CLOUD COMPUTING FOR MEDICAL RESEARCH AND HEALTHCAREYu-Chuan (Jack) Li, M.D., Ph.D., FACMI

Graduate Institute of Biomedical Informatics

College of Medical Science and Technology

Taiwan Medical University

Taipei Medical University (TMU)

• Top private medical university in Taiwan• 6000 students, 620 faculty members, 7 colleges• Closest to the world’s highest building – Taipei 101

• Largest JCI-Accredited teaching hospitals in Taipei• 3,150 beds• Over 10,000 Out-patient visit per day

3

北醫附醫 萬芳 雙和

TMU Healthcare Group

Taipei Medical University

4

Total Faculty6,102

Students: 6,059

Alumni: 31,214

Full-time Instructor 428

Part-time Instructor 649

7 Colleges13 Departments

16 Graduate Institutes

3 TMU Hospital

s

College of Medical Science and Technology - TMU

• Department of Biomedical Informatics• 80 master and Ph.D. students

• Department of Medical Technology• 60 master and Ph.D., 300 undergraduate

students• Department of Cancer Biology and Drug

Discovery• Ph.D. only

• Department of Neuro-regenerative medicine• Ph.D. only

Medical Cloud

Care Cloud

Wellness Cloud

CitizenHealthRecord

Clinical Care

Long-term Care

Prev

entio

n

NIST Definition v.15

• Cloud computing is a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.

Five Characteristics of Cloud

• On-demand self-service

• Broad network access

• Resource pooling

• Rapid elasticity

• Measured Service

“the kind of service that dry-lab biomedical researchers would always wanted…”

Other Terms related to Cloud• Service Model

• Cloud Software as a Service (SaaS)

• Cloud Platform as a Service (PaaS)

• Cloud Infrastructure as a Service (IaaS)

• Deployment Model

• Private cloud TMUH as an example

• Community cloud

• Public cloud

• Hybrid cloud

Private Cloud in TMUH

10 Gb10 Gb

2 Gb2 Gb2 Gb2 Gb2 Gb 2 Gb2 Gb 2 Gb

Fiber 10GbFiber 10Gb

Core Switch Engine

HIS主機 HIS主機

Core Switch Engine

Catalyst Express 500 Series

1 2

SETUP

SYSTEM

ALERTPOE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Catalyst Express 500 Series

1 2

SETUP

SYSTEMALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

100 Mb100 Mb

各機櫃之 Switch

100 Mb100 Mb 100 Mb100 Mb 100 Mb 100 Mb

住院Client

住院Client

門診Client

門診Client

行政Client

行政Client

醫療Client

醫療Client

連接至第一、二大樓網路主幹交換器

1

2

3

4

5

6

FANSTATUS

Power Supply 1 Power Supply 2

Catalyst 6500 SERIES

WS-SUP720-3BXL

SUPERVISOR 720 WITH INTEGRATED SWITCH FABRIC/PFC3BXL

EJECT

DISK 0

EJECT

DISK 1

CONSOLE PORT 2

PORT 1

W S - S U P 7 2 0 - 3 B X L

S U P E R V I S O R 7 2 0 W I T H I N T E G R A T E D S W I T C H F A B R I C / P F C 3 B X L

S YS TEM

S TA TU S

A C TIVE

P WR

M G M T E J E C T

D I S K 0

E J E C T

D I S K 1

C O N S O L E P O R T 2

P O R T 1

LI NK

LI NK

LI NK

R ES ET

WS-X6748-SFP

48 PORT GIGABIT ETHERNET SFP

STATUS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

WS-X6748-SFP

48 PORT GIGABIT ETHERNET SFP

STATUS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

1

2

3

4

5

6

FANSTATUS

Power Supply 1 Power Supply 2

Catalyst 6500 SERIES

WS-SUP720-3BXL

SUPERVISOR 720 WITH INTEGRATED SWITCH FABRIC/PFC3BXL

EJECT

DISK 0

EJECT

DISK 1

CONSOLE PORT 2

PORT 1

WS-SUP720-3BXL

SUPERVISOR 720 WITH INTEGRATED SWITCH FABRIC/PFC3BXL

S YS TEM

S TA TU S

A C TIV EP WR

M G M T EJECT

DISK 0

EJECT

DISK 1

CONSOLE PORT 2

PORT 1

LI NK

LI NK

LI NK

R ES ET

WS-X6748-SFP

48 PORT GIGABIT ETHERNET SFP

STATUS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

WS-X6748-SFP

48 PORT GIGABIT ETHERNET SFP

STATUS

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Catalyst Express 500 Series

1 2

SETUP

SYSTEMALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Catalyst Express 500 Series

