The Community Health aims at promoting digital health projects on an interdisciplinary basis.
The projects are divided in four groups. Projects A, B, and C correspond to different stages of development. Projects D present DSI/UZH digital know-how and expertise that could support other projects.
Do you have a Digital Health project? Please send your project information and contact name, using the following template (PPTX, 253 KB) to the coordinator of the DSI challenge area health Jana Sedlakova.
Please look at the projects from the members of the DSI challenge area health community (click at the beginning of the title for more information and the contact name).
Older Adults’ social communication and well-being during COVID
- COVID-19 pandemic can be seen as an extreme stressor that has potentially detrimental effects on health and well-being
- Older adults are particularly affected by social distancing measures and might be at risk of loneliness
- Maintaining social communication might have a buffering effect on wellbeing during this situation
- Did the social distancing measures have an effect on subjective well-being?
- Does social communication before and/or during social distancing affect changes in subjective well-being?
- Micro-longitudinal study April – November 2019 on older adults’ daily social communication
- Participants contacted again in March 2020 to fill in a weekly online-questionnaire on their communication and subjective well-being during social distancing
- Both questionnaires explicitly also included digital communication
- Social distancing had a negative effect on well-being: positive affect was reduced, negative affect and loneliness increased
- Satisfaction with communication during social distancing alleviated this effect: Those who were more satisfied with their communication showed reduced decline in subjective well-being
The language of the COVID-19 Pandemic: Investigating official communication and its relations with collective and individual emotions
What is the proportion of robots in Tweets? - BOTS
- Unclear how many bots influence the public conversation in social media in general and discourse on public health topics in particular
- Identify bots in georeferenced social media data from Twitter based on Botometer, the current gold standard in bot detection, and
- Apply a combined geographical and emotional trajectory analysis to evaluate potential improvements in bot detection of these data.
- Indiana University`s Botometer: Identify bots (current gold standard) https://botometer.iuni.iu.edu/#!/
- GIS: Activity spaces and mobility indicators
- EMOTIVE, Stresscapes, LIWC: Basic emotions, overall stress, thinking styles, social concerns, and parts of speech
- Just started. Looking for interested collaborators
PubliCo – an experimental online platform for COVID-19 related public perception
In the context of the COVID-19 crisis (and in many similar acute crises) citizens need personalized and reliable information, and policy-makers need to constantly sense the public perception in order to allow for continuous adaptations and improvement of emergency management and communication strategies.
- How do citizens perceive public health measures? What is the level of anxiety and the readiness to comply with current and potential public health measures? What are citizens’ preferences regarding potentially unavoidable trade-offs (e.g., restrictive measures, testing and surveillance, resource allocation) (empirical)
- How can RCC be improved in Switzerland? What can a platform like PubliCo contribute to RCC? What are best practice standards for such a platform? (methodological)
- What is good RCC? How should we communicate in crisis situations to avoid or minimize panic reactions? How can we address moral dilemmas, such as withholding or “spinning” information to affect behavior in the interest of public health? (theoretical)
- Establish PubliCo – an experimental, interactive online platform collecting data on public perception of COVID-19 and its implications, as a feedback loop for policymakers, health authorities, experts and media professionals engaged in providing information to the Swiss public.
- Design and apply toolkits to assess emotional state, behavioral dispositions, changes in social practices, and moral preferences of the public (e.g., regarding restrictive measures or resource allocation issues).
- Develop an ethical framework for public health crisis communication, including ethical criteria for “good” communication, preconditions for trust, cooperation and responsibility, as well as the role of public preferences in relation to policymaking and crisis management.
- Literature analysis and expert focus group to define the relevant subscales and items for the PubliCo survey, and the information to offer according to the scores;
- Definition of the automated analysis for specific subscales (descriptive and inferential statistics)
- Development and deploy of the PubliCo platform, collecting both survey data and diaries and providing real time monitoring features;
- Iterative adaptation of the platform based on qualitative analysis of the diaries;
- Publication of focused policy briefs based on the data.
Predicting the endorsement of preventive behaviors in the context of the Corona virus pandemic: Examining temporal dynamics and the role of risk communication
The current situation of the ongoing new Corona virus pandemic offers a rare chance to investigate the relation between perceived risks for oneself and one’s social environment, known health-behavior related factors (i.e., self-efficacy, response efficacy of different protective behaviors, perceived social norms), and the self-reported intention and adoption of protective behaviors. Different ways to communicate the risk of contracting the Covid-19 virus either to an individual or to a group of people (dilution effect, Slovic, 2007) should affect endorsement of preventive behaviors in general and social distancing in particular, as well as pandemic-related prosocial behavior.
