Social physics deals with the question of how to create organizations that are more cooperative, productive, and creative? The engine that drives social physics is the newly ubiquitous digital data that is becoming available about all aspects of human life. By using these data to build a predictive, computational theory of human behavior we can hope to engineer better social systems.
Emphasize – The goal is to passively collect the data. This gives us a more “honest” insight into the behavior of a person. In Facebook, for example (where people opt-in), people create an image that reflects how they want to be perceived. For Subaru– Naturalistic data
Important – Mentioned friends and family “FunF” (and others) – data on individuals. Gave devices to 100 people in return fo collecting detailed data using these phones. We call this a “social fMRI”
In our experiment we compare different method of getting people to exercise more.
Three conditions: Traditional approach: give direct subsidies to individuals for their action thus internalizing the externalities Let people see how well their peers have been doing Alternative approach: give subsidies to peers for individuals actions thus localizing the externalities (low cost, peer enforcement, emergence of good social norms)
In this case, the best results – when we gave people money based on the performance of their peers!
A cohort of patients with diabetes in an outpatient setting. Goal: to see if having behavior reflected back to the patient via an app can lead to: changes in health-related behaviors and improvements in health status
Continuous Measurements: Passive measurements (via Smartphone): Sleep – inactive phone at various time points Social – numbers of texts and call logs Activity – accelerometer that comes with phone Active measurements (via questions ‘pushed’ to phone): Morning glucose measurement Medication compliance Progress toward goal (1 of 4 - selected by a patient) Smoking Cessation Foot Care Stress Reduction Eating Better
Cogito is a computer assessment of interaction parameters (energy, mimicry, etc) and estimating patient engagement. Better conversations lead to more satisfied, engaged customers and increased revenue. Insurance firm Aetna estimates this saves them $20k/month/seat in deferred costs.
The competition rewards research conducted on the possible ways in which anonymous operating statistics for mobile networks could be used for socio-economic development. Through Data for Development, and in partnership with Orange Côte d’Ivoire, local authorities and regulatory bodies, the Group is considering how the international scientific community can use big data to improve the well-being of populations.
Human Patient Simulation Room is used to replicate an inpatient room, intensive care unit, emergency room, primary care office, labor and delivery, or community center environment with a sophisticated patient mannequin. The Mock Operating Room is configured to support realistic interprofessional operating room team training by interacting with other participants and utilizing a high-fidelity mannequin.
OpenPDS. Give people control over the data. Make them more comfortable “contributing” their data, understanding what they get in return
So this is exactly what SafeAnswers does. It allows researchers and services to ask individual questions in the form of pieces of code. If allowed, these pieces of code are then run against the data, the answer is validated and sent back to the service.
SafeAnswers is design to allows for services to leverage rich user data while respecting their privacy by reducing the dimensionality of the data on the fly to what is the service sufficient statistics, the pieces of information actually needed to provide the service.
mHealth Israel conference_Professor Erez Shmueli_MIT Media Lab_social physics in health_Dec 15, 2014
Social Physics in Health
Erez Shmueli, PhD
Society enabled by Big-Data
Social physics deals withthequestionof
how to create organizations that are more
cooperative, productive, and creative?
The enginethatdrives social physics is
thenewly ubiquitousdigital data thatis
becoming available about all aspects of
By usingthesedata to build a predictive,
computationaltheory of humanbehavior
we can hope to engineer better social
Call Data Records (CDRs)
Spending data (Credit and Debit cards)
Trading network (eToro)
Reality Mining (Living Labs)
Popularity vs. quality
Ties (e.g. trust relationships, reciprocity)
Sensing, Understanding and Shaping Social Behavior
The Friends and Family Living Lab
A revolution in social sciences
Higher measurement rate
Larger sample size
Data: 65 young families for one year, every 5 minutes
Open Source System: 1500+ research teams around the world
Social Influence incentive mechanism is 3.5 times
as efficient as standard incentive mechanism
65 young families, 3 months data
Using social incentives
The SMARTCATCH Living Lab
HistoryPushed questionnairesPassive collection
Social Signals of Engagement
It’s not what you say. It’s how you say it!
Agent Desktop Not Engaged
Balanced and Fluid Interaction
The sooner the better!
80% accuracy classification
The 2013 D4D Challenge – Ivory Coast
Orange contributed the 1st Data Commons
90 Research teams from around the world
20% reduction in infectious disease
Medical Teams’ Social Behavior
Sensing, Understanding and Shaping
The new deal on data
You have the right to possess,
control and dispose your data
Anonymization is hard
Share answers, instead of raw data
Open-source and Auditable
Sharing Social Health Data
Field trial application exploring the capacity for personal data to
predict mental illnesses
Social health metrics calculated from fine-grained sensor data
(calls, SMS, accelerometer, device state)
Collaboration with DARPA and Cogito
Diagnosing veterans with PTSD