DATA SCIENCE TRAINING

 

New to data science or looking to pick up a few new skills? Don’t miss these free webinars, guided practical tutorials and online resources featuring CANUE data.

Developed in partnership with Population Data BC


AN INTRODUCTION TO DATA SCIENCE – WEBINAR AND TUTORIAL SERIES

REGISTER

MODULE 1: Introduction to Machine Learning
January 15 | 11am to noon pacific: Overview webinar
January 17 | 11am to 1pm pacific: Guided practical tutorial

MODULE 2: Regression and Regularization Algorithms
January 29 | 11am to noon pacific: Overview webinar
January 31 | 11am to 1pm pacific: Guided practical tutorial

MODULE 3: Advanced Supervised Learning
February 12 | 11am to noon pacific: Overview webinar
February 14 | 11am to 1pm pacific: Guided practical tutorial

MODULE 4: Advanced Unsupervised Learning
February 26 | 11am to noon pacific: Overview webinar
February 28 | 11am to 1pm pacific: Guided practical tutorial

Dr. Aman Verma  is a Data Engineer with a PhD in Epidemiology from McGill University, and an undergraduate degree in Computer Science. He has experience in developing machine learning systems with large databases, particularly for scientific data in healthcare. While he’s comfortable learning any programming language, he’s recently become particularly interested in R. Aman is currently involved in a number of projects, including measuring how following opioid prescription guidelines can decrease the risk of opioid overdose, modelling trajectories of chronic obstructive pulmonary disease, and assessing how to best prioritize ambulance calls using secondary healthcare data.

 


AN INTRODUCTION TO DATA MANAGEMENT AND CLEANING FOR ANLAYSIS IN ‘R’  

REGISTER

This self paced free online course will provide you with an introduction to Data Management and Cleaning for Analysis using R Software. Each of the four modules includes a Power Point slide deck, CANUE training data, R code and associated exercises for practice.

Topics covered include:

  • Introduction and theory of data cleaning and management
  • Getting started with R software
  • Subsetting variables and data cleaning
  • Creating variables, subset observations and data cleaning
  • Merging, joining and reshaping data

 

Megan Striha currently works as a Data Analyst. She has a Masters of Public Health degree and three years of experience in health data analysis, including working with survey, administrative and census data.

Spotlight: Marianne Hatzopoulou

Marianne co-leads the Transportation group within CANUE. The main objective of the transportation group is to generate traffic volumes on road networks for various Canadian cities. The team is using various approaches including travel demand and network assignment models as well as statistical interpolation techniques.

Marianne Hatzopoulou is Associate Professor in the Department of Civil Engineering at the University of Toronto and Canada Research Chair in Transportation and Air Quality. Her expertise is in modelling road transport emissions and urban air quality, as well as evaluating population exposure to air pollution. Her research aims to capture the interactions between the daily activities and travel patterns of urban dwellers and the generation and dispersion of traffic emissions in urban environments. She has linked various traffic simulation models with tools for microscopic emission estimates and has published in the areas of traffic emission modeling, near-road air pollution, and greenhouse gas emissions from transport. Prof. Hatzopoulou serves on the National Academy of Science Transportation Research Board committees on “Transportation and Air Quality” and “Environmental Analysis in Transportation”.

University of Toronto profile

January 7 | 2019

Born to be Wise: a population registry data linkage protocol to assess the impact of modifiable early-life environmental exposures on the health and development of children. 

van den Bosch M, Brauer M, Burnett R, Davies HW, Davis Z, Guhn M, Jarvis I, Nesbitt L, Oberlander T, Rugel E, Sbihi H, Su JG, Jerrett M.

BMJ Open. 2018 Dec 14;8(12):e026954. doi: 10.1136/bmjopen-2018-026954

Abstract

INTRODUCTION:

Deficiencies in childhood development is a major global issue and inequalities are large. The influence of environmental exposures on childhood development is currently insufficiently explored. This project will analyse the impact of various modifiable early life environmental exposures on different dimensions of childhood development.