1 2

SETUP

SYSTEM

ALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Catalyst Express 500 Series

1 2

SETUP

SYSTEMALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Catalyst Express 500 Series

1 2

SETUP

SYSTEM

ALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Catalyst Express 500 Series

1 2

SETUP

SYSTEMALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Catalyst Express 500 Series

1 2

SETUP

SYSTEM

ALERT

POE

1 2 3 4 11 129 107 85 6 13 14 23 2421 2219 2017 1815 16

16X

13X

14X

23X

24X16X

1X

2X

11X

12X

Save 50-70% IT Cost Power saving 30% More efficient and effective

for resource applications

Non-Stop Intensive OLTP-type

Double-loop systemMore Security

Virtual MachinesMore Green

Business model

(Duplicability & profitability)

Operation model

(Sustainability)

Service model

(Feasibility & Value)

Me

dica

l C

loud

Care

Cloud

Wellness C

ould

CARE CLOUD

Care Cloud•Service model yet to be determined

•Need more evidence Systems Clinical Trial Insurance

•Useful for the aging population•HIT development – Taiwan Experience

Night time fall - bed occupancy

Day time fall

Lifeline PNC Monitoring Centre

Carer

Bed Sensor

Night time fall - bed occupancy

Telecare, UKExample: Telecare Helping Manage Falls Risks

West Lothian

22 min fall response

4hour Scotland average

15

16

Telehealth Service Model

Community Care

NHII Information Platform

Institutional CareHome Care

Telehealthcare Service Center(TSC)

Emergency care center

Telehealthcare Information Platform (TIP)

LTC managementcenter

0800-008-850

members

17

Telecare Service Center

A Randomized control trial for Cloud-based Blood Pressure Monitoring

Wireless Sphygmomanometer

Fully Integrated with CPOE

• 系統整合 - 導入醫院與醫令系統

WELLNESS CLOUD

Wellness Cloud•High possibility of Innovation

• e.g. +Social Networking• “Hey, it’s cool to stay healthy!”

• e.g. +gym, wellness, travel, sports and food industry

• e.g. + consumer electronics

Nike+ and iPhone, iPod

24

Mass Gathering Health Monitor

• 101 系統圖

25

CLOUD COMPUTING FOR MEDICAL RESEARCH

27

Personalization Participation

Prediction Prevention

4P Medicine

Clinical Data Repository (CDR)

• Often a backend of

Electronic Health

Record (EHR) for

efficient patient-

level data access

• High potential for

clinical research

From CC, NIH

From CC, NIH

NHIRDB• NHIRDB (National Health Insurance

Research Database)

• 12 years of de-identified claim database for 23 million people

• Cohort DB (Five 1-million people groups for 13 years)

• Disease-specific DB (16 disease groups)

• Random sample DB (outpatient 1/500, inpatient 1/20)

• generates >100 research papers a year

Biostatistics Value-add Cloud

What is MEUC (pronounced as “Milk”)

• MEUC (MEDICAL END-USER COMPUTING), an system that provide end-users a simple to use interface to access “Big Data” for biomedical research in a Cloud Computing framework.

Nutrition to researchers(Proudly made in TMU)

MEUC (Medical End-User Computing System)

The Size of MEUC

Year OPD IPD ER Drug Exam Lab Procedure

2007 2,810,380 84,415 164,313 8,377,165 878,392 12,185,057 963,163

2008 3,187,302 102,480 185,755 9,635,517 1,150,302 13,445,714 1,032,721

2009 3,666,632 164,355 252,414 11,370,756 1,368,141 11,435,334 1,274,819

2010 3,058,730 73,759 184,890 10,149,577 694,654 9,440,527 1,370,797

2011 3,580,217 80,253 188,372 10,607,887 769,530 7,929,639 1,442,294

Total 16,303,261 505,262 975,744 50,140,902 4,861,019 54,436,271 6,083,794

16M 505K 975K 50M 4.8M 54M 6M

DDQ-Disease and Disease Q value 

DDQ-Disease and Disease Q value• From NHI Database 2000 、 2001 and 2002 Data.

