwatch and smartphone). However, many HAR algorithms depend on fixed-size sampling windows that may poorly adapt to real-world conditions in which activity bouts are of unequal duration. A small sliding window can produce noisy predictions under stable conditions, whereas a large sliding window may miss brief bursts of intense activity.

Objective

We aimed to create an HAR framework adapted to variable duration activity bouts by (1) detecting the change points of activity bouts in a multivariate time series and (2) predicting activity for each homogeneous window defined by these change points.

Methods

We applied standard fixed-width sliding windows (4-6 different sizes) or greedy Gaussian segmentation (GGS) to identify break points in filtered triaxial accelerometer and gyroscope data. After standard feature engineering, we applied an Xgboost model to predict physical activity within each window and then converted windowed predictions to instantaneous predictions to facilitate comparison across segmentation methods. We applied these methods in 2 datasets: the human activity recognition using smartphones (HARuS) dataset where a total of 30 adults performed activities of approximately equal duration (approximately 20 seconds each) while wearing a waist-worn smartphone, and the Biomedical REAl-Time Health Evaluation for Pediatric Asthma (BREATHE) dataset where a total of 14 children performed 6 activities for approximately 10 min each while wearing a smartwatch. To mimic a real-world scenario, we generated artificial unequal activity bout durations in the BREATHE data by randomly subdividing each activity bout into 10 segments and randomly concatenating the 60 activity bouts. Each dataset was divided into ~90% training and ~10% holdout testing.

Results

In the HARuS data, GGS produced the least noisy predictions of 6 physical activities and had the second highest accuracy rate of 91.06% (the highest accuracy rate was 91.79% for the sliding window of size 0.8 second). In the BREATHE data, GGS again produced the least noisy predictions and had the highest accuracy rate of 79.4% of predictions for 6 physical activities.

Conclusions

In a scenario with variable duration activity bouts, GGS multivariate segmentation produced smart-sized windows with more stable predictions and a higher accuracy rate than traditional fixed-size sliding window approaches. Overall, accuracy was good in both datasets but, as expected, it was slightly lower in the more real-world study using wrist-worn smartwatches in children (BREATHE) than in the more tightly controlled study using waist-worn smartphones in adults (HARuS). We implemented GGS in an offline setting, but it could be adapted for real-time prediction with streaming data.

Automated Quality Control for Sensor Based Symptom Measurement Performed Outside the Lab

The use of wearable sensing technology for objective, non-invasive and remote clinimetric testing of symptoms has considerable potential. However, the accuracy achievable with such technology is highly reliant on separating the useful from irrelevant sensor data. Monitoring patient symptoms using digital sensors outside of controlled, clinical lab settings creates a variety of practical challenges, such as recording unexpected user behaviors. These behaviors often violate the assumptions of clinimetric testing protocols, where these protocols are designed to probe for specific symptoms. Such violations are frequent outside the lab and affect the accuracy of the subsequent data analysis and scientific conclusions. To address these problems, we report on a unified algorithmic framework for automated sensor data quality control, which can identify those parts of the sensor data that are sufficiently reliable for further analysis. Combining both parametric and nonparametric signal processing and machine learning techniques, we demonstrate that across 100 subjects and 300 clinimetric tests from three different types of behavioral clinimetric protocols, the system shows an average segmentation accuracy of around 90%. By extracting reliable sensor data, it is possible to strip the data of confounding factors in the environment that may threaten reproducibility and replicability.