METHODS:

Born to be Wise will study a Canadian cohort of approximately 34 000 children who have completed an early development test at the age of 5. Land use regression models of air pollution and spatially defined noise models will be linked to geocoded data on early development to analyse any harmful effects of these exposures. The potentially beneficial effect on early development of early life exposure to natural environments, as measured by fine-grained remote sensing data and various land use indexes, will also be explored. The project will use data linkages and analyse overall and age-specific impact, including variability depending on cumulative exposure by assigning time-weighted exposure estimates and by studying subsamples who have changed residence and exposure. Potentially moderating effects of natural environments on air pollution or noise exposures will be studied by mediation analyses. A matched case-control design will be applied to study moderating effects of natural environments on the association between low socioeconomic status and early development. The main statistical approach will be mixed effects models, applying a specific software to deal with multilevel random effects of nested data. Extensive confounding control will be achieved by including data on a range of detailed health and sociodemographic variables.

ETHICS AND DISSEMINATION:

The study protocol has been ethically approved by the Behavioural Research Ethics Board at the University of British Columbia. The findings will be published in peer-reviewed journals and presented at scholarly conferences. Through stakeholder engagement, the results will also reach policy and a broader audience.

December 17 | 2018

The influence of social networks and the built environment on physical inactivity: A longitudinal study of urban-dwelling adults.

Josey MJ, Moore S.

Health Place. 2018 Nov;54:62-68. Epub 2018 Sep 21. DOI: 10.1016/j.healthplace.2018.08.016

 

Abstract

Policies targeting the built environment to increase physical activity may be ineffective without considering personal social networks. Physical activity and social network data came from the Montreal Neighborhood Networks and Healthy Aging Panel; built environment measures were from geolocation data on Montreal parks and businesses. Using multilevel logistic regression with repeated physical inactivity measures, we showed that adults with more favorable social network characteristics had lower odds of physical inactivity. Having more physical activity facilities nearby also lowered physical inactivity, but not in socially-isolated adults. Community programs that address social isolation may also benefit efforts to increase physical activity.

 

IN MEMORIAM | FRANCES SILVERMAN

FRANCES SOMMERFREUND SILVERMAN

1942–2018


Frances Sommerfreund Silverman was born in Shanghai, China after her physician parents fled Vienna in 1942, narrowly escaping Hilter’s tyranny. Frances lived in Wuhu, China to the age of 6 before emigrating to Canada where her family settled in Montreal.

Frances enrolled in a doctoral program in respiratory physiology at McGill University in 1968 under the late Professor David Bates, widely recognized one of the founding figures in the field of air pollution and health. After several years of study, Frances moved to Toronto to direct the Pulmonary Function Laboratory at the Gage Research Institute which was at the time, a joint Centre of the University of Toronto (Department of Medicine) and Toronto Western Hospital.

After completing her doctoral work at McGill in 1978, Frances was immediately appointed Assistant Professor in the Department of Medicine at the University of Toronto. Both then and throughout the rest of her career, Frances was proud to be one of a very small group of non-clinical appointees in an otherwise clinical Department.

Frances remained at the Gage Research Institute as an early faculty member in the fledgling discipline of Environmental Health where her research continued to focus on air contaminants, staying true to her first publication in the CMAJ in 1970 – “Problems in studies of human exposure to air pollutants”.

Over the years, her research activities expanded to include many health-relevant air contaminants that remain important today, including ozone, cigarette smoke, allergens and particulate matter arising from industry and motor vehicle emissions. Frances’s earliest work on the health consequences of ozone exposure in the 1970s was formative and continues to be cited regularly. From that and her other insights, she is widely regarded as one of the founding researchers in this area.

