1. Improve Public Health
AI/ML Week 2019
Balaji Iyer, Global Business Development Manager, AI and ML
March 27th 2019
2. Rising importance of healthcare data
Payment reform (MACRA) mandates collection, analysis and reporting of quality
data for incentives, with penalties for non-adherence
Increasing specificity of digital radiology and pathology solutions and use of
personalized medicine including genomic data
Insights are increasingly dependent upon incorporation of richer, non-clinical
data sets, stored across different types of applications and data structures
Healthcare data is 100X the value of other personal information* making it a
target for cyberattacks like “Wannacry”
* Sources: Institute for Critical Infrastructure Technology
3. Security for Healthcare on AWS
Over 50 global
compliance
certifications and
accreditations
Benefit from AWS
industry leading
security teams 24/7,
365 days a year
Security infrastructure
built to satisfy military,
global banks, and other
high-sensitivity
organizations
Leverage security
enhancements from
1M+ customer
experiences
4. Storage and Archiving Core Operations and
Business Continuity
Care Coordination
Patient Engagement Clinical and Population
Health Analytics
Clinical
Information Systems
Where AWS Supports Healthcare
6. Healthcare data’s importance highlights specific priorities
Protecting Patient
Information
In 2016, 96 healthcare providers
reported a PHI breach, which was a
320% increase from the
previous year1
1 Source: Healthitsecurity.com, 2017, 2IDC Health Insights
Delivering on Data Needs
With 80% of data to go through the
cloud by 20202, healthcare IT needs
flexible, scalable infrastructures to
deliver insights
Business Continuity
Data resiliency is critical for business
continuity and healthcare providers
have the ultimate responsibility to
provide it
8. Healthcare Use cases
Patient & population
health analytics
Personalized Medicine
Clinical Trial
managementRevenue cycle
management (Medical
Coding)
Pharmacovigilance
PHI Compliance
Predictive Medicine
9. M L F R A M E W O R K S &
I N F R A S T R U C T U R E
A I S E R V I C E S
R E K O G N I T I O N
I M A G E
P O L L Y T R A N S C R I B E T R A N S L A T E C O M P R E H E N D
& C O M P R E H E N D
M E D I C A L
L E XR E K O G N I T I O N
V I D E O
Vision Speech Chatbots
A M A Z O N
S A G E M A K E R
B U I L D T R A I N
F O R E C A S TT E X T R A C T P E R S O N A L I Z E
D E P L O Y
Pre-built algorithms & notebooks
Data labeling (G R O U N D T R U T H )
One-click model training & tuning
Optimization (N E O )
One-click deployment & hosting
M L S E R V I C E S
F r a m e w o r k s I n t e r f a c e s I n f r a s t r u c t u r e
E C 2 P 3
& P 3 d n
E C 2 C 5 F P G A s G R E E N G R A S S E L A S T I C
I N F E R E N C E
Reinforcement learningAlgorithms & models ( A W S M A R K E T P L A C E
F O R M A C H I N E L E A R N I N G )
Language Forecasting Recommendations
13. Analyzing Unstructured Text in Healthcare
1.2 B unstructured clinical documents created per year
Critical information “trapped” in these documents
Difficult to extract insights
14. Accurately extract health information from patient
notes, clinical trial reports, and other electronic
health records using Amazon Comprehend
15. • Lipitor: Brand Name
• 20 mg: Dosage
• Once Daily: Frequency
Amazon Comprehend Medical
M r . S m i t h i s a 6 3 - y e a r - o l d
g e n t l e m a n w i t h c o r o n a r y
a r t e r y d i s e a s e a n d
h y p e r t e n s i o n . C U R R E N T
M E D I C A T I O N S : t a k i n g a d o s e
o f L I P I T O R 2 0 m g o n c e d a i l y
• Mr. Smith: Name
• 63 : Age
• Coronary artery : System Organ Site
• Coronary artery disease: Diagnosis Name
• Hypertension: Diagnosis Name
Relationship
extraction
16. Amazon Comprehend Medical
Entities
• Medication
• Medical condition
• Test, Treatments and Procedures
• Anatomy
• Protected Health Information (PHI)
Relationship Extraction
• Medication and dosage
• Test and result
• Many more
Entity Traits
• Negation
• Diagnosis, Sign or
Symptom
Protected Health Information Identification
(PHId API)
Distill a complex process into a simple API call
Medical Named Entity and
Relationship Extraction (NERe API)
17. Amazon Comprehend Medical
Extract text and data from virtually any document
K E Y F E A T U R E S
Medical
conditions
Anatomy
entities
PHI
identification
Medication and
dosage extraction
No ML experience
required
18. Patient & Population Health Analytics
Challenges Amazon Comprehend
Medical
Outcomes
Unstructured data is difficult to mine
Example: Clinical team in the ICU makes
over 120 decisions about care per day, how
do you keep up?