- to investigate the dynamics of the relation of theory-based determinants and the self-reported intention and adoption of protective behaviors over time in relation to developments of the pandemic and the associated recommendations / restrictions given by legal authorities in a representative sample of the Swiss population
- to examine age differences in the predictors of the protective behaviors and these behaviors themselves as well as their developments over time
- to examine different ways to communicate risk based on the dilution effect of Slovic (2007) and to investigate potential age differences.
- The first point of measurement took place in the weeks end of March – first week in April (shortly after the lockdown in Switzerland)
- The second point of measurement took place starting May 11, the second step in reopening after the lockdown
- Six more points of measurement will take place during 2020 and beginning of 2021, approx. every two months, depending on the development of the pandemic.
- Pilot study currently ongoing
- Launch of experiment approx. mid June
- In collaboration with gfs Zürich (https://gfs-zh.ch/), sampling 1000 cohabitants of Switzerland: 200 people living in the Italian speaking part, 300 people living in the French speaking part, and 500 adults living in the German speaking part of Switzerland
- telephone surveys, including random digit dialing (RDD) à highest likelihood of unbiased coverage of the general population
- All participants included in the first survey (see preregistration at OSF DOI 10.17605/OSF.IO/G6EHD) will be asked to continue their participation in the following surveys. For these participants only à online surveys
- In order to keep up power and representativeness, new participants will be recruited in every wave for compensating dropouts = representative longitudinal survey on the population level & analyzing individual change over time
- Online experiment via respondi.de, N = 600 Swiss participants
Visual Remote vitals
In the context of the Covid Pandemic
- Acute overwhelming of health services
- Diagnostic uncertainty pending test turn-around
- Objective monitoring of those in self isolation hard and not scalable
- Weaknesses of contact tracking apps reliant on slow or ambiguous inputs
Our proposed Solution:
- Changes in vital signs used clinically to inform diagnosis and estimate disease severity
- We have developed an algorithm to estimate pulse and resp rate on the basis of smartphone video
Estimation of vital signs form smartphone video
- Analysis of frame to frame changes in facial colour channels permits estimation of pulse rate
- Shoulder and thorax movement informs respiratory rate
- User input temp data
- Train CNL layer to sit on top of computer vision pre-analysis stage based on clinical data from CE marked monitoring devices using USZ patients
Brasil Sem Corona – COVID-19 Participatory Disease Surveillance
- Participatory surveillance has shown promising results from its conception to its application in several public health events. The use of a collaborative information pathway provides a rapid way for data collection on symptomatic individuals in the territory, in order to complement traditional disease surveillance systems. In Brazil, this methodology has been used at national level since 2014 during mass gatherings due to its great importance for related to potential public health emergencies.
- Objectives: to describe and analyze the priority risk areas for Covid-19 testing combining participatory surveillance and traditional surveillance data.
- Methods: Participatory surveillance, spatial-temporal scanning, anomaly detection, time series forecast and locally regressions.
- Pre-print: https://www.medrxiv.org/content/10.1101/2020.05.25.20109058v2
The Incidence of and Risk Factors for Coronavirus Disease 2019 (COVID-19) among Healthcare Workers – a Prospective Cohort Study"
- Hospital employees, being at increased risk for acquisition and transmission of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), are at the centre of the Coronavirus Disease 2019 (COVID-19) pandemic. In order to better understand and predict local COVID-19 activity, we developed an online platform to monitor viral respiratory symptoms among hospital employees.
- Objectives: to describe and analyze risk profiles for Covid-19 testing combining participatory surveillance and laboratory surveillance data.
- Methods: Participatory surveillance, spatial-temporal scanning, anomaly detection, time series forecast and locally regressions.
Digital visual information therapy (DigiVisIn) for orofacial pain
Communication of symptom-related information in nonprofessional terms plays an important role in disease management. The aim of information therapy is to inform patients about aspects of health maintenance and illness development so that they can make informed decisions based on a profound understanding. Imaging in medicine, for example, is not only used for diagnostics (e.g. intrauterine development of the fetus), but also to promote human understanding of complex processes. The didactic value of images is illustrated by the proverb: "A picture tells more than a thousand words".
Persistent complaints in the orofacial area (mouth and face) are mostly due to molecular changes in muscles and nerves, which cannot be visualized by imaging methods. In addition, psychosocial stressors commonly influence somatic symptom perception and vice versa, bodily symptoms negatively impact on psychological well-being. In current clinical practice, verbal explanations and hand-drawn sketches provide information regarding biological mechanisms and bio-psychosocial interactions.