The health outcomes she considered also expanded beyond airways measurements to increasingly more sophisticated measures such as genetic and epigenetic markers, inflammatory mediators, and vascular measures. Elegant and highly cited work in the early 2000s by Frances and her colleagues first established a mechanistic link between air pollution exposure and acute cardiovascular events.

Frances held appointments in the Department of Medicine (Division of Respirology), the Dalla Lana School of Public Health (Division of Occupational and Environmental Health), The School of the Environment, the Faculty of Kinesiology and Physical Education, the Li Ka Shing Knowledge Institute, and the University Health Network. She was always most proud of her affiliation with the Gage Research Institute (later the Gage Occupational and Environmental Health Unit), where she served as Acting Director and a member of the Board of Directors.

Despite starting her career as a basic scientist, Frances rapidly understood that truly transformative and impactful research can only be achieved through collaboration. She focused her efforts at the poorly explored nexus that exists between the basic sciences, health sciences and engineering. There, she built a network of collaborators and developed a world-class research program to study air health effects in healthy human subjects as well as those with mild asthma, children and adolescents, and those with chronic obstructive lung disease and obesity using controlled exposure challenges. Using this approach, Frances and her group bridged a critical gap between basic laboratory science and population health, providing much essential evidence needed for policy setting in Canada and abroad in relation to a range of contaminants from environmental tobacco smoke to vehicle emissions. Her work on air contaminants continued well past her retirement in 2012, and she remained actively engaged in research and mentorship until her death. Her curiosity and enthusiasm were infectious, and her level of energy unmatched. “Not bad for an old lady,” she would often observe.

Frances was a networker before networking was a thing, she prioritized the mentorship of young scientists long before it became a trend, and had a preternatural ability to see connections and seed innovative thinking. In her final year, Frances became an advocate for the rights of the elderly to health care access, arising from her own experiences in later life as a caregiver, her deep knowledge of the health care system, her drive to help others, and her talent for building relationships. Despite retirement, she actively mentored students and kept up the schedule of an active faculty member until her last day where her final effort was to advocate tenaciously at a faculty retreat on the importance of the environment as a determinant of health.

Frances was a person of great goodness and integrity – a true Mensch in the Yiddish sense. She continually challenged all who knew her to be better and do better by example. Her spirit, her wisdom, and her generosity will be greatly missed.

 

December 10 | 2018

Residential green space and pathways to term birth weight in the Canadian Healthy Infant Longitudinal Development (CHILD) Study.

Cusack L, Sbihi H, Larkin A, Chow A, Brook JR, Moraes T, Mandhane PJ, Becker AB, Azad MB, Subbarao P, Kozyrskyj A, Takaro TK, Sears MR, Turvey SE, Hystad P; CHILD Study Investigators.

Int J Health Geogr. 2018 Dec 4;17(1):43. doi: 10.1186/s12942-018-0160-x

Abstract

BACKGROUND:

A growing number of studies observe associations between the amount of green space around a mother’s home and positive birth outcomes; however, the robustness of this association and potential pathways of action remain unclear.

OBJECTIVES:

To examine associations between mother’s residential green space and term birth weight within the Canadian Healthy Infant Longitudinal Development (CHILD) study and examine specific hypothesized pathways.

METHODS:

We examined 2510 births located in Vancouver, Edmonton, Winnipeg, and Toronto Canada. Green space was estimated around mother’s residences during pregnancy using Landsat 30 m normalized difference vegetation index (NDVI). We examined hypothesized pathways of: (1) reduction of environmental exposure; (2) built environment features promoting physical activity; (3) psychosocial conditions; and (4) psychological influences. Linear regression was used to assess associations between green space and term birth weight adjusting first for a comprehensive set of confounding factors and then incrementally for pathway variables.

