Current standards of brain and mental care often rely on trials of insufficient scale, which not only limits our ability to diagnose, prevent, treat and personalize care but often leads to incorrect conclusions and undesirable results. What tools and data are becoming available via large-scale web-based and mobile applications, and how can researchers, innovators and practitioners connect with these initiatives?
- Chair: Alvaro Fernandez, CEO of SharpBrains, YGL Class of 2012
- Daniel Sternberg, Data Scientist at Lumosity
- Joan Severson, President of Digital Artefacts
- Robert Bilder, Chief of Medical Psychology-Neuropsychology at UCLA Semel Institute for Neuroscience
(👑VVIP ISHAAN ) Russian Call Girls Service Navi Mumbai🖕9920874524🖕Independent...
Big Data Unlocks Brain Care Insights
1. How can Big Data help upgrade
Brain Care?
brain
brain
brain
brain
brain
brain
brain
brain
brain
brain
brain
brain
brain
brain brain
brain
brain
2. Chaired by: Alvaro Fernandez,
CEO of Sharp Brains, YGL Class of 2012
Daniel Sternberg,
Data Scientist at Lumosity
Joan Severson,
President of Digital Artefacts
Robert Bilder,
Chief of Medical Psychology Neuropsychology at
UCLA Semel Institute for Neuroscience
How can Big Data help upgrade brain care?
4. LUMOS LABS, INC.
Daniel Sternberg, PhD
Data Scientist, Lumosity
A Large Scale Approach to Studying
and Enhancing Cognitive
Performance
5. LUMOS LABS, INC.
Why we need scale
Studying the relationship between health and lifestyle factors and cognitive
performance
• Individual effects tend to be small, especially for within-individual
variables
• Unequal sampling from different demographic groups
• Large scale can allow you to estimate the actual functional relationship
between continuous variables (like age) and cognitive performance
Measuring enhancement
• Effect sizes are generally small to medium
• Small RCTs can provide basic validation for a particular cognitive training
intervention, but a larger scale can allow continuous iteration and testing
of smaller changes to the approach in order to maximize efficacy
6. LUMOS LABS, INC.
The Lumosity Dataset
Measures of performance and enhancement
Over 45 million users
Over 1 billion gameplays
Cognitive training games
40+ exercises
Brain Performance Test
An online, repeatable assessment battery
7. LUMOS LABS, INC.
Health and lifestyle
Real world cognition
Social behavior
Personality
Profile
Surveys
IP address
Location estimate
Location-related
covariates
Date of Birth
Gender
Education Level
The Lumosity Dataset
Demographics
8. LUMOS LABS, INC.
Preliminary examples of insights
1. Relationships between lifestyle factors and cognitive
performance
2. Decline in cognitive performance with age and the impact
of training
Sternberg, Ballard, Hardy, Katz, Doraiswamy & Scanlon, 2013
9. LUMOS LABS, INC.
Effects of lifestyle factors on cognitive
performance
• Questions cover a variety of
lifestyle and health variables that
have been shown to affect
cognition
• Survey data for 750,000 users who
took the survey between May 2011
and January 2012
• Looked at how survey responses
related to performance the first
time a user played one of the
following games
Sternberg, Ballard, Hardy, Katz, Doraiswamy & Scanlon, 2013
Spatial memory span
Memory Matrix
Measure = threshold memory span
N = 161,717
males = 65,095 (40.3%), females = 96,662 (59.7%)
mean age = 37.97 yrs. (sd=15.7)
Arithmetic
Raindrops
Measure = Number correct before 3 errors
N = 127,048
males = 53,169 (41.8%), females = 73,879 (58.2%)
mean age = 37.34 yrs. (sd=15.6)
1-back matching task
Speed Match
Measure = Number correct
N = 162,462
males = 65,285 (40.2%), females = 97,177 (59.8%)
mean age = 37.98 yrs. (sd=15.7)
10. LUMOS LABS, INC.
Sternberg, Ballard, Hardy, Katz, Doraiswamy & Scanlon, 2013
B
C
FIGURE 2 | (A) Exercises used in the analysis of the health and lifestyle survey. (B) The effect of reported sleep on game performance. (C) The effect of
reported alcohol intake on game performance (controlling for age, gender, and level of education).
