IT SUPPORT FOR THE AUSTRALIAN SCHIZOPHRENIA

RESEARCH BANK

F. A. Henskens

1,2

, C. M. Loughland

1,2

, M. S. Aphale

1,2

, D. Paul

, J. M. Richards

, P. Rasser

V. J. Carr

1,2

, S. V. Catts

2,4

, A. Jablensky

2,3

, P. T. Michie

1,2

, B. J. Mowry

2,4

, C. Pantelis

2,5

U. Schall

1,2

and R. J. Scott

1,2

University of Newcastle, NSW 2308, Australia;

Schizophrenia Research Institute, NSW 2010, Australia

University of Western Australia, WA 6009, Australia;

University of Queensland, Queensland 4072, Australia

University of Melbourne, Victoria 3010, Australia

Keywords: Globus grid, Web services, Certificate-based security, JSP, ASRB, Schizophrenia.

Abstract: Schizophrenia represents one of the most perplexing and challenging problems confronting both researchers

and health-care providers today. An Australian coalition of researchers has commenced construction of a

resource termed the Australian Schizophrenia Research Bank (ASRB), which intends to support a wide

spectrum of schizophrenia research studies. Software has been written to implement of a series of clinical

assessment instruments, and a purpose-built Globus Grid constructed to provide secure aggregation and

storage of the collected data. This paper describes the design, technology selections and implementation of

the extant computer infrastructure, and discusses planned extensions and enhancements.

1 INTRODUCTION

Schizophrenia represents one of the most perplexing

and challenging problems confronting both

researchers and health-care providers today. With a

prevalence of 4 per 1,000 (Saha et al., 2005),

significant advances in schizophrenia research have

been limited by the difficulty in achieving

sufficiently large samples of probands and

unaffected first degree relatives, in order to study the

causal role of multiple genetic factors each of small

effect. To overcome this limitation, a national (to

Australia) coalition of researchers has commenced

construction of a resource, termed the Australian

Schizophrenia Research Bank (ASRB). The

National Health and Medical Research Council

(NHMRC), the Schizophrenia Research Institute

(SRI), The Pratt Foundation, and other benefactors

fund the ASRB, which intends to support a wide

spectrum of schizophrenia research studies for

Australian and international investigators.

The specific aims of the ASRB include:

1. To develop a research bank with

comprehensive, cross-referenced data from

a large sample (initially 2,000) of

volunteers with schizophrenia and healthy

controls (initially 2,000). The data collected

includes clinical and cognitive

characterisation, blood banking involving

DNA extraction and establishment of

immortal cell lines and MRI brain scans of

schizophrenia using standardised imaging

and analysis methodology

2. To establish infrastructure that makes the

data set available to Australian and

international researchers in a form that is

comprehensive, cross-referenced, and able

to support schizophrenia research across the

clinical, cognitive, genetic, brain imaging

and linked domains.

Funding for the establishment of the ASRB

commenced in 2006, and following the official

launch in 2007, the project commenced recruitment

of sufferers and healthy controls. Construction of the

IT support infrastructure commenced in 2006, and is

currently supporting the collection and storage of

data. This paper describes the design, technology

selections and implementation of the extant

computer infrastructure, and discusses planned

extensions and enhancements.

405

Henskens F., Loughland C., Aphale M., Paul D., Richards J., Rasser P., Carr V., Catts S., Jablensky A., Michie P., Mowry B., Pantelis C., Schall U. and

Scott R. (2009).

IT SUPPORT FOR THE AUSTRALIAN SCHIZOPHRENIA RESEARCH BANK.

In Proceedings of the International Conference on Health Informatics, pages 405-410

DOI: 10.5220/0001429504050410

 SciTePress

2 REQUIREMENTS

The principles underpinning construction of the

ASRB are: useful access to the data set will be

provided to all authorised Internet-connected

researchers without a requirement that they possess

local high performance equipment; access will

strictly accord with ethics approvals, local and

national; collection of the data will be efficient, and

will avoid double-handling as far as possible;

collection of data will be ethical, respectful and

considerate of all participants.

