Dynamics of the COVID-19 Spread by Modified Graphical Network Analysis
Yigit Aydede & Mutlu Yuksel
abstract: Our objective in this study is to develop a model that can directly use the observed data (tests, cases, or deaths) to recover the time-varying efficacy of local non pharmaceutical interventions by dynamically identified delays. There are several recent examples in the literature that also use positivity rates and case numbers based on multinational panel datasets, where the delays in the effect of mobility changes, however, are not estimated dynamically but assumed to be constant during the entire period of analysis. We have developed and trained a non parametric algorithm inspired by Gaussian Graphical Models that have been recently and extensively used in genomics, finance, neuro imaging, and other fields that require network analysis on high-dimensional data. In our initial application, we have tested our method and compared Montreal, Toronto, and New York. We find that the mobility restrictions are least capable of fighting the spread in Montreal than in Toronto and New York, although the average mean-maximum correlations are similar in these cities. Our counterfactual simulations reveal that the low efficacy of mobility restrictions in Montreal might be related to a lower public sensitivity toCOVID-19 and that the average reduction in mobility relative to the spread might not have been enough in terms of its magnitude and speed compared to other cities
The Role of Environmental Determinants and Social Mobility in Viral Infection Transmission in Halifax
About the Funding:
A collaborative partnership to develop a COVID-19 response strategy has formed among NSHA Research & Innovation (Nova Scotia Health Authority), Dalhousie University, Dalhousie Medical Research Foundation, the QEII Health Sciences Centre Foundation, the IWK Foundation, the IWK Health Centre, the Dartmouth General Hospital Foundation (DGHF) and Research Nova Scotia (RNS). These partners have collectively committed $1.5 million to support the Nova Scotia research community. This research effort will inform the best COVID-19 practices and support healthcare decision making and planning that benefits the population of Nova Scotia.
Proposed study that received the award:
This study will analyze the relationships between COVID-19 transmission rates, meteorological and air quality, and the fluctuations in social mobility in Nova Scotia to allow for better calibration of spatial spread in simulation models that are used to inform policy making.
Dynamics of Social Mobility during the COVID19 Pandemic in Canada
Yigit Aydede, Francisko Begolli , Mutlu Yuksel
Abstract: As the number of cases increases globally, governments and authorities have continued to use mobility restrictions that were, and still are, the only effective tool to control for the viral transmission. Yet, the relationship between public orders and behavioral parameters of social distancing observed in the community is a complex process and an important policy question. The evidence shows that adherence to public orders about the social distancing is not stable and fluctuates with degree of spatial differences in information and the level of risk aversion. This study aims to uncover the behavioural parameters of change in mobility dynamics in major Canadian cities and questions the role of people’s beliefs about how contagious the disease is on the level of compliancy to public orders. Our findings reveal that the degree of social distancing under strict restrictions is bound by choice, which is affected by the departure of people’s beliefs from the public order about how severe the effects of disease are. Understanding the dynamics of social distancing thus helps reduce the growth rate of the number of infections, compared to that predicted by epidemiological models.
Predicting Asthma Pervasiveness and Exacerbations without Surveys in Canada
by Yigit Aydede, Mutlu Yuksel, Andrea Giusto
Abstract: Asthma is one of the most common chronic diseases in Canada that has a profound impact on more than 3.8 million people’s life and the well-being of societies. Lack of accurate and high-frequency real-time surveillance data prevents timely and targeted interventions at the community or individual level. Canadian Chronic Disease Surveillance System, which is the main public health infobase on chronic diseases, provides data on incident and prevalent cases of asthma with long delays as the health care utilization is the only source of information. Thus, oscillations in respiratory symptoms are mostly underreported or reported at the time of health care service utilization, not at the time of exacerbations. The objective of this study is to provide nontraditional back-stage information on asthma that might not be available from traditional surveys. Obtained from Twitter and Google, we test the validity and reliability of high-frequency real-time information by predicting their response to meteorological and air quality data with advance machine learning methods at the provincial and city level.
What can be predicted from a national health survey? Is cancer one of them?
by Yigit Aydede, Mutlu Yuksel, Andrea Giusto
Abstract: A consideration of predicting cancer or any chronic disease using publicly available data is a taunting thought experiment and, in fact, any attempt may seem to be unscientifically overpromising. Canada has a national survey, Canadian Community Health Survey (CCHS), that has been used as a major health surveillance system. With recently available and digitally recorded health care utilization linked to CCHS, it has become obvious that there may be treasures of meaningful information buried in those data, which, however, is way too large and too complex for humans to make sense of. We first create a unique panel data that follows the same person 10 years from 2001 to 2011 by the linked CCHS and the Discharge Abstract Database (DAD). This data records every hospital visit with more than 33450 disease codes in hourly timing. Learning to detect meaningful patterns in large and complex data sets, such as this, with the help of every-increasing memory capacity and processing speed of super computers opens up new horizons. Therefore, we ask: can we predict cancer or any chronic disease by using this data? We first used Public Use Microdata Files (PUMF) of 2012-213 CCHS without the DAD link to see the predictive power of the data when it is used with advance machine learning models. In our first attempt, we identified 26 predictive factors and 71% success in predicting incident of cancer. Can this be used as a recommender system? Our results are promising that, with more and better data that can be achieved with adjusted new survey designs, and the panel data that we built, we can expand the prediction power to new unexplored territories. This study is design to see the predictive power of data that can be extracted by advance machine learning models and to suggest tangible improvements in surveys to help researchers who are willing to take this challenge.