quadratic effects of alcohol for all three tasks. Low to moder- as alcohol intake increased from there. The presence of nega-
Effects of lifestyle factors on cognitive
performance
11. LUMOS LABS, INC.
Preliminary examples of insights
1. Relationships between lifestyle factors and cognitive
performance
2. Decline in cognitive performance with age and the impact
of training
Sternberg, Ballard, Hardy, Katz, Doraiswamy & Scanlon, 2013
12. LUMOS LABS, INC.
Aging and learning
20,000+ users who played each
game at least 25 times
Looked at how baseline
performance and the learning
trajectory in the tasks differed
by age
Normalized game scores based
on a separate dataset of
approximately 1,000,000 users
per game in order to compare
games to each other
Sternberg, Ballard, Hardy, Katz, Doraiswamy & Scanlon, 2013
Spatial memory span
Memory Matrix
Measure = threshold memory span
N = 23,109
males = 11,156 (49.1%), females = 11,562 (50.9%)
mean age = 44.63 yrs. (sd=15.5) range = 18-74
Working memory 2-back
Memory Match
Measure = n correct in 45 seconds
N = 22,718
males = 11,294 (48.7%), females = 11,855 (51.3%)
mean age = 44.59 yrs (sd=15.4) range = 18-74
Arithmetic
Raindrops
Measure = n correct before 3 errors
N = 41,338
males = 19,444 (47.0%), females = 21,894 (53.0%)
mean age = 41.21 yrs. (sd=15.3) range = 18-74
V
Word Bubbles
Measure = n words correct in 3 minutes
N = 107,478
males = 34,339 (31.9%), females = 73,139 (68.1%)
mean age = 38.82 yrs. (sd=14.7) range = 18-74
14. LUMOS LABS, INC.
Change in performance with training
Sternberg, Ballard, Hardy, Katz, Doraiswamy & Scanlon, 2013
15. LUMOS LABS, INC.
Other lifestyle-cognition relationships
that we’re exploring
- Persistence of training gains over time as a function of age
Ballard, Sternberg, Hardy and Scanlon (2012) SfN poster
- Time of day and circadian rhythms
Sternberg, Hardy and Scanlon (2013) EScoNS poster
- Cognitive profiles related to job categories
- Structure of individual learning curves
- Geographic and education-related variables
16. LUMOS LABS, INC.
External Collaborations
We encourage researchers with
interesting research questions
and well-formed analysis plans
to apply to access portions of
our dataset. You can learn more
by visiting
http://hcp.lumosity.com and
clicking on Get Involved.
All data shared with researchers
is de-identified in adherence with
our Privacy Policy and Terms of
Service.
A few of our ongoing data
collaborations:
•
• Effects of cognitive training on
emotion regulation
• Identifying meaningful cognitive
decline in aging populations
• Applying decision-making models to a
multi-alternative forced choice
task, and relationships to personality
• Interactions of age and gender in the
effects of sleep on cognitive
performance
17. LUMOS LABS, INC.
Data Partnerships
We are also interested in exploring partnerships with other organizations to
gain new insights that are only made possible by combining efforts.
A few possible data “mashups”:
- Genetic factors related to cognitive profiles
- Real-time health and activity monitoring
- Cognitive training and academic performance
- Cognitive training and job performance
37. Robert Bilder,
Chief of Medical Psychology
Neuropsychology at UCLA
Semel Institute for Neuroscience
How can Big Data help upgrade brain care?
38. Big Data Get Personal
Robert M Bilder, PhD
Michael E. Tennenbaum Family Professor of Creativity Research, and
Chief of Medical Psychology – Neuropsychology,
UCLA Jane & Terry Semel Institute for Neuroscience & Human Behavior,
Stewart & Lynda Resnick Neuropsychiatric Hospital,
Departments of Psychiatry & Biobehavioral Sciences and Psychology
David Geffen School of Medicine at UCLA, and College of Letters & Science at
UCLA
September 19th, 2013
41. Current Sources: Clinical
Data
• NIH & Academic Research Aggregators
• dbGaP and other NCBI resources
• BD2K (“Big Data to Knowledge”) RFA ($24M/yr)
• Neurosynth, Enigma, ADNI, FITBIR
• Clinically-oriented NPO’s
• TranSmart (tune in tomorrow to hear: Pete Chiarelli, CEO
of One Mind for Research)
• Patients Like Me; MJ Fox Foundation; others
• Affordable Care Act – EMR mandate
• Cross-overs: personal genomics, health aggregators
42. NIH BD2K
summary of major challenges
• Locating data and software tools
• Gaining access to data and software tools
• Standardizing data and metadata
• Extending policies and practices for data and software
sharing
• Organizing, managing, and processing biomedical Big
Data
• Developing new methods for analyzing biomedical Big
Data
• Training researchers for analyzing and for designing
tools for analyzing biomedical Big Data effectively
43. Multilevel Models from
Biology to Psychology:
Mission Impossible?
Bilder RM, Howe AG, Sabb FW
Journal of Abnormal Psychology, 2013 Aug;122(3):917-27.
It might be argued that the task of the psychologist, the
task of understanding behavior and reducing the vagaries
of human thought to a mechanical process of cause and
effect, is a more difficult one than that of any other
scientist.