To this end, it was decided that the electronic

dataset should be centrally stored, and it should

include metadata about the tissue samples. Gathering

of the data, and subsequent access to it would be

best facilitated using the Internet. It became apparent

that Grid technology (Foster & Kesselman, 1999)

could be used in a rather non-conventional way, to

provide a web-based secure, single entry-point

environment for interaction with the data. In

accordance with a requirement of the Australian

Research Council, the Globus 4 Grid system (Foster,

2005) is used.

Each volunteer providing data to the Bank meets

with a Clinical Assessment Office (CAO) and

undergoes a comprehensive assessment process,

involving multiple instruments and lasting for some

three and a half hours. Traditionally, trained

psychologists using paper and pen collect clinical

and neuropsychological assessment data; the data is

hand-scored, and checked for errors, prior to

entering into a software tool for analysis. It was

decided that the ASRB assessment data should be

collected directly into a purpose-built software

package.

In Australia, research that involves humans is

subjected to close scrutiny, under conditions

specified in per-project ethics approval documents.

Protection of the privacy of the volunteers is of

paramount importance. While the ultimate objective

of the ASRB is provision of de-identified data of

high usefulness for schizophrenia research, it is

necessary, particularly in the collection phase, that

identifying data is included and accessible to

authorised persons. This fine-grained control over

access to data is not typically required in Grid

environments, and presented a challenge to the

project.

The remainder of this paper discusses the

techniques used to achieve these requirements.

3 GRID-BASED SECURITY

Ethics approvals are necessarily associated with the

collection of data and samples from live patients.

These approvals typically specify the project for

which the data is to be used, and limit the group of

researchers permitted to access the data, for

example, to those at a particular institution, or in a

particular research group. It is also common that

researchers permitted to use and analyse patient data

are typically prevented from being able to identify

patients from their data (i.e. the data is de-

identified). Whilst the ASRB data ultimately belongs

to a single entity, it is collected at five principal

locations spread across four Australian states, and

each of the collection sites is covered by a local

ethics approval. Thus, a major and important aim of

the ASRB Grid is to provide controlled access to the

data available to each particular user of the Grid.

The most obvious need is to allow all authorized

users to access the de-identified data, but it is also

important to allow access to any other data (e.g.

identifying data) for which the user has approval,

either through their institution, research group, or

personally. A further consideration is that it should

be possible for selected personnel to identify patients

from their data in the circumstance that analysis has

discovered potentially beneficial treatments for those

patients.

Typical Grids require strong security to

determine whether a user should have access to a

given system, or set of systems, without the need for

any fine-grained security; a user is either allowed to

access the system, or they are not. The ASRB Grid

is different because users have different access rights

to the resources provided by the Grid, even those on

an individual component system.

As version 4 of the Globus toolkit is mainly built

on the Web Services Resource Framework (WSRF),

it can be easily used over the Web. As more fully

discussed in (Paul et al., 2006), a Web portal is used

to access the Grid systems, eliminating the need for

researchers to install special software on their

machines, and providing flexibility with respect to

client location and host computer. The portal

framework is Gridsphere (Novotny et al.), with

GridPortlets (Russell et al., Accessed 2006) used to

access the Grid. Gridsphere is an open-source portal

framework completely compliant with the JSR 168

specifications, so that any standards-compliant

portlet can be used. GridPortlets are a set of portlets

for Gridsphere that allow access to Grid resource

and user credential management, as well as GridFTP

operations, and many other useful Grid activities.

HEALTHINF 2009 - International Conference on Health Informatics

406

The GT4Portlets extension to this allows the

execution of jobs on remote Globus Toolkit 4

systems, and further enhances GridPortlet’s

compatibility with the newest version of Globus.

A SimpleCA (SimpleCA, Accessed 2008)

certificate authority supplies users with credentials

to access ASRB Grid resources. To further facilitate

the researcher’s use of the system, PURSe portlets

(Christie, 2007) are used to eliminate the user’s need

to knowingly interact with this system. Using these

portlets, a user fills in a Web-based form to request

an account. The user is then sent an email to verify

their request and an administrator is informed of the

request. The administrator can accept or reject the

user, and has the capability to provide the user with

access to an account on the Grid; ultimately the user

is informed by email of the result. When a user is

accepted, appropriate Grid credentials are

automatically created for him/her and a personal

proxy certificate stored the MyProxy server. The

user can then log in to the Web portal, using the

single password supplied by them in their initial

request, and a proxy certificate is automatically

retrieved from the MyProxy server. This proxy

certificate is then available for access by the portlets

in the Web portal; the portlets use these credentials

to authenticate with any Grid resources in a manner

that is completely transparent to the user.

