ICBO 2014 Proceedings
Applications of OBI ‘assay’
Mark Jensen*, Alexander P. Cox, Jonathan P. Bona, William Duncan, Patrick L. Ray, Alexander D. Diehl
The State University of New York at Buffalo
Buffalo, NY, USA
*
mpjensen@buffalo.edu
All authors contributed equally to this work
Abstract—We discuss the applicability of using the OBI assay being evaluated. Also, cognitive functions, such as short-term
paradigm for representing patient questionnaires, memory, cannot be the bearer of measureable qualities. The
neuropsychological tests, and neurological exams, as well to solution in NPT is to connect a cognitive process to the
annotate data generated from these assessments. We conclude function it realizes in the assay process using a new
that the specification for OBI ‘assay’ employs a broad enough relationship between a data item and a function.
notion of evaluation to allow for these uses. However, it would be
preferable to introduce subclasses of OBI ‘planned process’ or The Multiple Sclerosis Patient Data Ontology (MSPD) has
OBI ‘assay’ that explicitly addresses these types of use cases and been developed to represent both clinical measures and patient
provides clear groupings for general types of assays. reported outcomes (PRO) associated with the New York State
Multiple Sclerosis Consortium (NYSMSC) patient data
Keywords—assay; OBI; questionnaire; neuropsychological registry [4]. A PRO is generally considered to be an assessment
test; neurological exam; clinical history of any aspect of a patient's health status that comes directly
from the patient and without any interpretation by a clinician
I. BACKGROUND [5]. The data registry uses standardized forms addressing
The Ontology for Biomedical Investigations (OBI) is an demographic and clinical information, disease status and
integrated ontology for the description of biological and progression. It also includes data pertaining to patients’
clinical investigations [1]. OBI is a domain ontology that perception of their quality of life and wellbeing, which
provides a set of terms and relations to support precise includes assessment of physical and psychosocial impairment.
annotation and querying of the data generated in biomedical During the enrollment process patients are asked to rate their
investigations. It represents the design, types of analyses and perception of their own functional abilities and affective states.
assays performed, specifications, and data generated, resulting A difficulty in using the assay framework has been in
in classes such as ‘assay’, ‘plan specification’, and reconciling what qualifies as a physical examination and
‘measurement datum’. OBI defines ‘assay’ as “a planned subsequent evaluation. An output of a survey in which a patient
process with the objective to produce information about the is asked to make a judgment about his or her perceived
material entity that is the evaluant, by physically examining it limitation in a particular limb or visual acuity may indeed
or its proxies” [2]. All assays have a specified output, an qualify in this case as a sort of post-hoc physical exam which
information content entity, which is about the evaluant. allows the evaluant to also be the evaluator. The OBI ‘self-
Examples of usage are: “Assay the wavelength of light emitted reported handedness assessment’ supports the application of
by excited Neon atoms. Count of geese flying over a house.” ‘assay’ to cases where a patient self-evaluates outside the
Subclasses of OBI ‘assay’ include many laboratory-specific context of a direct physical exam.
examples, such as ‘sequencing assay’ and ‘metabolite
profiling’. However, other types include ‘performing a clinical However, it is less clear how questionnaires and forms that
assessment’, ‘age measurement assay’, and ‘handedness assay’. obtain basic demographic data fit within OBI’s account of
assays. A patient responding to questions such as date of birth,
Several projects are underway which seek to represent and marital status, insurance provider, etc. pushes one to reconsider
annotate data generated from different types of forms, what is being evaluated, especially since no physical
questionnaires, and tests. Each of these uses-cases broaden the examination is involved.
application of OBI ‘assay’ in one or more ways.
A related project is the development of an ontology-based
Neuropsychological tests are used to assess cognitive medical history module to extend a legacy clinical information
domains such as attention, visual-spatial ability, memory, management system. This module collects, structures, and
executive function, and language comprehension and stores data using OBO Foundry ontologies and semantic web
expression. In addition to representing the structure of these technology. Part of this work involves the development of an
neuropsychological tests, it is crucial to capture the cognitive ontological model for health history questionnaires, each
processes and functions that they evaluate as well as the data consisting of a series of questions to be answered by the patient
they produce. The neuropsychological Testing Ontology (NPT) during a medical history interview session. While many
utilizes OBI’s assay paradigm to represent these tests [3]. The question answers that make up a patient's clinical history are
handedness assay was used as a starting point to model these clearly about the patient’s body or are the result of some
tests. However, difficulties have been encountered in relating physical examination of the patient, others do not seem to fit
the assay process to the cognitive processes and functions the OBI assay framework. Family history questions are
96
ICBO 2014 Proceedings
problematic in this regard. So are questions about the existence using OBI ‘assay’. As a result, they raise interesting questions
of a previous diagnosis, such as “Has a doctor ever told you about what modifications or additions to OBI are required.
