Bnaic2013.tudelft.nl

Eligibility Criteria for Clinical Trials1 a Department of Computer Science,VU University Amsterdam, The Netherlands b {huang,annette,Frank.van.Harmelen}@cs.vu.nl In this extended abstract, we propose a rule-based formalization of eligibility criteria for clinical trials.
The rule-based formalization is implemented by using the logic programming language Prolog. Comparedwith existing formalizations such as pattern-based and script-based languages, the rule-based formaliza-tion has the advantages of being declarative, expressive, reusable and easy to maintain. Our rule-basedformalization is based on a general framework for eligibility criteria containing three types of knowledge:(1) trial-specific knowledge, (2) domain-specific knowledge and (3) common knowledge. This frameworkenables the reuse of several parts of the formalization of eligibility criteria. We have implemented the pro-posed rule-based formalization in SemanticCT, a semantically-enabled system for clinical trials, showingthe feasibility of using our rule-based formalization of eligibility criteria for supporting patient recruitmentin clinical trial systems.
Eligibility criteria consist of inclusion criteria, which state a set of conditions that must be met, and exclusioncriteria, which state a set of conditions that must not be met, in order to participate in a clinical trial.
Take the example of the trial NCT00002720, the eligibility criteria are: Histologically proven stage I, invasive breast cancer Progesterone receptor positive or negative Age: 65 to 80, Sex: Female, Menopausal status: Postmenopausal Other: - No serious disease that would preclude surgery - No other prior or concurrent malignancy except basal cell carcinoma or carcinoma in situ of the cervix Those inclusion criteria (such as ’invasive breast cancer’ ) and exclusion criteria (such as ’No serious disease that would preclude surgery’) are trial specific. However, in order to check whether or not a requireditem (i.e., a criterion) has been met by a patient record, we need some domain knowledge to interpret therequirement and make it directly checkable from patient data. For example, ’invasive breast cancer’ can bedefined as either ’invasive ductal carcinoma’ or ’invasive lobular carcinoma’ in the diagnosis. Furthermore, 1This is an extended abstract of the paper in the Proceedings of the 14th Conference on Artificial Intelligence in Medicine (AIME 2013), N. Peek, R. Marin, and M. Peleg (eds.), Springer, LNAI2013, pp.38-47, 2013 we need some knowledge, such as temporal reasoning knowledge, to deal with temporal aspects of criteria,and service interface knowledge, to get the corresponding patient data from the EHR or CMR servers.
We can formalize the knowledge rules of the specification of eligibility criteria of clinical trials with respect to the following different re-usable knowledge types: (1) Trial-specific Knowledge: this is the formalization of which specific inclusion criteria and exclusion criteria are required for a particular clinical trial. (2) Domain-specific Knowledge: an example of this typeof knowledge is a patient of breast cancer is triple negative if the patient has estrogon receptor negative,progesterone receptor negative and protein HER2 negative status. (3) Common Knowledge: the specifica-tion of the eligibility criteria may involve some knowledge which is domain independent, like for exampleknowledge about temporal reasoning.
Related work; [2] translates each free-text eligibility criterion into a machine executable statement using a derivation of the Arden Syntax. In our work, we use a more expressive rule-based language and thenstructured the eligibility criteria as RDF. [1] presents a method entirely based on standard semantic webtechnologies and tools, that allows the automatic recruitment of a patient to available clinical trials. Althoughwe propose an even more expressive language for modeling the eligibility criteria this is in the same spiritas our approach. The empirical analysis in [3] shows that the vast majority (85%) of trial criteria is of”significant semantic complexity”. This justifies our choice for an expressive rule-based formalism. Thepaper also observes that temporal data play a role in 40% of all criteria, justifying our choice for a separatelayer for this in our formalization.
SemanticCT2 is a semantically enabled system for clinical trials. The goals of SemanticCT are not onlyto achieve interoperability by semantic integration of heterogeneous data in clinical trials, but also to facil-itate automatic reasoning and data processing services for decision support systems in various settings ofclinical trials. We have implemented the rule-based formalization of eligibility criteria as a component ofSemanticCT for the service of automatic identification of eligible patients for clinical trials.
Our feasibility study shows how two important tasks can in principle be supported by the formalization and implementation. Our experiments concern a patient recruitment task (= finding patients that qualify fora given trial), and a trial feasibility task (= checking if a set of inclusion and exclusion criteria for a newlydesigned trial results in a sufficient number of recruitable patients). The Patient Recruitment experimentshows on a (simulated) patient recruitment scenario, that we can check maximally 83.33% of the criteria,and minimally 34.48% of the criteria, based on the given patient data. The feasibility experiment shows thatdependent on the target we can find candidate patients who meet the checked criteria. These experimentsshow that conditions of realistic trials can be formalized and implemented in such a way that, at least on ourartificially generated but medically and statistically plausible patient data, both patient recruitment and trialfeasibility can be supported.
[1] Paolo Besana, Marc Cuggia, Oussama Zekri, Annabel Bourde, and Anita Burgun. Using semantic web technologies for clinical trial recruitment. In International Semantic Web Conference, pages 34–49,2010.
[2] L. Ohno-Machado, S. J. Wang, P. Mar, and A. A. Boxwala. Decision support for clinical trial eligibility determination in breast cancer. Proc AMIA Symp, pages 340–344, 1999.
[3] J. Ross, S. Tu, S. Carini, and I. Sim. Analysis of eligibility criteria complexity in clinical trials. AMIA Summits Transl Sci Proc, 2010:46–50, 2010.

Source: http://bnaic2013.tudelft.nl/proceedings/papers/paper_47.pdf