As previously described, it is vitally important

that researchers are restricted to access only that data

for which they are approved. The Globus Toolkit

includes a component that can be used for this

purpose: the Community Authorization Service

(CAS) (Globus, Accessed 2008) (not to be confused

with JA-SIG’s Central Authentication Service (JA-

SIG, Accessed 2006)). CAS allows resource

providers to give course-grained access to various

systems, handing finer-grained access control

management to the community of users. This is

important for the ASRB Grid because there are very

complex levels of access for different data resources,

so fine-grained control is needed, and the users

themselves (or the administrator, on their behalf) can

best handle the complexities of these relationships.

CAS is based on the notion of assigning users to

roles, which then have different sets of permissions

associated with them. Each user can have multiple

roles, and his or her access to a particular resource is

only allowed if at least one of the roles allows access

to the resource. Support for CAS roles has been

added to the PostgreSQL database (PostgreSQL,

Accessed 2006) used to store volunteers’ responses

to the clinical assessment battery.

Whenever a new role is needed (for example

when a new ethics approval is granted), the system

administrator creates a database view that is

accessible only to users in the new role. The view

can provide fine-grained control, specifying whether

the users with a role have permission only to read

data, or to insert new data, update existing data or

delete data.

When a user (or group of users) is provided with

a new role, the system checks to see whether the

view allowed by the user’s combination of roles

currently exists. If it does not, a new view is

automatically created, including rules that allow

database operations such as insertion of new data or

updating of existing data, but only if those

permissions are included in one or more of the roles

assigned to the user.

Users are then able to access the database using

the Web portal, and any of their queries are

automatically restricted to the views permitted by

their roles. For example, the role of clinical

assessment officer (CAO) allows the user access to

complete volunteer information collected at the

officer’s site. Thus, if a Newcastle CAO performs

the same database search as a Sydney CAO, they are

returned different results; the Newcastle CAO will

‘see’ identifying data for Newcastle volunteers,

while the Sydney CAO will be returned similar data

about Sydney volunteers.

4 CLINICAL ASSESSMENT

After considering various structured numbering

schemes for volunteer identification, it was decided

that any such scheme could potentially breach

privacy. For example, representation of familial

relationships, as occurs in some numbering schemes,

could allow a CAO with access to a volunteer’s

identity to discern and possibly identify that person’s

relation(s) for whom the CAO does not have

identification permission. Accordingly a simple

scheme was adopted whereby numbers are centrally

and sequentially allocated.

Clinical Assessments (CAs) are performed at

locations chosen in part with regard to the

convenience of volunteers, so CAOs may, or may

not be Internet connected while performing the

assessment. Each CAO has been provided with a

notebook computer, with installed CA software. To

minimise the possibility of discomfort to volunteers

who may feel threatened by the sight of the back of a

screen, the tablet form of computer was chosen, so

that the CA can be conducted with direct computer

input using a stylus (very similar to use of a pen and

paper). The lack of Internet connectivity during CA

requires initial storage of the collected data on the

notebook computer. It also requires allocation of the

identification numbers prior to commencement of

the CA. The process is that the CAO, in preparation

IT SUPPORT FOR THE AUSTRALIAN SCHIZOPHRENIA RESEARCH BANK

407

for a CA, logs on to the Grid, selects the ‘number

allocation’ portlet that accepts basic volunteer

identification and contact information (obtained

when the appointment was made) and returns that

volunteer’s identification number. Centralised

allocation ensures uniqueness of volunteer

identifiers.