The objectives of this project are the development and implementation of an interactive, digital tool to support patient-centered information therapy (DigiVisIn). DigiVisIn visualizes expert knowledge by means of biological illustrations and graphical models, which are adaptable to the individual needs for information therapy. The structure is modular so that different aspects are presentable based on requirements. Due to the didactic focus on comprehension by layperson, also specialists from various medical and paramedical disciplines with little expertise in the area of orofacial problems are able to convey the necessary information at a high quality level. In particular, the projects aim at empowering students in conveying complex biopsychosocial relationships in a way that they are easily understandable for laypersons. Once approved by the ethics committee, the application will be tested in a clinical context. The DSI application serves as financial bridge for the preparation of an Innosuisse application.
PD Dr. Dominik Ettlin, Prof. Gerhard Schwabe, Dr. Mateusz Dolata, Dr. Markus Wolf
The mental health consequences of COVID-19: Geographic variation and influencing factors
- Assess mental health (along with socio-demographic influencing factors and residence locations) in the Swiss population as a response to the current COVID-19 public health crisis over time (three waves),
- Explore innovative ways to assess emotional stress in the population, based on social media in Switzerland and neighboring countries,
- Investigate geographic variation in influencing factors and mental health from survey responses and from social media,
- Disseminate the findings with an interactive and dynamic geovisualization tool on a website, available to the knowledge users
- Three wave self-administered online mental health survey: symptoms of anxiety, depression, coping, socio-demographics, lifestyle, place of residence, etc.
- Regression models: Associations between influencing factors and mental health outcomes
- EMOTIVE, Stresscapes, LIWC, GIS: Emotional stress and spatial trajectories in Twitter tweets before, during and after the crisis
- GIS, spatial regression, spatial scan statistics: Spatially covarying survey outcomes and emotional stress from social media
Submitted to SNSF
The role of communication during the COVID 19 outbreak
- Understanding the role of COVID-19-related communication during the disease outbreak in Switzerland
- Online survey, March 19 – March 24, 2020 (first week after the lockdown)
- Sample provided by GfK Switzerland, quoted for age, gender, and residential region (N = 1005)
- Information sources/communication channels regarding the coronavirus crisis (news media, social media, personal communication): relevance, functions, evaluation
- Social distancing: behavior, attitudes, perceived efficacy, perceived norms, perceived threat
- Panic buying (“Hamsterkäufe”): behavior, attitudes, perceived norms, presumed influence of media
- Further: trust (e.g., in media, government), neuroticism, conformism
Wie wichtig ist Anwesenheit? Psychotherapie und psychotherapeutische Beziehung im Übergang von Ko-Präsenz zu Tele-Präsenz
- The COVID-19 pandemic has led to an unexpected acceleration of the rise of telemedicine in mental health care. This situation offers a unique opportunity to study therapists’ and patients’ adaptation to a wide-ranging substitution of co-presence by tele-presence. The central concern is the effect of tele- instead of co-presence on the therapeutic relationship – which is one of the main effective factors in psychotherapy.
- Centre for Social Psychiatry, University Hospital of Psychiatry, Zurich
- naturalistic, longitudinal mixed-methods design
- Online questionnaire on therapists’ experience with teletherapy (132 answers so far)
- video-recorded psychotherapy sessions (face-to-face, via phone and via videoconferencing) alongside routine clinical data of the participating patients
- Until now, the possibility to record teletherapy sessions is limited technically and ethically, because sessions are not recorded routinely (primarily for quality control) and because the videoconferencing tool currently used routinely saves records on dropbox (which is not compatible with data safety for patient data)
Semantic activity analytics of social reminiscence in older adults’ everyday conversations
- Reminiscence is the act of thinking or talking about personal experiences that occurred in the past
- Reminiscence is a real-life activity in the WHO healthy aging model: it is essential for healthy aging, serving multiple functions, such as decision-making and introspection, transmitting life lessons, and bonding with others
- real-time prediction of reminiscence in real-life conversations of older adults (in German, English and French),
- design of digital health interventions to improve older adults‘ well being
- gerontopsychology (e.g. reminiscence theory), natural language processing and machine learning
- first publication (German data) submitted to the Journal of Medical Internet Research (Impact Factor: 4.945). Second publication (English data) underway. Planning of the third (French data) underway
- discussions around the submission of an SNF grant in Fall 2019 underway
Digitising historic epidemics - Establishing an interdisciplinary database/tool to learn from infectious disease outbreaks in Switzerland, ca. 1850-1950
- The emergence of epidemics is a challenge for public health.
- Health policy makers benefit from historical experience to increase risk awareness and inform decision making (historical epidemiology).
- The reconstruction of regional nuances is important for understanding how epidemics spread in the population to prevent ecological fallacy.
