Accurate prediction of kinase-substrate networks using knowledge graphs.

Phosphorylation of specific substrates by protein kinases is a key control mechanism for vital cell-fate decisions and other cellular processes. However, discovering specific kinase-substrate relationships is time-consuming and often rather serendipitous. Computational predictions alleviate these challenges, but the current approaches suffer from limitations like restricted kinome coverage and inaccuracy. They also typically utilise only local features without reflecting broader interaction context. To address these limitations, we have developed an alternative predictive model. It uses statistical relational learning on top of phosphorylation networks interpreted as knowledge graphs, a simple yet robust model for representing networked knowledge. Compared to a representative selection of six existing systems, our model has the highest kinome coverage and produces biologically valid high-confidence predictions not possible with the other tools. Specifically, we have experimentally validated predictions of previously unknown phosphorylations by the LATS1, AKT1, PKA and MST2 kinases in human. Thus, our tool is useful for focusing phosphoproteomic experiments, and facilitates the discovery of new phosphorylation reactions. Our model can be accessed publicly via an easy-to-use web interface (LinkPhinder).

Facilitating prediction of adverse drug reactions by using knowledge graphs and multi-label learning models.

Timely identification of adverse drug reactions (ADRs) is highly important in the domains of public health and pharmacology. Early discovery of potential ADRs can limit their effect on patient lives and also make drug development pipelines more robust and efficient. Reliable in silico prediction of ADRs can be helpful in this context, and thus, it has been intensely studied. Recent works achieved promising results using machine learning. The presented work focuses on machine learning methods that use drug profiles for making predictions and use features from multiple data sources. We argue that despite promising results, existing works have limitations, especially regarding flexibility in experimenting with different data sets and/or predictive models. We suggest to address these limitations by generalization of the key principles used by the state of the art. Namely, we explore effects of: (1) using knowledge graphs-machine-readable interlinked representations of biomedical knowledge-as a convenient uniform representation of heterogeneous data; and (2) casting ADR prediction as a multi-label ranking problem. We present a specific way of using knowledge graphs to generate different feature sets and demonstrate favourable performance of selected off-the-shelf multi-label learning models in comparison with existing works. Our experiments suggest better suitability of certain multi-label learning methods for applications where ranking is preferred. The presented approach can be easily extended to other feature sources or machine learning methods, making it flexible for experiments tuned toward specific requirements of end users. Our work also provides a clearly defined and reproducible baseline for any future related experiments.

Using Drug Similarities for Discovery of Possible Adverse Reactions.

We propose a new computational method for discovery of possible adverse drug reactions. The method consists of two key steps. First we use openly available resources to semi-automatically compile a consolidated data set describing drugs and their features (e.g., chemical structure, related targets, indications or known adverse reaction). The data set is represented as a graph, which allows for definition of graph-based similarity metrics. The metrics can then be used for propagating known adverse reactions between similar drugs, which leads to weighted (i.e., ranked) predictions of previously unknown links between drugs and their possible side effects. We implemented the proposed method in the form of a software prototype and evaluated our approach by discarding known drug-side effect links from our data and checking whether our prototype is able to re-discover them. As this is an evaluation methodology used by several recent state of the art approaches, we could compare our results with them. Our approach scored best in all widely used metrics like precision, recall or the ratio of relevant predictions present among the top ranked results. The improvement was as much as 125.79% over the next best approach. For instance, the F1 score was 0.5606 (66.35% better than the next best method). Most importantly, in 95.32% of cases, the top five results contain at least one, but typically three correctly predicted side effect (36.05% better than the second best approach).

Physicians' Attitudes Towards the Advice of a Guideline-Based Decision Support System: A Case Study With OncoDoc2 in the Management of Breast Cancer Patients.

