• 501(c)(3) nonprofit organization• Many strategic partners (SFASA, BAES, etc.) • A young organization founded 3 years ago• Membership is FREE• <1% administrative cost. All funding goes to serve the
community• Provide a virtual community for people interested in
data science to exchange ideas and help each other• Organize
• International scientific conferences• Monthly journal clubs• Technical seminars• Career development events• Educational lectures such as legal series
• Actively seeking members, volunteers and new ideas
Any questions, please contact email: [email protected]
November's Onsite EventKeynote Speakers:Dr. Lisa LaVange
Dr. Lisa LaVange•Professor and Associate Chair, Department of
Biostatistics; Director, Collaborative Coordinating
Center; University of North Carolina
Dr. Steven Shakr. Steven Shak
Co-Founder, Chief Scientific Officer, and Chief Medical
Officer, Genomic Health
Dr. Mingxiu Hu. Mingxiu Hu
Senior Vice President, Nektar Therapeutics
:
Featured Sessions:• Innovative Trial Design • Innovative Technology and Applications in Clinical Trials • Reflection on Recent Regulatory Guidance • Successful Examples in Statistical Innovation and Leadership • Machine Learning, Al, Big Data, and Applications in Clinical Trials • Panel Discussion on Innovation, Impact, and Leadership
Featured Sessions:
July ’s Speaker and Topic
Speaker: Dr. Li Zhou, MD, PhD, FACMI
Dr. Zhou is an Associate Professor at Harvard Medical School (HMS) and a Lead Investigator at the Division of General Internal Medicine and Primary Care of the Brigham and Women’s Hospital. Dr. Zhou’s primary research areas include natural language processing (NLP), temporal reasoning, knowledge representation, clinical decision support, and health information systems. Dr. Zhou served as a Senior Medical Informatician at Partners HealthCare Systems for more than 10 years. She has served as Principal Investigator and co-Investigator on many research programs funded by AHRQ, NIH, PCORI, CRICO, etc. Dr. Zhou directs the MTERMS Lab (http://mterms.bwh.harvard.edu/) and has led the design and development of multiple NLP systems.
4
Big Data and AI
in Healthcare
Li Zhou, MD, PhD, FACMI
Associate Professor, Division of General Internal Medicine and Primary Care,
Brigham and Women’s Hospital, Harvard Medical School
Big Data & Artificial Intelligence
in Healthcare
5
Big Data and AI
in Healthcare
How much data is generated every minute?
Source: http://www.iflscience.com/technology/how-much-data-does-the-world-generate-every-minute/
Source: https://www.forbes.com/sites/andrewcave/2017/04/13/what-will-we-do-when-the-worlds-data-hits-163-
zettabytes-in-2025/#5a7fc9f349ab
90% of the data in the world
today has been created in the
past few years.
16.3 zettabytes of data per year
163 zettabytes per year by 2025
(a zettabytes = one trillion
gigabytes)
5 V’ s
Volume
Velocity
Variety
Veracity
Value
6
Big Data and AI
in Healthcare
Big Data in Healthcare
Sources: The “big data” revolution in healthcare”. The Mckinsey & Company. 2013.
https://www.medgadget.com/2018/04/big-data-in-healthcare-market-value-share-of-20-69-with-
cerner-co-cognizant-dell-philips-siemens-and-business-forecast-to-2022.html
“Pools” of healthcare data
Clinical data/genetic data
Claims and cost data
Pharmaceutical research data
Patient behavior and sentiment data
7
Big Data and AI
in Healthcare
Electronic Health Record (EHR)
Source: https://dashboard.healthit.gov/evaluations/data-briefs/non-federal-acute-care-hospital-ehr-
adoption-2008-2015.php#figure1
8
Big Data and AI
in Healthcare
Precision Medicine
200 terabytes = 16 million file cabinets filled with text = 30,000 DVDs
Publicly available on the Amazon Web Services cloud since 2012
Allows any researcher to access and analyze the data at a fraction
of the cost and analyze the data much more quickly
Sources: https://www.nih.gov/news-events/news-releases/1000-genomes-project-data-available-amazon-cloud
http://www.1000genomes.org/; https://allofus.nih.gov/
1000 Genomes Project (2008-2015) – a large freely
accessible public catalogue of human variation and
genotype data
Genomes of 2,504 people across 5 continental regions
All of US (2016- ) - $215 million in funding aimed to collect genetic and health data
from one million subjects.
9
Big Data and AI
in Healthcare
Health Data Networks
Source: Curtis LH, et al. Four health data networks illustrate the potential for a shared national multipurpose big-
data network. Health Affairs July 2014: 33:7.
10
Big Data and AI
in Healthcare
Emerging Technology Breakthroughs
Fourth Industrial Revolution
(Industry 4.0)
Artificial Intelligence
“AI is the new electricity”
Robotics
the Internet of Things
3D printing
Quantum Computing
Nanotechnology
Autonomous Vehicles
Sources: http://www.dbta.com/Editorial/Trends-and-Applications/Powering-the-Internet-of-Things-with-Real-Time-Hadoop-103469.aspx
http://thefutureofthings.com/8973-7-major-advancements-3d-printing-is-making-in-the-medical-field/
https://en.wikipedia.org/wiki/Industry_4.0
13
Big Data and AI
in Healthcare
Pathways to Revolutionize Healthcare
Right living: prevention; informed lifestyle choice
Right care: evidence-based care; identification of
patients at high-risk; personalized medicine
Right provider: most appropriate provider and setting
Right value: increased quality and reduced costs
Right innovation: new knowledge discovery
Source: The “big data” revolution in healthcare”. The Mckinsey & Company. 2013.
