Towards Patterns for Defining and Changing Data Collection Instruments
in Mobile Healthcare Scenarios
Johannes Schobel, R¨
udiger Pryss, Marc Schickler, Manfred Reichert
Institute of Databases and Information Systems, Ulm University, Ulm, Germany
{johannes.schobel, ruediger.pryss, marc.schickler, manfred.reichert}@uni-ulm.de
Abstract—Especially in healthcare scenarios and clinical
trials, a large amount of data needs to be collected in a rather
short time. In this context, smart mobile devices can be a
feasible instrument to foster data collection scenarios. To enable
domain experts to create and maintain mobile data collection
applications themselves, the QuestionSys framework relies on a
model-driven approach to digitize paper-based questionnaires.
This digital transformation is based on manual as well as
automated tasks. The manual tasks applied by the domain
experts can be eased by the use of change patterns. They
describe features to easily add or delete the elements of a
questionnaire. This work summarizes crucial change patterns
and shows how they can be applied in practice. We believe
that the patterns constitute an important means to implement
sophisticated mobile data collection applications by domain
experts themselves.
Keywords-Mobile data collection, process-driven data collec-
tion, electronic questionnaires, change patterns.
I. INTRODUCTION
In various application domains, paper-based question-
naires (e.g., instruments) are frequently used to collect
large amounts of data. Recently, many hard-coded mobile
applications were introduced, which enable data collection
with smart mobile devices. These applications, in turn,
increased the efficacy of the data collection procedure on one
hand. On the other, frequent changes to these applications
have revealed to be challenging. A more flexible support is
provided by the QuestionSys framework, enabling domain
experts to create data collection instruments on their own [1].
To achieve this goal, a sophisticated configurator component,
Page
Alcohol
Consumption
......
First Level
Second Level
Page
Demographic
Information
Demographic Information
Welcome (HEADLINE)
Introduction (TEXT)
Name (QUESTION, FREETEXT)
Age (QUESTION, SLIDER)
Sex (QUESTION, SINGLE CHOICE)
name
age
sex
Figure 1. Different Levels of Change Patterns
combining end-user programming techniques with process
management technology, was introduced [2].
In order to better understand the digital transformation
of traditional paper-based questionnaires to complex mobile
data collection instruments, structured interviews with 10
experts from various domains were conducted. In this con-
text, more than 40 instruments from healthcare, psychology,
logistics, and education were evaluated. Finally, a set of
change patterns could be identified that shall enable domain
experts to define or change data collection instruments.
II. CHANGE PATTERNS
Table I illustrates the change patterns, which have been
identified in the realized mobile data collection applica-
tions.1The patterns can be assigned to different levels. Each
level, in turn, reflects different aspects of the data collection
instrument (cf. Fig. 1). Patterns of the first level solely
correspond to the structure (e.g., the flow) of the instrument,
whereas change patterns of the second level refer to the
content of questionnaire pages.
Structural Change Patterns (S) provide features to
create and maintain the logic of a data collection instrument.
For example, pages may be added or previously defined
instruments may be inserted.
Content Change Patterns (C) enable the management
of elements (e.g., headlines or questions) of a specific
1Estimated values, the projects are still ongoing.
Tinnitus Research
Risk Factors during
Pregnancy
Adverse Childhood
Experiences
PTSD in
War Regions
Learning Deficits
among Students
Supporting Children
after Accidents
S1 Insert Page >5>25 >80 >150 >10 >5
S2 Insert Block 0 >15 >80 >100 >15 >5
S3 Insert Empty Path 0 >15 >90 >250 >20 >5
S4 Embed Instrument 0 1 1 >20 0 0
C1 Insert Element >60 >120 >200 >1500 >100 >40
C2 Move Element >10 >40 >50 >500 >20 >10
R1 Cut Page 0 >5>10 >30 0 0
R2 Merge Page 0 >5>10 >15 0 0
R3 Move Page 0 >10 >15 >60 0 0
R4 Update Decision 0 >5>30 >80 >15 0
Table I
IDENTIFIED CHANGE PATTERNS
page. Data elements for capturing answers are automatically
created when using these patterns.
Refactoring Change Patterns (R) allow modelers to
adapt an instrument to new requirements without violating
validity constraints. For example, a page containing demo-
graphic questions may be moved, while other pages referring
to these questions are updated accordingly.