Another major benefit of this allocation

technique is that accidental (or intentional) duplicate

assessment of volunteers can be avoided. When a

CAO requests allocation of a new identifier, the

system checks the database to determine whether the

provided volunteer information is substantially

similar to that of an existing participant. In some

cases (if access to the potential duplicate entry is

permitted for the CAO) the issue can be resolved

straight away; otherwise the situation is referred to

the Project Manager, whose ‘super user’ permissions

allow a determination to be made.

The CA software is written in the object-oriented

Java programming language (Arnold & Gosling,

2000), chosen in part because the compiled code will

execute on any computer with an implemented Java

Virtual Machine (JVM), most significantly the CA

notebooks and within the Unix-based central

server’s web site. The data collected during

assessment, and its structure, is described in XML

(Bray et al., 2006). JSP technology (Sun Developer

Network, 2008) is used to facilitate alignment of

software screens with the database fields. This

architecture has already proven its value as user

requirements evolved during the software

development.

The assessments performed in a CA were

determined after consultation with the principal

schizophrenia researchers in Australia and abroad,

and include the entire Diagnostic Interview for

Psychoses (DIP) (Castle et al., 2006). Scoring of the

DIP is achieved using the OPCRIT dynamic link

library (Craddock et al., 2006) on the Windows-

based notebooks, however the OPCRIT functionality

will require re-coding for the server implementation.

Also included in the CA are: Socio-demographic and

Clinical History Schedule; Neurological,

Personality, and Cognitive Functioning evaluations,

and a psychosis screening tool.

Whilst clinical assessment has been automated

previously, for example (Peters et al., 1998), this has

largely been for one instrument only, not for a

complete series as conducted by ASRB, making this

process unique.

The benefits of an electronic clinical assessment

system are that, by and large, it reduces the need for

paper copies. Because assessments can be scored

and the data checked automatically, this reduces the

amount of time assessors spend second-handling the

data, and the one data input occasion (at the time of

assessment) reduces human error due to data scoring

and entry. In addition, data is automatically checked

to ensure all questions have been answered before

the assessment is uploaded to the servers at the

ASRB website. This automated data transfer allows

for the quick and efficient aggregation of the scored

dataset.

Security of data transfer is ensured by the use of

Transport Layer Security (Freier et al., 1996),

supported by a commercially provided Secure

Socket Layer (SSL) certificate. This system uses

key-based encryption (National Institute of

Standards and Technology, 2006), with the ASRB

server having its own private key, and client

browsers using the matching public key (certificate)

that is certified as belonging to the ASRB by an

international authority trusted by all modern Web

browsers. In this way, clients can be assured that all

communication is encrypted and that only the ASRB

system can decipher/understand it.

Computer usage is now commonplace, so the use

of electronic means to collect data no longer

intimidates participants or assessors. However, when

required, either because of equipment malfunction or

volunteer discomfort, assessors can revert to using

traditional paper and pen versions by printing a copy

from the electronic version.

At this time two user manuals have been

produced. One comprises step-by-step instructions

about installing the software, and updates, on the CA

notebook computers. The other is a "How To..."

manual that instructs CAOs on how to use the

software. Experience shows that CAOs require some

initial training (for example running through a

training assessment with them, for which a practice

web portal has been provided) and troubleshooting,

but they have quickly adapted to the system and

feedback is very positive, as shown in the next

section.

5 USER FEEDBACK

Users of the systems have already reported the

following benefits of computerisation:

1. It is quick, because assessors only have to

handle the data once, at collection.

2. Labour costs are reduced because assessors can

complete assessments more quickly. This is

because they do not have to spend time scoring

or on multiple data entry.

3. Accuracy is enhanced. The system allows

consistent data collection across sites.

Assessment is structured but flexible, in that it

asks questions in the same way and order, but it

HEALTHINF 2009 - International Conference on Health Informatics

408

is flexible enough to allow CAOs to return to

questions if new information comes to light

throughout the interview.

4. Comment sections are available for each

question to record verbatim responses by the

participant or any information of relevance.

That information can then be checked by a

psychiatrist for quality assurance, diagnostic,

and staff supervision purposes.

5. Set data parameters reduce input errors during

the interview. All assessments are checked for

completion at the end of the assessment,

reducing missing data. Data is scored

automatically, reducing scoring errors.