that you had a myocardial infarction or heart attack?” [6]. The
Our poster details the discussed uses of OBI ‘assay’ and
planned process of soliciting an answer to this question is
intended to produce information about physical entities (the summarizes the difficulties encountered. We offer alternatives
and suggest the inclusion of a general set of assay and planned
patient; her heart) as well as information about related entities
such as diagnoses. However, asking and answering this process types which will aid in recognizing distinctions
between the various assessment strategies. Our hope is that this
question and recording the answer does not directly involve a
physical examination. An answer of “yes” to this question most work will promote development in OBI and assist others who
are using the assay paradigm in OBI.
likely indicates that a previous assay resulted in the original
diagnosis; however it is much more difficult to argue for any
connection between an answer of “No” (or “I'm not sure”) and
any sort of physical examination. REFERENCES
[1] Brinkman, R.R., et al., Modeling biomedical experimental processes
II. CONCLUSION with OBI. J Biomed Semantics, 2010. 1 Suppl 1: p. S7.
As it is currently defined, OBI ‘assay’ allows for a broad [2] http://purl.obolibrary.org/obo/OBI_0000070
interpretation of what it means to physically examine or [3] Cox AP, Jensen M, Ruttenberg A, Szigeti K, Diehl AD. (2013)
evaluate a patient. While neuropsychological tests and clinical “Measuring Cognitive Functions: Hurdles in the Development of the
NeuroPsychological Testing Ontology” in Proceedings 4th International
exams can be made to fit within the assay framework, Conference on Biomedical Ontology, Montreal, Canada, July 7-9, 2013.
modification is required. Subjects being asked to evaluate [4] Jacobs, L.D., et al., A profile of multiple sclerosis: the New York State
aspects of their own bodily functioning or cognitive and Multiple Sclerosis Consortium. Mult Scler, 1999. 5(5): p. 369-76.
affective status provides another challenge for understanding [5] Deshpande, P.R., et al., Patient-reported outcomes: A new era in clinical
and implementing OBI ‘assay’, yet this ontological class can research. Perspect Clin Res, 2011. 2(4): p. 137-44.
still provide a plausible solution. However, questionnaires, [6] https://www.phenxtoolkit.org/index.php?pageLink=browse.protocoldeta
demographic information, and factual tests with no interpretive ils&id=40801
or summary outputs go beyond what can be accomplished
97
ICBO 2014 Proceedings
Applications of OBI assay!
Mark Jensen, Alexander P. Cox, Jonathan Bona, Patrick L. Ray, William Duncan, Alexander D. Diehl
Introduction! Example of Canonical Use Case! NeuroPsychological Testing Ontology (NPT)! Problems Encountered in Applications of OBI ‘assay’!
The Ontology for Biomedical Investigations (OBI) is an Neuropsychological tests are used to assess cognitive domains • OBI lacks diversity in the types of assays represented.
integrated ontology for the description of biological and DVVD\ such as attention, visual-spatial ability, memory, executive function, While the original scope of ‘assay’ seems grounded in
clinical investigations. It represents the design, types of DQDO\WHDVVD\
LVD
and language. NPT uses the assay paradigm to represent these prototypically “wet” laboratory assays, its definition does
analyses and assays performed, specifications, and data LVD
tests. The OBI handedness assay was used as a starting point to not restrict its application to these cases.
PHDVXULQJJOXFRVH KRPR
generated during an investigation. Thus, it provides classes FRQFHQWUDWLRQLQEORRGVHUXP VDSLHQV model neuropsychological tests. However, difficulties arose when
such as assay, plan specification, and measurement LQVWDQFHRI LQVWDQFHRI
relating the results of neuropsychological assays to cognitive • Elucidation for the concepts of evaluation and
datum. An assay is a planned process which produces processes and functions. In particular, a cognitive function – such as measurement is needed. If possible, formal definitions
information about an evaluant. Examples of assays include: PHDVXULQJJOXFRVH
FRQFHQWUDWLRQLQEORRGVHUXP
KDV
RXWSXW
PHDVXUHPHQW
GDWDLWHP
short-term memory or executive function – cannot be the bearer of a should be provided.