14
Big Data and AI
in Healthcare
“Disruptive Dozen: 12 AI technologies
that will reinvent care”
Sources: World Medical Innovation Forum - AI 2018.
https://www.youtube.com/channel/UCauKpbsS_hUqQaPp8EVGYOg
AI at the bedside
A picture is worth a thousand words
Can personal devices improve your health?
Risky business: using EHRs to predict disease risk
Reading the tea leaves of cancer immunotherapy
Bringing “smart” machines to medicine
Harnessing the power of digital pathology
Minimizing the treats of antimicrobial resistance and infections associated
with antibiotic use
Getting back to face time: AI tools that help reduce physicians’ computer use
Disseminating medical expertise to areas that need it most
Next-gen radiology
Melding mind and machine
15
Big Data and AI
in Healthcare
Source: Kalis B, et al. Harvard Business Review. May 2018. https://hbr.org/2018/05/10-promising-ai-
applications-in-health-care#comment-section
16
Big Data and AI
in Healthcare
AI, Machine Learning and Deep Learning
Source: http://houseofbots.com/news-detail/2754-1-a-take-on-deep-learning
18
Big Data and AI
in Healthcare
An artificial intelligence
trained to classify images of
skin lesions as benign
lesions or malignant skin
cancers achieves the
accuracy of board-certified
dermatologists.
Esteva A, Kuprel B, Novoa RA, Ko J,
Swetter SM, Blau HM, Thrun S.
Dermatologist-level classification of skin
cancer with deep neural networks.
Nature. 2017;542(7639):115-8.
26
Big Data and AI
in Healthcare
Natural Language Processing (NLP)
NLP lies at the intersection of artificial intelligence and linguistics
NLP aims to create intelligent agents to understand and manipulate human
languages
Any system that analyzes or synthesizes text or speech (e.g., voice to text;
text to structured data, translations, text analytics)
NLP is in High Demand
The global NLP market will be worth $13.4 Billion by 2020.
The global healthcare NLP market is estimated to be worth $2.7 billion by 2020 and
$4.3 billion by 2024, growing significantly from the $936 million reported in 2015.
The market is projected to rise at a compound annual growth rate of 18.8%.
(EHR market is expected to reach $23.98 billion globally in 2020)
source: https://hitinfrastructure.com/news/healthcare-natural-language-processing-expects-steady-growth
27
Big Data and AI
in Healthcare
A significant portion of biomedical information is
stored in textual (unstructured) form
Electronic Health Records, such as clinic notes, progress notes, radiology reports, pathology reports, discharge summaries, free-text entries and comments
Patient Health Record/Gateway
Biomedical literature, such as journal articles and abstracts
Social Media
Others, such as emails, guidelines, books, and surveys
29
Big Data and AI
in Healthcare
Clinical Document-
ation
Info
Extraction Encoding
Data Mining
Knowledge Discovery
Clinical Decision Support
Research, Innovation, and others
• Information reconciliation
o medication, problem
and allergy
reconciliation
• Predictive models
o High risk
o Hospital readmission
o Mortality
• Diagnosis/treatment
• Quality measures
• Clinical information
o Problems, medications,
allergies, socio-
behavioral info,
functional status
• Contextual information
o Family histories,
temporal information,
negation
• Standard/ interoperability
o Terminology encoding
o Information modeling
• Document classification
• Clustering
• Relation identification
• Topic modeling
• Active learning
• Dictation: speech recognition
• Document quality: misspelling checker
• Template / Similarity
• Summarization• New methods development
• Application development
• Data and knowledge sharing
• Other areas: pharmacovigilance
Medical NLP - Ecosystem
NLP
(MTERMS)
http://mterms.bwh.harvard.edu/mterms/© 2018 Li Zhou
30
Big Data and AI
in Healthcare
Additional Information Extracted from
Notes using NLP
Navathe AS, Zhong F, Lei V, Chang F, Sordo Sanchez M, Navathe S, Topaz M, Rocha RA, Zhou L Hospital
Readmission and Social Risk Factors Identified from Physician Notes. Health Serv Res. 2017 Mar 13.
32
Big Data and AI
in Healthcare
Speech Recognition
SR becomes the bridge between human-machine interaction
33
Big Data and AI
in Healthcare
Ways to Document in EHR
Optical
Character
Recognition
Speech
RecognitionOnsite
ScribesTranscriptionistsTyping Remote
Scribes
© 2018 Li Zhou
Zhou L, Blackley SV, Kowalski L, Doan R, Acker WW, Landman AB, Kontrient E, Mack D, Meteer M, Bates DW, Goss
FR. Analysis of Errors in Dictated Clinical Documents Assisted by Speech Recognition Software and Professional
Transcriptionists. JAMA Network Open. 2018 Jul 6;1(3):e180530-.t
36
Big Data and AI
in Healthcare
Language Processing & Diseases
Christie is the best-selling novelist of all time. Her works come third in the rankings
of the world's most-widely published books, behind only Shakespeare's works and
the Bible.
37
Big Data and AI
in Healthcare
Linguistic Analysis Claims Agatha
Christie Had Alzheimer’s
Compared a selection of Christie’s novels between the
ages of 28 and 82, counting numbers of different words,
indefinite nouns and phrases used in each
Statistically significant drops in vocabulary and increases in
repeated phrases and indefinite nouns
o A book she wrote aged 81 showed 30% fewer word
types than another book she wrote aged 63, 18% more
repeated phrases, and almost three times as many
indefinite words
These language effects are recognized as symptoms of
memory difficulties associated with Alzheimer's disease
41
Big Data and AI
in Healthcare
AI vs. Clinician
Will some of our clinicians be out of jobs?
How to integrate AI solutions into clinical practice?
What are the legal liabilities if the machine makes mistake?
How much will we rely on AI?