6. It is efficient. The software is installed on

laptop computers, making the assessment fully

portable. In addition, the software has inbuilt

automated skips for redundant questions so

CAOs do not waste time on irrelevant

questions.

6 FUTURE WORK

The next phase of development involves

management of the growing number of collected

MRI scans. Work is in progress on determination of

the metadata that will be stored in the database about

each scan. Additionally, standard structures are

being designed to store the scans, both in original

form (in excess of one gigabyte per scan), plus a set

of standard mutations of the scan data. For example,

prior to release of scans to researchers, the facial

features must be removed to prevent identification.

Another aspect of the MRI data is pre-processing

in preparation for analysis. Previous research

suggests that variations in the volumes of

cerebrospinal fluid, grey matter and white matter are

related to brains diseases such as schizophrenia (da

Silva, 2007). It follows that volumetric analysis of

the brain structures is of interest to the research

community. In preparation of such measurement, it

is necessary to delineate the regions, after which the

images can be normalised. At present the delineation

is performed manually, since computerised

algorithms, e.g. (Hata et al., 2000, Pham & Prince,

1999) do not yet produce the required level of

accuracy. This is time-consuming, and not practical

when thousands of images are involved. Work is in

progress on development of new techniques for

computerised segmentation of the brain MRIs.

Ultimately, the ASRB Grid infrastructure aims to

provide researchers with the ability to analyse

(subsets of) the data collection, leading to advances

in the understanding and treatment of schizophrenia.

While at present the Grid is primarily used for its

security features, some researcher tasks will benefit

from access to the parallel resources it makes

available. For example, transfer of sets of MRI data

over the Internet for local (to the researcher) analysis

is expensive with respect to time (noting that some

of the member sites are up to 4,000 kilometres

apart). It is much more efficient to carry out the

analysis by performing the computation close to the

data source, with high bandwidth data path(s)

joining the storage and compute nodes.

It is not easy to automatically execute a task on a

set of remote machines. Projects such as

GT4Portlets allow the execution of jobs on a single

remote machine, and projects such as the Gridbus

Broker (Venugopal et al., 2006) automatically

allocate tasks to servers, but the interfaces to these

are very general. Thus a further task for this project

is to create a portlet wizard that allows the easy

creation of a portlet to execute a particular

application. These portlets will likely be based on

the Gridbus Broker, but will enable researchers to

choose input files and set parameters using a simple,

easily understandable Web form. Provision of a

wizard will make it easy for developers to create

portlets for many different programs. By way of

support for computations with special needs, more

knowledgeable developers (or their programmers)

will have access to the full source code so the portlet

can be modified as needed. In this way researchers

and developers will be able to use the processing

capabilities of the distributed compute servers much

more easily than would otherwise be possible.

7 CONCLUSIONS

This paper presents the current state of development

of IT-based infrastructure to support the Australian

Schizophrenia Research Bank. A Globus-based Grid

underpins the centralised data storage, providing a

secure (using CA certificates) processing and access

system that presents a Web browser interface to

users. Purpose-built software executing on

(optionally non-networked) Windows tablet

computers is described. This software automates a

comprehensive set of clinical assessment

instruments; each collected data item, together with

the diagnoses, is later securely uploaded to the

central repository using a Web portal.

The use of Grid technology for its security,

rather than its resource sharing capabilities, is

somewhat unusual. Techniques are described that

provide fine-grained control over user access to the

centralised dataset. This granularity is essential to

achieve compliance with the ethics conditions under

IT SUPPORT FOR THE AUSTRALIAN SCHIZOPHRENIA RESEARCH BANK

409

which research that involves humans is conducted in

Australia.

Various extant and future Grid facilities and

capabilities are discussed, together with user

affirmation of the current system. Experience with

the design and implementation of further

functionality will be the subject of future

publications.

ACKNOWLEDGEMENTS

This work was supported by the Australian Research

Council (ARC) grant SR0566756 (2005-2006), the

National Health & Medical Research Council

(NHMRC) grant AIP/ERP #1679 (2006-2010), the

Schizophrenia Research Institute (SRI) utilising

infrastructure funding from NSW Health, and a

grant from the Pratt Foundation (2007-2011).

