The BMS-LM ontology for biomedical data reporting throughout the lifecycle of a research study: From data model to ontology.

Biomedical research data reuse and sharing is essential for fostering research progress. To this aim, data producers need to master data management and reporting through standard and rich metadata, as encouraged by open data initiatives such as the FAIR (Findable, Accessible, Interoperable, Reusable) guidelines. This helps data re-users to understand and reuse the shared data with confidence. Therefore, dedicated frameworks are required. The provenance reporting throughout a biomedical study lifecycle has been proposed as a way to increase confidence in data while reusing it. The Biomedical Study - Lifecycle Management (BMS-LM) data model has implemented provenance and lifecycle traceability for several multimodal-imaging techniques but this is not enough for data understanding while reusing it. Actually, in the large scope of biomedical research, a multitude of metadata sources, also called Knowledge Organization Systems (KOSs), are available for data annotation. In addition, data producers uses local terminologies or KOSs, containing vernacular terms for data reporting. The result is a set of heterogeneous KOSs (local and published) with different formats and levels of granularity. To manage the inherent heterogeneity, semantic interoperability is encouraged by the Research Data Management (RDM) community. Ontologies, and more specifically top ontologies such as BFO and DOLCE, make explicit the metadata semantics and enhance semantic interoperability. Based on the BMS-LM data model and the BFO top ontology, the BioMedical Study - Lifecycle Management (BMS-LM) core ontology is proposed together with an associated framework for semantic interoperability between heterogeneous KOSs. It is made of four ontological levels: top/core/domain/local and aims to build bridges between local and published KOSs. In this paper, the conversion of the BMS-LM data model to a core ontology is detailed. The implementation of its semantic interoperability in a specific domain context is explained and illustrated with examples from small animal preclinical research.

Predicting treatment resistance in schizophrenia patients: Machine learning highlights the role of early pathophysiologic features.

Detecting patients with a high-risk profile for treatment-resistant schizophrenia (TRS) can be beneficial for implementing individually adapted therapeutic strategies and better understanding the TRS etiology. The aim of this study was to explore, with machine learning methods, the impact of demographic and clinical patient characteristics on TRS prediction, for already established risk factors and unexplored ones. This was a retrospective study of 500 patients admitted during 2020 to the University Hospital Group for Paris Psychiatry. We hypothesized potential TRS risk factors. The selected features were coded into structured variables in a new dataset, by processing patients discharge summaries and medical narratives with natural-language processing methods. We compared three machine learning models (XGBoost, logistic elastic net regression, logistic regression without regularization) for predicting TRS outcome. We analysed feature impact on the models, suggesting the following factors as markers of a high-risk TRS profile: early age at first contact with psychiatry, antipsychotic treatment interruptions due to non-adherence, absence of positive symptoms at baseline, educational problems and adolescence mental disorders in the personal psychiatric history. Specifically, we found a significant association with TRS outcome for age at first contact with psychiatry and medication non-adherence. Our findings on TRS risk factors are consistent with the review of the literature and suggest potential in using early pathophysiologic features for TRS prediction. Results were encouraging with the use of natural-langage processing techniques to leverage raw data provided by discharge summaries, combined with machine leaning models. These findings are a promising step for helping clinicians adapt their guidelines to early detection of TRS.

An Ontology Design Pattern for Modeling Experimental Paradigms.

We present an ontology design pattern for modeling scientific experiments and examinations conducted in a clinical research study. Integrating heterogeneous data into a common ontological model is a challenge, redoubled if we want them to be explored later. In order to facilitate the development of dedicated ontological modules, this design pattern relies on invariants, is centered on the event of the experiment, and keeps the link to the original data.

Ontology-Based Semantic Annotation of French Psychiatric Clinical Documents.

Building a timeline of psychiatric patient profiles can answer many valuable questions, such as how important medical events affect the progression of psychosis in patients. However, the majority of text information extraction and semantic annotation tools, as well as domain ontologies, are only available in English and cannot be easily extended to other languages, due to fundamental linguistic differences. In this paper, we describe a semantic annotation system based on an ontology developed in the PsyCARE framework. Our system is being manually evaluated by two annotators on 50 patient discharge summaries, showing promising results.

Ontological Modeling of Clinical Study Forms.

The use of eCRFs is now commonplace in clinical research studies. We propose here an ontological model of these forms allowing to describe them, to express their granularity and to link them to the relevant entities of the study in which they are used. It has been developed in a psychiatry project but its generality may allow a wider application.

Temporal Medical Knowledge Representation Using Ontologies.

Representing temporal information is a recurrent problem for biomedical ontologies. We propose a foundational ontology that combines the so-called three-dimensional and four-dimensional approaches in order to be able to track changes in an individual and to trace his or her medical history. This requires, on the one hand, associating with any representation of an individual the representation of his or her life course and, on the other hand, distinguishing the properties that characterize this individual from those that characterize his or her life course.

Performances of a Solution to Semi-Automatically Fill eCRF with Data from the Electronic Health Record: Protocol for a Prospective Individual Participant Data Meta-Analysis.

Clinical trial data collection still relies on a manual entry from information available in the medical record. This process introduces delay and error risk. Automating data transfer from Electronic Health Record (EHR) to Electronic Data Capture (EDC) system, under investigators' supervision, would gracefully solve these issues. The present paper describes the design of the evaluation of a technology allowing EHR to act as eSource for clinical trials. As part of the EHR2EDC project, for 6 ongoing clinical trials, running at 3 hospitals, a parallel semi-automated data collection using such technology will be conducted focusing on a limited scope of data (demographic data, local laboratory results, concomitant medication and vital signs). The evaluation protocol consists in an individual participant data prospective meta-analysis comparing regular clinical trial data collection to the semi-automated one. The main outcome is the proportion of data correctly entered. Data quality and associated workload for hospital staff will be compared as secondary outcomes. Results should be available in 2020.