When wrongly used, guideline-based clinical decision support systems (CDSSs) may generate inappropriate propositions that do not match the recommendations provided by clinical practice guidelines (CPGs). The user may decide to comply with or react to the CDSS, and her decision may finally comply or not with CPGs. OncoDoc2 is a guideline-based CDSS for breast cancer management. We collected 394 decisions made by multidisciplinary meeting physicians in three hospitals where the CDSS was evaluated. We observed a global CPG compliance of 86.8% and a global CDSS compliance of 75.4%. Non-CPG compliance was observed in case of a negative reactance to the CDSS, when users did not follow a correct CDSS proposition (8.6% of decisions). Because of errors in patient data entry, OncoDoc2 delivered non-recommended propositions in 21.3% of decisions, leading to compliances with CDSS and CPGs of respectively 21.4% and 65.5%, whereas both compliances exceeded 90% when CDSS advices included CPG recommendations. Automation bias, when users followed an incorrect CDSS proposition explained the remaining non-compliance with CPGs (4.6% of decisions). Securing the use of CDSSs is of major importance to warranty patient safety and benefit of their potential to improve care.

LORD: a phenotype-genotype semantically integrated biomedical data tool to support rare disease diagnosis coding in health information systems.

Characterizing a rare disease diagnosis for a given patient is often made through expert's networks. It is a complex task that could evolve over time depending on the natural history of the disease and the evolution of the scientific knowledge. Most rare diseases have genetic causes and recent improvements of sequencing techniques contribute to the discovery of many new diseases every year. Diagnosis coding in the rare disease field requires data from multiple knowledge bases to be aggregated in order to offer the clinician a global information space from possible diagnosis to clinical signs (phenotypes) and known genetic mutations (genotype). Nowadays, the major barrier to the coding activity is the lack of consolidation of such information scattered in different thesaurus such as Orphanet, OMIM or HPO. The Linking Open data for Rare Diseases (LORD) web portal we developed stands as the first attempt to fill this gap by offering an integrated view of 8,400 rare diseases linked to more than 14,500 signs and 3,270 genes. The application provides a browsing feature to navigate through the relationships between diseases, signs and genes, and some Application Programming Interfaces to help its integration in health information systems in routine.

Implementation and management of a biomedical observation dictionary in a large healthcare information system.

OBJECTIVE: This study shows the evolution of a biomedical observation dictionary within the Assistance Publique Hôpitaux Paris (AP-HP), the largest European university hospital group. The different steps are detailed as follows: the dictionary creation, the mapping to logical observation identifier names and codes (LOINC), the integration into a multiterminological management platform and, finally, the implementation in the health information system. METHODS: AP-HP decided to create a biomedical observation dictionary named AnaBio, to map it to LOINC and to maintain the mapping. A management platform based on methods used for knowledge engineering has been put in place. It aims at integrating AnaBio within the health information system and improving both the quality and stability of the dictionary. RESULTS: This new management platform is now active in AP-HP. The AnaBio dictionary is shared by 120 laboratories and currently includes 50 000 codes. The mapping implementation to LOINC reaches 40% of the AnaBio entries and uses 26% of LOINC records. The results of our work validate the choice made to develop a local dictionary aligned with LOINC. DISCUSSION AND CONCLUSIONS: This work constitutes a first step towards a wider use of the platform. The next step will support the entire biomedical production chain, from the clinician prescription, through laboratory tests tracking in the laboratory information system to the communication of results and the use for decision support and biomedical research. In addition, the increase in the mapping implementation to LOINC ensures the interoperability allowing communication with other international health institutions.

Implementation of a platform dedicated to the biomedical analysis terminologies management.

BACKGROUND AND OBJECTIVES: Assistance Publique - Hôpitaux de Paris (AP-HP) is implementing a new laboratory management system (LMS) common to the 12 hospital groups. First step to this process was to acquire a biological analysis dictionary. This dictionary is interfaced with the international nomenclature LOINC, and has been developed in collaboration with experts from all biological disciplines. In this paper we describe in three steps (modeling, data migration and integration/verification) the implementation of a platform for publishing and maintaining the AP-HP laboratory data dictionary (AnaBio). MATERIAL AND METHODS: Due to data complexity and volume, setting up a platform dedicated to the terminology management was a key requirement. This is an enhancement tackling identified weaknesses of previous spreadsheet tool. Our core model allows interoperability regarding data exchange standards and dictionary evolution. RESULTS: We completed our goals within one year. In addition, structuring data representation has lead to a significant data quality improvement (impacting more than 10% of data). The platform is active in the 21 hospitals of the institution spread into 165 laboratories.