Hippocrates is a search tool that contains 1.6 million of biomedical references from Pubmed, related to the Oncology domain. It makes use of the data model incorporated into the European project EURECA and is oriented in extending the current framework with biomedical searching functionality. EURECA framework is used by oncologists for multiple use cases. Having as input of Hippocrates the workflow state of the oncologist's workflow, but also the Electronic Health Record (EHR) of the patient, the objective of the current project was defined; to improve the searching process of the oncologist by making use of the contextual information of the case. During this project, the concept of collaboration among doctors became more and more obvious that was needed and necessary to be supported in the searching process. Different contextualization and collaboration techniques were discovered, analyzed and finally implemented as Hippocrates components. Three different contextualization approaches were implemented; the pre-filtering in which concepts included in the EHR of the patients are used by the oncologist in order to filter the documents and get relevant only results, the query expansion that expands the user's query with the patient's observations including that of the diagnosis and the re-ranking that re-ranks the top results based on a Boolean score of concepts of the EHR included in the references. In the collaboration perspective, an algorithmic intervention based on the I-SPY equations for scoring documents rated by similar users was implemented among multiple User Interface (UI) based intervention components such as group history ranked by rating. The experiment/evaluation of Hippocrates besides the effort and resources devoted towards that direction, was limited to three participants, one of which is an Oncologist. The results though, even not statistically confident, indicate that the pre-filtering contextualization improves the searching process, and the rest two were observed to improve it in at least one case. Limitations were discovered during the internal evaluation of the collaborative algorithm and the I-Spy equations but also prove that could at least improve the Inverse Cumulative Gain of the result set. The Usability of the system was highly rated and several components averagely rated were accompanied with valuable feedback that could be used for future development. Overall, Hippocrates received positive evaluation and could be further developed and tested to prove the improvement on the searching process of the oncologists.
|Date of Award||31 Aug 2014|
|Supervisor||Mykola Pechenizkiy (Supervisor 1) & A.I.D. Bucur (External coach)|