• Visits Count more than 230(total population 1/100,000) Disease , It means exclude rare diseases , we can get 4,000 disease.

• 16,000,000 Q value of each Group of 20 Group.

CTM

Hospitals

URS(non-radiology images, signals

and report)PIS

(Pathology Information System)

EMR (Electronic

Medical Record)

PACS(Radiology

images and report)

LIS(Lab Information System)

MEUC (Medical End-User

Computing)

CBIS(Bio-repositoryInformation

System)

Tier 2 & 3Bio-repository

Bioinformatics Platform

Drug-DiscoveryPlatform

IAS(IntelligentAcquisitio

nSystem)

Translational Bio-Informatics Road Map

Tier 1 & 0Bio-repository

EMR Integration Gateway

CPOE(Computer Physician Order Entry)

PCS(Proactive Consent System)

from Prof. Yu-Chuan (Jack) Li, 2009-12, last revised 2012-07-24

Translational Bio-Informatics Road Map

from Prof. Yu-Chuan (Jack) Li, 2009-12, last revised 2012-07-24

Translational Research Cloud

Researchers

8 Center of Excellencefor Cancer Research

National Cancer InstituteNCI

Lab & Exam DB

Exome, SNP, Microarray, Proteomic 2D page…etc.

Radiology, Pathology(MTA...)

Nuclear Medicine Image

Therapeutic Information DB

Data Mining Tool Box

Systems Epidemiology for Cancer Research

CaBIG Gateway

Secondary Cancer Research Database

Master Patient Index (Pseudo ID)

PrivatePublicLimited

Electronic Medical Record

NHIRDB Cancer Data

Tissue Bank Data

Cancer Registry Data

Cancer Screening Array Data

Bio-signal DB

Privacy & Security Firewall

UI, Access

Control

Challenges• ELSI and privacy issues

• IRB or no IRB or a Joint IRB• More complex with more partners

• Lack of data standard for EMR data• caBIG, caDSR (Cancer Data Standards Registry and

Repository) for hosting and managing metadata)

• Scale and complexity of EHR• High demand of computational resource to maintain

multiparty private computation (shared results without sharing data)

• Not RCT clinical trials (but much more available)

CLOUD COMPUTING FOR HEALTH CARE

Medical Cloud

Cloud Computing for Major Hospitals• Virtualization (IaaS)

• Cut the maintenance cost in half

• Much less server room space and much greener

• More flexibility and portability of services

• Service-Oriented Architecture (SaaS)

• An overhaul of the basic system architecture

• Highly flexible and efficient new architecture

• Fast deployment of new applications

• Web-friendly

Cloud Computing for Healthcare

Current

Future

From Service Provider

Cloud Computing for Personalized Health Care

• PHR or ePHR (electronic Personal Health Record) is at the the core of the Next-Generation Health IT

• PHR =• Personal part of the EMR from all the providers • + self-measured bio-signals • + self-entered health related information like family history,

exercise, food consumption, food allergies, OTC drugs, cigarette consumption…etc.

• PHR will be the basis of Personalized Healthcare

EMR

PHR

EMR vs. PHR

Intakerecord

Exercise record

BP, Glucose, ..

etc.

others

Personalized Medicine

• It is estimated in 2014, a personal

Genome can be sequenced under

$1,000 USD

• 3 billion DNA and 33K genes

more than 100K proteins

metabolic pathways

all the functions of body

Some estimated 4GB to store all the

short-reads (before compression)

Personalized Medicine (cont.)

• Need a place to • store it• review it• make sense out of it by linking to a person’s health information

• PHR on the Cloud will be the ideal place

Cloud Computing for Rural Heath Care

• The lack of IT resource of rural health stations and small hospitals in China (>16,000)

• Pioneered since 2007 by Steve Chan (designer of Cray-2)

• Deployed into two medium-sized cities in the western part of China (200 health stations, 800 care-givers)

• A new sustainability model for HIT in developing countries and resource-poor areas

Conclusion• Cloud Computing will change the face of Biomedical

Research Data Service

• Cloud Computing, with privacy-enhanced, could change the future of HIT delivery in developing countries and resource-poor areas

• Fits the needs of many healthcare sectors due to flexibility and cost-effectiveness

• Cloud Computing will be a “liberator” for scalability/accessibility limitations

Thank you for your attention