Create ”single-lens” on a single
patient
19. Revenue Cycle Management – Medical Coding
Challenges Amazon Comprehend
Medical
Outcomes
Process of coding or classifying
patient records according to the
International Classification of
Diseases (ICD) is one of the most
complex transactions
Impact coding efficiency and
reduce burden on clinical
staff
20. Clinical Trial Management
Challenges Amazon Comprehend
Medical
Outcomes
Identify the right patients for
clinical trials quickly
Allow for quick and
accurate indexing across
large patient
populations
22. PHI Compliance
Challenges Amazon Comprehend
Medical
Outcomes
Difficult to maintain HIPAA
compliance and technical
requirements for PHI
Accurate way to create
inventory of sensitive PHI
23. OCR++ service to easily extract text and data from
virtually any document. No ML experience required.
24. Amazon Textract
TEXT method. Then, the proteins were clustered
using the k- medoids method with the
optimal number of clusters.
The performance of the various clusterings
was evalu- ated using two types of measures.
The first is the average silhouette width itself,
which is a measure of the clus- ter
compactness and separation. In general,
clustering is based on the assumption that
the underlying data form compact clusters of
similar characteristics. Larger aver- age
silhouette width means that the result of a
clustering algorithm consists of compact
clusters which are well sep- arated from each
other, i.e. probably close to the actual data
distribution. A small average silhouette width
means e.g. that one of the clusters …
26. Graceland, Memphis
Presley, Elvis Aaron
TCB Limited
12-12-1234
TN
01 08 1935 X
901 987-6543
3765 Elvis Presley Blvd.
38116
X RCA Records
Rock n Roll Health
X
Presley, Elvis Aaron
Government forms (e.g. FDA new drug
application, financial disclosure form,
incident reporting)
Tax forms (US – e.g. W2, 1099-MISC, 990,
1040; UK – e.g. P45; Canada – e.g. T4, T5)
Amazon Textract forms
28. Deep Learning for Pulmonary Nodules in
Lung Cancer
• Deep learning algorithms assess the
malignancy risk of pulmonary nodules
based on factors such as nodule size,
shape, density, volume, as well as patient
demographics.
• Use AWS Deep Learning AMI and the
TensorFlow machine learning framework
to train computer vision algorithms for CT
scans.
31. Deep Learning to Detect Coronary Artery
Disease
• Accelerated by GPUs, HeartFlow’s
solution analyzes CT scans to create a
3D model of a patient’s heart and
coronary arteries
• In addition to creating an accurate 3D
model, the system simulates the flow of
blood in each vessel
• Uses the Caffe deep learning framework
on P2 instances; exploring TensorFlow
on G3
33. Advancing Cardiac Visualization with Deep
Learning
• Enabled seamless visualization of 3D
medical images and solved limited
computation power available to doctors
today by moving from CPU to GPU
computing.
• Reduced time for medical imaging analysis
from 30 minutes to seconds using deep
learning and Amazon G2 and S3.
34. Early Detection of Diabetic Retinopathy
Analyzed more than 80,000
fundus photos captured during
retinopathy screenings.
Student Researchers used AWS
EC2, S3, and EBS to manage,
analyze, and review the many
gigabytes of data.
35. Machine Learning
in Healthcare w/ Voice & Data
Canary Speech uses GPU-accelerated deep
learning to detect signs of brain injury by
analyzing voices for vocal tremors, slower
speech and gaps between words.
Used Tesla K80 GPU accelerators on AWS and
a speech recognition tool developed in-house.
36. “Amazon Lex represents a great
opportunity for us to deliver a better
experience to our patients. Everything we
do at OhioHealth is ultimately about
providing the right care to our patients at
the right time and in the right place.
Amazon Lex’s next generation technology
and the innovative applications we are
developing using it will help provide an
improved customer experience. We are just
scratching the surface of what is possible,”
Michael Krouse
Senior Vice President and CIO – Ohio Health
“ • Delivers Personalized Care
Recommendations
• Makes Customer
Appointments
• Drives Urgent Care Referrals
OhioHealth – Lex Integration
37. Next Steps for Building a Pilot
1. Get Started:
https://aws.amazon.com/blogs/architecture/store-protect-optimize-your-healthcare-data-with-aws/
ml.aws
2. Get Trained:
https://aws.amazon.com/training/learning-paths/machine-learning/
3. Start building: https://aws.amazon.com/getting-started/
38. “Why Not?”
"The common question that gets asked in
business is, 'why?' That's a good question, but
an equally valid question is, 'why not?'" - Jeff
Bezos
Thinking small is a self-fulfilling prophecy. Leaders create and
communicate a bold direction that inspires results. They think
differently and look around corners for ways to serve customers.