(D. O. Hebb, 1949, p. xi)
44. Managing assertions about brain-behavior relations
using a neural circuit description framework
Bilder, Howe & Sabb, 2013 - JAP
45. Non-clinical data
• Personal data generation and monitoring
• Active & passive monitoring (non-clinical
apps, GPS, communication [including Web usage and wifi
network usage, cameras, microphones, social net
analysis)
• Brain-training performance and neurofeedback data
• Google, ISP’s & telecoms, automotive
• Macro-monitoring: Energy and resource utilization
• Power grid, water use
• Aerial reconnaissance and other emission sensing
(gases, light, heat, other RF)
46. Implications for brain health
• Lots of data about your traits
(genes, phenotypes, habits, … )
• Lots of data about your past states
(experiences, exposures, performance, …)
• Lots of predictions about your future
(health, wealth, wisdom, …)
• Opportunity: promote behavior change through
action planning in line with values and goals
48. The only thing missing
is an appropriate
conceptualization of the
known universe.
49.
50. Before I share my data …
• How will I benefit personally?
• Health? Wealth? Happiness? Prestige? Fun?
Achievement? Stimulation? Understanding? Security?
• What are the risks?
• Wasting my time
• Wasting my money
• Compromising my privacy
• Exposing myself to marketers
• Exposing myself to scams or piracy
• Exposing weaknesses to insurers or governments
It was not long ago that computers and internet search were controlled by experts
It was not long ago that cognitive testing was controlled by experts
The BrainBaseline platform and consumer based mobile technologies makes once specialized tools available to everyone
The BrainBaseline platform and consumer based mobile technologies makes once specialized tools available to everyone
The data from these tests has been combined with user-generated demographic data to develop the worlds largest normative dataset of self-administered cognitive tests.
Ecologically valid measures of cognitive functionAbility to measure effective of therapeutic and behavior based interventions through high frequency measures meeting construct and test-retest reliability HIV-CNS clinical data analysis indicates BrainBaseline testing instruments exceed traditional clinician administered testingTesting instruments are designed to take advantage of Multiple modalities of interaction via tablet and mobile phone computing interfaces – capacitive touchscreen, camera, multi-touch, audio, voice recording,
These are a handful of collaborations that would be of relevance to Novartis
These are a handful of collaborations that would be of relevance to Novartis
Datasets so large and complex that they become awkward to work with using on-hand database management tools.degree of complexity within the data set amount of value that can be derived from innovative vs. non-innovative analysis techniques use of longitudinal information supplements the analysis Mike2.0DATA IN THE WORLD – ESTIMATES OF 600 EXABYTES CIRCA 2010-2012http://www.quora.com/How-much-data-is-in-the-world-and-what-is-the-rate-of-new-data-being-addedhttp://phys.org/news/2011-02-world-scientists-total-technological-capacity.html AS OF 2011, 295 EXABYTES
So what I am proposing is that we emphasize the crowd-sourcing of big data about human phenotypes.We are just completing work on a massive PHENOMICS study examining cognitive, behavioral, and imaging in a few thousand individuals, and great though that has been it is very expensive and time-consuming. And we still may not have enough data to answer core genetic questions.But if we can crowd-source data acquisition for these phenotypes – we can acquire data on millions of people and dramatically accelerate the revision cycle of knowledge.What we need are systems that people actually want to use and that balance utility to the user with utility to the researcher.This is a model I described in a TEDx talk in 2010, and which aims to provide users with the capacity to gather and organize information about their favorite subjects – themselves!; and then use it to make predictions about their own futures and choose actions that are compatible with their stated goals.The implementation of an engine like this is already being targeted by multiple software firms, for example Huff Po has contracted a local firm to create a GPS for the Sould to help people stay on track.I propose that we use the fabulous tranSmart infrastructure described yesterday by Dr. Athey, to reach out and gather the phenotypes of the healthy, for the good of all.
Thanks very much Dan and One Mind team for giving me an opportunity to share a few words. I’m bob bilder and I love logos, acronyms, and slogans.This slide is a candidate bumper sticker for a program we call BruinBrains. The idea of this program is to use knowledge about the brain to help our students, faculty, and staff use our brains better.Why do I bring this up today?I do so because I think this work is riding the next wave of human evolution – where we use our brains to change our brains - and I believe this can have the single greatest impact on brain and mental health.And I know the one thing you are looking for as you have your first cup of coffee this morning is… more statistics and probability theory!
So we are trying to advance this vision here on our campus, where the rubber meets the road. This slide illustrates a new offering this summer – our summer institute in Brain-Mind-Wellness that brings together Mindfulness practice and theory, integrative east west medicine, and the one I teach is “personal brain management.” We are already gathering web-based cognitive data on people who participate in a Bruin exercise programs, and are developing further platforms to integrate and kick the tires on systems that engage healthy people and encourage their sharing of knowledge about themselves for the common good. We believe this is an important alternative to the for profit organizations that are emerging to help aggregate data from individuals on a massive scale (for example Lumosity for cognitive data, or Facebook for everything else). What we really need are more trusted alternatives, where individual can share these valuable data about themselves for the good of all. I believe 1 Mind 4 Research can do the same – and develop itself as a trusted resource - on a national scale, and I hope we can help.Thanks for listening!
So I hope you’ll write.And also visit us at the Tennenbaum Center for the Biology of Creativity and Like us on Facebook.Thank you.