assay the wavelength of light emitted by excited neon quality. To resolve this issue, we created a new relationship, is
atoms” and “count the number of geese flying over a
KDVVSHFLILHG
LQSXW
LVDERXW
functional measurement of, to connect neuropsychological test • The exact relationship between assays and their outputs
house. Subclasses of OBI assay include laboratory- UHDOL]HV
UHDOL]HV
EORRGVHUXP
results to the cognitive functions being evaluated. is unclear. All measurement data items have to be the
JOXFRVH KRPR
GHULYHVIURP
specific examples, such as sequencing assay and SDUWRI
VSHFLPHQ VDSLHQV
output of some assay, but not all assays have to output a
measurement data item. Thus, assays can have outputs
KDVUROH
metabolite profiling. However, other types include
KDVUROH KDVUROH
HYDOXDQW
performing a clinical assessment, age measurement DQDO\WHUROH
UROH SDWLHQWUROH
" ! that are not measurement data items. Furthermore, all
assay, and handedness assay. Several projects at the !!
assay output data must be about a material entity that
University at Buffalo seek to represent and annotate data bears an evaluant role. This complicates the
generated from different types of questionnaires, forms, and OBI ‘assay’ has many subclasses. Among these, the ‘analyte assay’ !#
!#
" ! %
!#
" !
$
! ! representation of assays designed to evaluate non-
tests. Each of these provide a use case that broadens the classes represents “classic” laboratory assays in which a substance material entities.
current application of OBI assay in one or more ways, with an analyte role is detected in a mixture, which bears the ! ! ! ! " ! !
! &
possibly stretching its applicability. evaluant role. Other OBI assays omit naming of the analyte and its " !
• It is not clear how filling out questionnaires or forms that
! ! "! $ ! ! "!
role, but follow a similar design pattern, where the evaluant role is !#
! ! % !#" ! $ !% obtain basic demographic data fit within OBI’s account of
reserved for the entity under study. assay. A patient responding to questions such as date of
OBI Assay! ! !
birth, marital status, insurance provider, etc. pushes one
! ! &
A key question is whether the material entity bearing the evaluant ! !
& & "! "!
to reconsider what is being evaluated—especially since
The class ‘assay’ is central to OBI’s purpose and utility. The role can be a sentient creature, a person, who may be assayed via no physical examination is involved. Can a patient
paradigm for representing assays involves several key observation or direct questioning to yield information that is about
! ! "! ! ! ! ! #"! #" !
evaluate oneself? Also, does an ordinal ranking of pain
components that relate to the assay class, as shown below. non-material aspects of that person. A precedent for this in OBI is
count as a measurement?
The evaluant role specifies the mode of participation in the the ‘handedness assay’ and its subclasses, which represent assays Above is a partial representation of the Clock-Drawing Test in NPT. Above
assay for the entity under study. The measurement data about the handedness of a person. that are examples of common mistakes made by test participants.
item represents information derived from executing an
Solutions!
assay. The assay objective specifies the goal of the assay. Multiple Sclerosis Patient Data Ontology (MSPD)! Medical History Collection!
Each of these is essential to representing and differentiating • Examples of non-assay planned processes that produce
subtypes of assay. MSPD has been created to SDWLHQWUHSRUWHG
One of our projects is an ontology-based medical history module information about evaluants should be provided to
represent clinical measures DVVD\ that is part of a larger clinical information management system. It illuminate the distinction between these classes.
MXGJHPHQWDERXW
and patient reported outcomes IXQFWLRQDVVD\ LVD stores structured representations of questions and answers about
2%,DVVD\LVDSODQQHGSURFHVVZLWKWKHREMHFWLYHWRSURGXFHLQIRUPDWLRQDERXWWKH obtained from enrollment forms YLVLRQ
DVVD\ LVD 1<606&
HQUROOHH
patients’ medical histories. The process of completing a medical • Develop paradigmatic assay applications and make
PDWHULDOHQWLW\WKDWLVWKHHYDOXDQWE\SK\VLFDOO\H[DPLQLQJLWRULWVSUR[LHV
(TXLYDOHQWWR DFKLHYHVBSODQQHGBREMHFWLYHVRPH DVVD\REMHFWLYH used by centers participating in history questionnaire has as its parts assay-like planned processes current applications consistent in their representation.
the New York State Multiple LQVWDQFHRI LQVWDQFHRI
to produce information about the patient, but many do not involve
2%,DVVD\REMHFWLYHLVDQREMHFWLYHVSHFLILFDWLRQWRGHWHUPLQHDVSHFLILHGW\SHRI
LQIRUPDWLRQDERXWDQHYDOXDWHGHQWLW\ WKHPDWHULDOHQWLW\EHDULQJHYDOXDQWUROH Sclerosis Consortium. YLVLRQ
KDV PHDVXUHPHQW
physically examining the patient or anything else. Example • Objective specification should be specified to relate to the
OLPLWDWLRQ
Enrollees asked to rate, for DVVD\
RXWSXW GDWDLWHP
questions derived from the PhenX Toolkit [1] appear below. evaluation in the assay, not just the type of information.