The combined use of the identified change patterns allows
domain experts to define or change mobile data collection
instruments in a more flexible manner. Furthermore, they
foster the continuous development and evolution of instru-
ments (e.g., making use of the refactoring patterns). Fig.
2 illustrates how a set of change patterns may change an
existing mobile data collection application.
In general, each identified pattern is described using a
visual representation with an example from the considered
real-world scenarios. Moreover, pre- and post-conditions for
applying the patterns are listed (cf. Table II).
III. RELATED WORK
The use of patterns in the context of reoccurring tasks
is promising in various fields of computer science. [3]
suggests a set of refactoring patterns for code fragments,
while preserving the original behaviour of the source code.
Furthermore, [4] proposes more than 20 patterns (e.g., Ob-
server pattern). [5] proposes patterns for the communication
between services. In the context of business process manage-
ment, [6] introduces common process change patterns. The
latter allow for comparing the expressiveness of respective
information systems capable of executing process models.
In [7], cloud architectural patterns for developing scalable
applications are presented. These pattern allow for an easy
installation and management on various environments.
Name Insert Block. Add a new block to an existing instrument. Available
types are IF,ALL, and REPEAT.
Signature insertBlock(type, before, after)
Example Depending on the type of the block, various scenarios are possible:
•IF blocks solely select one path based on already given
answers during run time.
•ALL blocks select all paths to be executed, however, the
person interacting with the smart mobile device may choose
its order of execution.
•REPEAT blocks allow for repeating the content of the block
multiple times. The amount of repetitions may be determined
at run time (e.g., based on given answers) or are pre-defined
by the domain expert (e.g., ntimes).
A B A B
Pre-
Condition
The position to insert the block must be exactly specified; i.e.,
after must directly follow before.
Post-
Condition
An empty block comprising a split and join gateway that are
directly connected is inserted; For IF and REPEAT blocks, data
elements for evaluating the conditions need to be connected using
READ data edges.
Table II
STRUCTURAL CHANGE PATTERN S2: INSERT BLOCK
Page
Alcohol
B
Page
Intro
A
Page
Alcohol
B
Page
Intro
A
Page
Drugs
C
Apply Change Patterns:
insertBlock(ALL, ...)
insertPage(C, ...)
insertEmptyPath(...)
movePage(B, ...)
Figure 2. Applying Change Patterns to a Data Collection Instrument
IV. SUMMARY
In the QuestionSys framework, automated and manual
tasks are required to digitize paper-based questionnaires.
Structured interviews identified a meaningful set of change
patterns (i.e., manual tasks) needed in the various scenar-
ios. The long-running projects from the healthcare domain
revealed that the provision of sophisticated change patterns
is indispensable. Providing change patterns in the context of
mobile data collection applications is an important step to
cope with the frequent application changes in practice on one
hand. On the other, more longitudinal data can be gathered
in a rather short project time. Altogether, the patterns allow
domain experts without programming skills to better create
and maintain instruments on their own.
REFERENCES
[1] J. Schobel, R. Pryss, M. Schickler, M. Ruf-Leuschner, T. El-
bert, and M. Reichert, “End-User Programming of Mobile
Services: Empowering Domain Experts to Implement Mobile
Data Collection Applications,” in IEEE 5th Int’l Conf on
Mobile Services, June 2016.
[2] J. Schobel, R. Pryss, M. Schickler, and M. Reichert, “A
Configurator Component for End-User Defined Mobile Data
Collection Processes,” in Demo Track of the 14th Int’l Conf
on Service Oriented Computing, October 2016.
[3] W. F. Opdyke, “Refactoring Object-Oriented Frameworks,”
Ph.D. dissertation, University of Illinois, 1992.
[4] E. Gamma, Design Patterns: Elements of Reusable Object-
Oriented Software. Pearson Education India, 1995.
[5] A. Barros, M. Dumas, and A. H. ter Hofstede, “Service Interac-
tion Patterns,” in Int’l Conf on Business Process Management.
Springer, 2005, pp. 302–318.
[6] S. Rinderle-Ma, M. Reichert, and B. Weber, “On the Formal
Semantics of Change Patterns in Process-aware Information
Systems,” in Proc. 27th Int’l Conf on Conceptual Modeling,
ser. LNCS, no. 5231. Springer, October 2008, pp. 279–293.
[7] C. Fehling, F. Leymann, J. R¨
utschlin, and D. Schumm,
“Pattern-based Development and Management of Cloud Ap-
plications,” Future Internet, vol. 4, no. 1, pp. 110–141, 2012.