RESULTS:

Fully adjusted models showed non-statistically significant increases in term birth weight with increasing green space. For example, a 0.1 increase in NDVI within 500 m was associated with a 21.5 g (95% CI - 4.6, 47.7) increase in term birth weight. Associations varied by city and were most robust for high-density locations. For the two largest cities (Vancouver and Toronto), we observed an increase in birth weight of 41.2 g (95% CI 7.8, 74.6) for a 0.1 increase in NDVI within 500 m. We did not observe substantial reductions in the green space effect on birth weight when adjusting for pathway variables.

CONCLUSION:

Our results highlight the need to further characterize the interactions between green space, urban density and climate related factors as well as the pathways linking residential green space to birth outcomes.

December 3 | 2018

A picture tells a thousand…exposures: Opportunities and challenges of deep learning image analyses in exposure science and environmental epidemiology.

Weichenthal S, Hatzopoulou M, Brauer M. 

Environ Int. 2018 Nov 22. pii: S0160-4120(18)32200-1. [Epub ahead of print] DOI:10.1016/j.envint.2018.11.042

 

Abstract

BACKGROUND:

Artificial intelligence (AI) is revolutionizing our world, with applications ranging from medicine to engineering.

OBJECTIVES:

Here we discuss the promise, challenges, and probable data sources needed to apply AI in the fields of exposure science and environmental health. In particular, we focus on the use of deep convolutional neural networks to estimate environmental exposures using images and other complementary data sources such as cell phone mobility and social media information.

DISCUSSION:

Characterizing the health impacts of multiple spatially-correlated exposures remains a challenge in environmental epidemiology. A shift toward integrated measures that simultaneously capture multiple aspects of the urban built environment could improve efficiency and provide important insights into how our collective environments influence population health. The widespread adoption of AI in exposure science is on the frontier. This will likely result in new ways of understanding environmental impacts on health and may allow for analyses to be efficiently scaled for broad coverage. Image-based convolutional neural networks may also offer a cost-effective means of estimating local environmental exposures in low and middle-income countries where monitoring and surveillance infrastructure is limited. However, suitable databases must first be assembled to train and evaluate these models and these novel approaches should be complemented with traditional exposure metrics.

CONCLUSIONS:

The promise of deep learning in environmental health is great and will complement existing measurements for data-rich settings and could enhance the resolution and accuracy of estimates in data poor scenarios. Interdisciplinary partnerships will be needed to fully realize this potential.

November 26 | 2018

Association of Prenatal Exposure to Air Pollution With Autism Spectrum Disorder. 

Pagalan L, Bickford C, Weikum W, Lanphear B, Brauer M, Lanphear N, Hanley GE, Oberlander TF, Winters M. 

JAMA Pediatr. 2018 Nov 19. doi: 10.1001/jamapediatrics.2018.3101. [Epub ahead of print] 10.1001/jamapediatrics.2018.3101

 

Abstract

IMPORTANCE: The etiology of autism spectrum disorder (ASD) is poorly understood, but prior studies suggest associations with airborne pollutants.

OBJECTIVE: To evaluate the association between prenatal exposures to airborne pollutants and ASD in a large population-based cohort.

DESIGN, SETTING, AND PARTICIPANTS: This population-based cohort encompassed nearly all births in Metro Vancouver, British Columbia, Canada, from 2004 through 2009, with follow-up through 2014. Children were diagnosed with ASD using a standardized assessment with the Autism Diagnostic Interview-Revised and Autism Diagnostic Observation Schedule. Monthly mean exposures to particulate matter with a diameter less than 2.5 µm (PM2.5), nitric oxide (NO), and nitrogen dioxide (NO2) at the maternal residence during pregnancy were estimated with temporally adjusted, high-resolution land use regression models. The association between prenatal air pollution exposures and the odds of developing ASD was evaluated using logistic regression adjusted for child sex, birth month, birth year, maternal age, maternal birthplace, and neighborhood-level urbanicity and income band. Data analysis occurred from June 2016 to May 2018.