- However, valuable experiences from the past are not sufficiently accessible for research, policy makers, teaching, and the society.
- The digital society is running danger to be blind to the historical eye, because analogue past information may be forgotten in the archives.
- Two showcase topics with rich and underused information in Swiss archives: Cholera (1850s/1860s) & influenza (1889/1890 & 1918/1919).
- To digitise examples of past outbreak experiences (quantitative and qualitative information)
- To develop an interdisciplinary online dash board (data base and geo-visualisation tool) for
a) researchers (open access to data)
b) teachers/students (MSc projects)
c) science communication (data- and geo-visualisation for policy makers and the society)
- Evolutionary Medicine, Epidemiology, Geography, History, etc.
- Database, R Shiny Tool, webpage, Hist-GIS, geovisualisation, epidemiological modelling, impact of interventions, etc.
- a) Looking for seed money to expand pilot data and to start application process (SNSF, DSI, UZH Lehrkredit, foundations, etc.).
- b) Ongoing MSc thesis on spatial-temporal distribution of the Spanish Flu
Large-Scale Literature-Based Entity Recognition for COVID-19
- We have developed over the years efficient and reliable methods for entity recognition across the scientific literature.
- Recently we applied them to publications about COVID19. The methods can be applied to any text discussing medical or biological aspects of a given disease.
- Our dictionary-based lookup tool OGER is used in conjunction with a pretrained BioBERT model and a vocabulary specific to COVID-19.
- NER and NEN used to be performed sequentially, but performing both simultaneously yields better results.
- ~3000 abstracts per week on PubMed related to COVID-19: We need efficient, reliable tagging of these publications to help health researchers
- Our pipeline tags PMC and PubMed articles for COVID-19 related and other medical concepts, and outputs a variety of formats (EuroPMC, PubAnnotation, BioC…)
Multiple-path literature exploration in the context of COVID 19: LitExplorer
- Context: COVID-19 literature published from January to the middle of May was more than 23,000 papers and doubling every 20 days.
- Problem: Experts across domains have an urgent need for information up-to-date and pertinent to their respective fields but they struggle to keep up with the publication volume.
- Hypothesis: Exploring literature and discovering knowledge in it can be facilitated by an environment that exposes a semantically enriched version of the literature.
- Build an environment for analysis of scientific literature related to COVID-19, which offers various modalities for aggregation of information across multiple papers.
- Our aim is to provide a web tool where COVID-19 literature is enriched by:
- NER annotations (provided by OGER)
- text and NER searches
- a network of semantically connected publications (by sentence similarity)
- extractive summarization
- a reading strategy powered by semantic hyperlinks among sentences.
Social Media Mining for Covid-19
Development of a system that focuses on the analysis of social media-based health data with regard to COVID-19.
Motivation and goals:
- Discovering of emerging preprints and trending drugs
- Monitoring of the distribution of sources and topics of posts
- Detection and tracking of outbreaks
- A subset of 500K tweets from the 361M tweets of the Covid-19 Twitter chatter dataset of the Panacea Lab.
- Calculation of language, hashtag and domain distribution
- LDA topic models
- Calculation of number of posts per day/hashtag
Methods for individual posts and the entire dataset:
- Sentiment analysis by hashtag
- Detection of mentions of paper preprints
- Detection of mentions of drugs
Digital health and the COVID-19 epidemic
- Vokinger et al. (Research Paper): This study aimed to develop a framework to provide guidance on the alignment of a specific app with epidemiological principles to contain the epidemic spread, as well as on compliance with the current legal basis in Switzerland concerning privacy and constitutional rights.
- Method: Literature review, Checklist Development
- Nittas et al. (Viewpoint): Disrupted healthcare systems and the need for physical distancing seem to open a window of opportunity for a broader exposure to telehealth solutions, many of which might have the potential to improve care long after the pandemic passes.
- Method: Literature review
AutoDISCERN: Using AI to assist patients in assessing health web page quality
- Patients increasingly turn to search engines and online content before, or in place of, talking with a health professional. However, the quality of online health information is not regulated. To address this, the NHS developed the DISCERN criteria to evaluate the quality of online health information. An AI version of the tool could run in the background and alert patients when they are viewing low quality information.
- Develop an AI system to evaluate the quality of online health information according to the DISCERN criteria.
- We built an automated implementation of the DISCERN instrument using Natural Language Processing & Machine Learning models.
- The AI models achieved accuracy scores of 81%. In comparison, human raters achieve a manual rating accuracy of 94%. Our research suggests that it is feasible to automate online health information quality assessment, which is an important step towards empowering patients to become informed partners in the healthcare process.