REFERENCES

Arnold, K. & Gosling, J. (2000) The Java Programming

Language, 3rd ed., Addison Wesley.

Bray, T., Paoli, J., Sperberg-Mcqueen, C. M., Maler, E. &

Yergeau, F. (2006) Extensible Markup Language

(XML) 1.0 (Fourth Edition).

http://www.w3.org/TR/2006/REC-xml-20060816.

Castle, D. J., Jablensky A., Mcgrath J. J., Carr V., Morgan

V., Waterreus A., Valurig G., Stain H., Mcguffin P. &

Farmer A. (2006) The diagnostic interview for

psychoses (DIP) : development, reliability and

applications. Psychological Medicine 36(1), 69-80.

Christie, M. (2007) PURSe Portlets Website.

http://www.extreme.indiana.edu/portals/purse-portlets.

Craddock, M., Asherson, P., Owen, M. J., Williams, J.,

Mcguffin P. & Farmer A.E. (2006) The OPCRIT

Webpage. The British Journal of Psychiatry, 169, 58-

63.

Da Silva, F. (2007) A Dirichlet process mixture model for

brain MRI tissue classification. Medical Image

Analysis, 11, 168-182.

Foster, I. (2005) Globus Toolkit Version 4: Software for

Service-Oriented Systems. IFIP International

Conference on Network and Parallel Computing.

Springer-Verlag.

Foster, I. & Kesselman, C. (1999) The Grid: Blueprint for

a New Computing Infrastructure, Morgan Kaufmann.

Freier, A., Karlton, P. & Kocher, P. (1996) The SSL

Protocol Version 3.0.

http://wp.netscape.com/eng/ssl3/draft302.txt.

Globus (Accessed 2008) GT 4.0: Security.

http://www.globus.org/toolkit/docs/4.0/security/.

Hata, Y., Kobashi, S., Hirano, S., Kitagaki, H. & Mori, E.

(2000) Automated segmentation of human brain MR

images aided by fuzzy information granulation and

fuzzy inference. IEEE Transactions on Systems, Man,

and Cybernetics, Part C: Applications and Reviews,

30(3), 381-395.

JA-SIG (Accessed 2006) JA-SIG Central Authentication

Service. http://www.ja-sig.org/products/cas.

National Institute of Standards and Technology (2006)

Secure Hashing.

http://csrc.nist.gov/groups/ST/toolkit/secure_hashing.h

tml.

Novotny, J., Russell, M. & Wehrens, O. Gridsphere: A

Portal Framework for Building Collaborations.

Gridsphere Project Website.

Paul, D., Henskens, F. A., Johnston, P. & Hannaford, M.

R. (2006) Portal-based Support for Mental Health

Research. Grid Computing Environments Workshop,

Supercomputing '06. Tampa, Florida, IEEE Computer

Society.

Peters, L., Clark, D. & Carrol, F. (1998) Are computerised

interviews equivalent to human interviewers?: CIDI-

Auto vs. CIDI. Psychological Medicine, 28, 893-901.

Pham, D. L. & Prince, J. L. (1999) Adaptive fuzzy

segmentation of magnetic resonance images. IEEE

Transactions on Medical Imaging, 18(9).

Postgresql (Accessed 2006) The world's most advanced

open source database. http://www.postgresql.org/.

Russell, M., Novotny, J. & Wehrens, O. (Accessed 2006)

The Grid Portlets Web Application: A Grid Portal

Framework. Gridsphere Project Website.

Saha, S., Chant, D., Welham, J. & Mcgrath, J. (2005) A

systematic review of the prevalence of schizophrenia.

PLoS Medicine, 2, 413-433.

SimpleCA (Accessed 2008) SimpleCA Instructions

. http://www.vpnc.org/SimpleCA/.

Sun Developer Network (2008) JavaServer Pages

Technology. http://java.sun.com/products/jsp/.

Venugopal, S., Buyya, R. & Winton, L. (2006) A Grid

Service Broker for Scheduling e-Science Applications

on Global Data Grids. Concurrency and Computation:

Practice and Experience, 18(6), 685-699.

HEALTHINF 2009 - International Conference on Health Informatics

410