EXPOSURES: Mean monthly concentrations of ambient PM2.5, NO, and NO2 at the maternal residence during pregnancy, calculated retrospectively using temporally adjusted, high-resolution land use regression models.

MAIN OUTCOMES AND MEASURES: Autism spectrum disorder diagnoses based on standardized assessment of the Autism Diagnostic Interview-Revised and Autism Diagnostic Observation Schedule. The hypothesis being tested was formulated during data collection.

RESULTS: In a cohort of 132 256 births, 1307 children (1.0%) were diagnosed with ASD by the age of 5 years. The final sample size for the PM2.5-adjusted model was 129 439 children, and for NO and NO2, it was 129 436 children; of these, 1276 (1.0%) were diagnosed with ASD. Adjusted odds ratios for ASD per interquartile range (IQR) were not significant for exposure to PM2.5 during pregnancy (1.04 [95% CI, 0.98-1.10] per 1.5 μg/m3 increase [IQR] in PM2.5) or NO2 (1.06 [95% CI, 0.99-1.12] per 4.8 ppb [IQR] increase in NO2) but the odds ratio was significant for NO (1.07 [95% CI, 1.01-1.13] per 10.7 ppb [IQR] increase in NO). Odds ratios for male children were 1.04 (95% CI, 0.98-1.10) for PM2.5; 1.09 (95% CI, 1.02-1.15) for NO; and 1.07 (95% CI, 1.00-1.13) for NO2. For female children, they were for 1.03 (95% CI, 0.90-1.18) for PM2.5; 0.98 (95% CI, 0.83-1.13) for NO; and 1.00 (95% CI, 0.86-1.16) for NO2.

CONCLUSIONS AND RELEVANCE: In a population-based birth cohort, we detected an association between exposure to NO and ASD but no significant association with PM2.5 and NO2.

URBAN FORM METRICS FROM HIGH RESOLUTION SATELLITE AND STREET-LEVEL IMAGERY

 

 

 

 

 

 

 

 

We are getting ready to start some intensive work on developing urban form metrics from high resolution satellite and street-level imagery. We will be hosting two virtual meetings to develop a comprehensive list of metrics and applications that are high on the priority list for CANUE members. Your feedback will help us develop a work program for CANUE staff in the coming months.

We will be hosting the same meeting on Dec 4th  (9am to 10:30 am pacific) and Dec 6th (12 noon to 1:30pm pacific).

REGISTER HERE  for your preferred date.

If you are unable to attend on either date, please email info@canue.ca with the following message: “please add me to the working group for machine learning and high resolution imagery”. You can provide feedback prior to the meetings, and we will send a summary report in early January.

November 19 | 2018

The health risks and benefits of cycling in urban environments compared with car use: health impact assessment study.

David Rojas-Rueda, Audrey de Nazelle, Marko Tainio, Mark J Nieuwenhuijsen

 

BMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d4521

 

Abstract

Objective To estimate the risks and benefits to health of travel by bicycle, using a bicycle sharing scheme, compared with travel by car in an urban environment.

Design Health impact assessment study.

Setting Public bicycle sharing initiative, Bicing, in Barcelona, Spain.

Participants 181 982 Bicing subscribers.

Main outcomes measures The primary outcome measure was all cause mortality for the three domains of physical activity, air pollution (exposure to particulate matter <2.5 μm), and road traffic incidents. The secondary outcome was change in levels of carbon dioxide emissions.

Results Compared with car users the estimated annual change in mortality of the Barcelona residents using Bicing (n=181 982) was 0.03 deaths from road traffic incidents and 0.13 deaths from air pollution. As a result of physical activity, 12.46 deaths were avoided (benefit:risk ratio 77). The annual number of deaths avoided was 12.28. As a result of journeys by Bicing, annual carbon dioxide emissions were reduced by an estimated 9 062 344 kg.

Conclusions Public bicycle sharing initiatives such as Bicing in Barcelona have greater benefits than risks to health and reduce carbon dioxide emissions.