2%,HYDOXDQWUROHLVDUROHWKDWLQKHUHVLQDPDWHULDOHQWLW\WKDWLVUHDOL]HGLQDQDVVD\ LVDERXW
LQZKLFKGDWDLVJHQHUDWHGDERXWWKHEHDUHURIWKHHYDOXDQWUROH example, aspects of their LVSDUWRI
LV
1<606&
HQUROOHH is about
6XEFODVVRI µLVUHDOL]HGE\¶RQO\DVVD\ bodily functioning, the extent of EHDUHU
• Providing general subtypes of assay to group its current
OLPLWDWLRQ RI
YLVXDO
,$2PHDVXUHPHQWGDWDLWHP LVDGDWDLWHPWKDWLVDUHFRUGLQJRIWKHRXWSXWRIDQDVVD\ their pain, or life satisfaction DVVD\
IXQFWLRQ
KDVUROH Family history?
subtypes would help address these shortcomings. For
P
could be said to be producing LVSDUWRI
UHDOL]HV
is about
Substance use? example, assays could be grouped by the nature of their
,$2LQIRUPDWLRQ information about themselves 1<606&SDWLHQW
UHSRUWHG
YLVXDO
SHUFHSWLRQ
1<606&
HQUROOHHUROH is about
Health history?
evaluants, the type of evaluation process, or their
FRQWHQWHQWLW\ as the evaluant. HQUROOPHQWDVVD\ Medications?
objectives. Membership in these groupings could be
inferred by enforcing the use of consistent logical
UHDOL]HV is about
,$2GLUHFWLYH
LQIRUPDWLRQ KDVVSHFLILHG $ERYHLVDQH[DPSOHRI063'VDSSOLFDWLRQRI2%, DVVD\
HQWLW\
,$2GDWDLWHP
RXWSXW
LVDERXW
%R[HVEHORZWKHEOXHOLQHDUHLQVWDQFHVRIFODVVHVQDPHG ? definitions for assay subtypes.
2%,SODQQHG LQVLGHWKHER[
SURFHVV P
,$2PHDVXUHPHQW 2%,HYDOXDQW /HIWLVDH[DPSOHRIDQD[LRPLQWKHYLVLRQOLPLWDWLRQDVVD\
GDWDLWHP UROH
Contact Information!
%HORZLVDSRUWLRQRIWKHSDWLHQWUHSRUWHGFRPSRQHQWRIWKH
,$2REMHFWLYH HQUROOPHQWIRUP
VSHFLILFDWLRQ
Mark Jensen: mpjensen@buffalo.edu
LVVSHFLILHG UHDOL]HV
RXWSXWRI 2%,DVVD\ KDVUROH Alexander P. Cox: apcox@buffalo.edu
•Has any of your first degree relatives ever had melanoma? Jonathan Bona: jpbona@buffalo.edu
DFKLHYHV
SODQQHG
•Has a doctor or nurse ever said that you have high blood pressure Patrick L. Ray: plray@buffalo.edu
2%,DVVD\ REMHFWLYH KDVVSHFLILHG
%)2PDWHULDO or hypertension? William Duncan: wdduncan@buffalo.edu
LQSXW
REMHFWLYH HQWLW\ •Have you smoked at least 100 cigarettes in your entire life? Alexander D. Diehl: addiehl@buffalo.edu
•In the past 3 years, please indicate if you have taken either of the
%ODFNDUHSURFHVVHVJUHHQDUHLQIRUPDWLRQFRQWHQWHQWLWLHV following types of medications: NYS Center of Excellence in Bioinformatics & Life Sciences
SXUSOHDUHPDWHULDOHQWLWLHVDQGUHGDUHUHDOL]DEOHHQWLWLHV
Statin medications such as lovastatin, … University at Buffalo
[1] https://www.phenxtoolkit.org/ 701 Ellicott Street, Buffalo, New York 14203 www.buffalo.edu
98