Download Social Networks in Human Disease

Social Networks in
Human Disease
Douglas Luke
Introduction to Network Medicine
October 28, 2013
http://cphss.wustl.edu
Goals


The role of ‘above-the-skin’ social networks in
human disease
Three domains




Infectious diseases
Chronic diseases
Network-relevant disease interventions
Future directions


Making the social-physiological link
Frontiers – statistical and computational modeling
of social network disease processes
The brain as a large network:
• 1011 neurons
• 1015 synaptic connections


http://humanconnectome.org/
http://13pt.com/projects/nyt110621/
Biological networks: Food webs
(Wikipedia:
Summerhayes &
Elton’s 1923 food
web of Bear Island)
Social networks
MLK Genogram
Moreno Sociogram
What is the connection between social networks and human
diseases?
CONCEPTUAL MODEL
Social networks and human disease conceptual model
Adapted from Berkman, et al, 2000, SSM
Dynamic version of model
Where has network analysis been
used in public health?
From Luke & Harris (2007) ARPH
Multiple
analytic
approaches
Luke & Stamatakis, 2012, ARPH
Social networks as platforms for contagion
INFECTIOUS DISEASE
Social networks structure
contagious flows

Two broad types of social network theories
(Borgatti, 2011)



Network flow model (pipeline)
Network coordination model (social position,
bonds)
Modern infectious disease control informed
by network analysis

From traditional S-I-R models to network informed
models
First HIV/AIDS network graphic
(Auerbach et al, 1984; Luke & Stamatakis, 2012)
Traditional S-I-R models ignore social
structure
Jun,2002
(http://dimacs.rutgers.edu/Workshops/EpidTutorial)
Traditional S-I-R models ignore social
structure
Assumes random mixing
Jun,2002
(http://dimacs.rutgers.edu/Workshops/EpidTutorial)
Modern epidemiology recognizes
importance of social networks
High school romantic contacts
Peter S. Bearman,
James Moody, and
Katherine Stovel,
Chains of affection:
The structure of
adolescent
romantic and
sexual networks,
American Journal
of Sociology 110,
44-91 (2004).
Need to take social network
structure into account
(Dimitrov & Meyers, 2010)
Sex is scale-free
From Liljeros, et al. (2001). The web of human
sexual contacts. Nature, 411, 907-908.
Risk of infection based on network
properties
From Christley, et al, 2005, AJE
Modeling dynamic infectious
disease network processes
(Dimitrov & Meyers, 2010)
Social networks as environments that promote or inhibit health
behavior and disease risk
CHRONIC DISEASE
Social networks & chronic
disease

Common applications




Cancer, CVD, and other major chronic conditions
Smoking, drinking, and other substance use
Obesity and physical activity
Theoretical challenge


Homophily – tendency for connected actors in a
social network to look like each other (e.g.,
smoking status)
How to disentangle social influence from social
selection as causes of homophily
Social networks implicated in
chronic disease

Primary prevention



Secondary prevention


Social networks related to wide variety of behavioral
risk factors
Smoking, drinking, exercise, breast-feeding, etc.
Peer and family networks can influence cancer
screening
Tertiary prevention

Numerous studies show that social support and size
of social network increase life expectancy after
cancer, heart disease, stroke
From Kroenke, 2006, JCO
From Kroenke, 2006, JCO
General association of social support and
network size with cancer mortality
From Pinquart, 2010, Oncology/Hematology
From Ennett & Bauman, 1993, JHSB
From Alexander, et al., 2001, JAH
From Alexander, et al., 2001, JAH
Theoretical challenge Homophily


Homophily – tendency for people who are
connected in a network (e.g., friends) to be
more similar to each other (e.g., smoking
status); Birds of a feather flock together
Challenge is to disentangle two potential
causes of homophily


Social selection
Social influence
Underlying cause of homophily:
Selection vs. influence
Disentangling peer influence and
selection-obesity
Clustering of obesity (yellow circles) in a social
network (Christakis & Fowler, 2007)
Disentangling peer influence and
selection-smoking
From Hall & Valente, 2007, AB
From Mercken, et al., 2010, Addiction
Developing more effective interventions and treatments that
operate on social networks or use social network information
DISEASE INTERVENTIONS
Disease interventions


Conceptual model
Two types


Direct intervention to the social network itself (1)
Use social network information to enhance an
intervention or disease treatment (2)
Who are the critical players in a
pediatric hospital ward?
Original graphic by Jan Willem Tulp.
Based on Isella, 2011, PLOS One.
Can disease
networks be
modified?
Alcoholics Anonymous - best example of
an effective network disease intervention
From Kelly, et al, 2011, DAD
From Valente, et al., 2003, AJPH
Targeted social distancing design for
pandemic influenza
“For influenza as infectious as the 1957-58 Asian
flu…, closing schools and keeping children at
home reduced the attack rate by >90%.”
From Glass, et al., 2005, EID
Research gaps and methodological advancements
FUTURE DIRECTIONS
Methodological challenges and opportunities



Much early work based on large, self-report surveys
using extremely simple measures of social support and
network size
Current work uses more sophisticated network
visualization and description, using whole network data
New statistical network modeling techniques allows for
sophisticated hypothesis testing


ERGM (exponential random graph modeling)
New frontier is integrating social network information into
computational modeling

ABM (agent-based modeling)
Types of network methods (n = 76
empirical network studies)
Modeling dissemination of Best
Practices in Tobacco Control
From Luke, et al., 2013, HEB
Modeling dissemination of Best
Practices in Tobacco Control
From Luke, et al., 2013, HEB
Interaction of social network
characteristics and tobacco
control messaging
From Hammond, 2006, Brookings Report
Bahr, et al., 2009, Obesity
Encounter types reported by a social network
From Read, et al, 2008, JRSI
Simulated epidemics based on social
networks and isolation strategies
From Read, et al, 2008, JRSI
For more information:
Douglas Luke
http://cphss.wustl.edu
[email protected]