Klarigi: Characteristic explanations for semantic biomedical data

Annotation of biomedical entities with ontology classes provides for formal semantic analysis and mobilisation of background knowledge in determining their relationships. To date, enrichment analysis has been routinely employed to identify classes that are over-represented in annotations across sets of groups, such as biosample gene expression profiles or patient phenotypes, and is useful for a range of tasks including differential diagnosis and causative variant prioritisation. These approaches, however, usually consider only univariate relationships, make limited use of the semantic features of ontologies, and provide limited information and evaluation of the explanatory power of both singular and grouped candidate classes. Moreover, they are not designed to solve the problem of deriving cohesive, characteristic, and discriminatory sets of classes for entity groups. We have developed a new tool, called Klarigi, which introduces multiple scoring heuristics for identification of classes that are both compositional and discriminatory for groups of entities annotated with ontology classes. The tool includes a novel algorithm for derivation of multivariable semantic explanations for entity groups, makes use of semantic inference through live use of an ontology reasoner, and includes a classification method for identifying the discriminatory power of candidate sets, in addition to significance testing apposite to traditional enrichment approaches. We describe the design and implementation of Klarigi, including its scoring and explanation determination methods, and evaluate its use in application to two test cases with clinical significance, comparing and contrasting methods and results with literature-based and enrichment analysis methods. We demonstrate that Klarigi produces characteristic and discriminatory explanations for groups of biomedical entities in two settings. We also show that these explanations recapitulate and extend the knowledge held in existing biomedical databases and literature for several diseases. We conclude that Klarigi provides a distinct and valuable perspective on biomedical datasets when compared with traditional enrichment methods, and therefore constitutes a new method by which biomedical datasets can be explored, contributing to improved insight into semantic data.

Machine learning of COVID-19 clinical data identifies population structures with therapeutic potential

Clinical outcomes for patients with COVID-19 are heterogeneous and there is interest in defining subgroups for prognostic modeling and development of treatment algorithms. We obtained 28 demographic and laboratory variables in patients admitted to hospital with COVID-19. These comprised a training cohort (n = 6099) and two validation cohorts during the first and second waves of the pandemic (n = 996; n = 1011). Uniform manifold approximation and projection (UMAP) dimension reduction and Gaussian mixture model (GMM) analysis was used to define patient clusters. 29 clusters were defined in the training cohort and associated with markedly different mortality rates, which were predictive within confirmation datasets. Deconvolution of clinical features within clusters identified unexpected relationships between variables. Integration of large datasets using UMAP-assisted clustering can therefore identify patient subgroups with prognostic information and uncovers unexpected interactions between clinical variables. This application of machine learning represents a powerful approach for delineating disease pathogenesis and potential therapeutic interventions.

Evaluating semantic similarity methods for comparison of text-derived phenotype profiles

Background

Semantic similarity is a valuable tool for analysis in biomedicine. When applied to phenotype profiles derived from clinical text, they have the capacity to enable and enhance ‘patient-like me’ analyses, automated coding, differential diagnosis, and outcome prediction. While a large body of work exists exploring the use of semantic similarity for multiple tasks, including protein interaction prediction, and rare disease differential diagnosis, there is less work exploring comparison of patient phenotype profiles for clinical tasks. Moreover, there are no experimental explorations of optimal parameters or better methods in the area.

Methods

We develop a platform for reproducible benchmarking and comparison of experimental conditions for patient phentoype similarity. Using the platform, we evaluate the task of ranking shared primary diagnosis from uncurated phenotype profiles derived from all text narrative associated with admissions in the medical information mart for intensive care (MIMIC-III).

Results

300 semantic similarity configurations were evaluated, as well as one embedding-based approach. On average, measures that did not make use of an external information content measure performed slightly better, however the best-performing configurations when measured by area under receiver operating characteristic curve and Top Ten Accuracy used term-specificity and annotation-frequency measures.

Conclusion

We identified and interpreted the performance of a large number of semantic similarity configurations for the task of classifying diagnosis from text-derived phenotype profiles in one setting. We also provided a basis for further research on other settings and related tasks in the area.

Multi-faceted semantic clustering with text-derived phenotypes

Identification of ontology concepts in clinical narrative text enables the creation of phenotype profiles that can be associated with clinical entities, such as patients or drugs. Constructing patient phenotype profiles using formal ontologies enables their analysis via semantic similarity, in turn enabling the use of background knowledge in clustering or classification analyses. However, traditional semantic similarity approaches collapse complex relationships between patient phenotypes into a unitary similarity scores for each pair of patients. Moreover, single scores may be based only on matching terms with the greatest information content (IC), ignoring other dimensions of patient similarity. This process necessarily leads to a loss of information in the resulting representation of patient similarity, and is especially apparent when using very large text-derived and highly multi-morbid phenotype profiles. Moreover, it renders finding a biological explanation for similarity very difficult; the black box problem. In this article, we explore the generation of multiple semantic similarity scores for patients based on different facets of their phenotypic manifestation, which we define through different sub-graphs in the Human Phenotype Ontology. We further present a new methodology for deriving sets of qualitative class descriptions for groups of entities described by ontology terms. Leveraging this strategy to obtain meaningful explanations for our semantic clusters alongside other evaluation techniques, we show that semantic clustering with ontology-derived facets enables the representation, and thus identification of, clinically relevant phenotype relationships not easily recoverable using overall clustering alone. In this way, we demonstrate the potential of faceted semantic clustering for gaining a deeper and more nuanced understanding of text-derived patient phenotypes.

Perceptions of anonymised data use and awareness of the NHS data opt-out amongst patients, carers and healthcare staff

BACKGROUND

England operates a National Data Opt-Out (NDOO) for the secondary use of confidential health data for research and planning. We hypothesised that public awareness and support for the secondary use of health data and the NDOO would vary by participant demography and healthcare experience. We explored patient/public awareness and perceptions of secondary data use, grouping potential researchers into National Health Service (NHS), academia or commercial. We assessed awareness of the NDOO system amongst patients, carers, healthcare staff and the public. We co-developed recommendations to consider when sharing unconsented health data for research.

METHODS

A patient and public engagement program, co-created and including patient and public workshops, questionnaires and discussion groups regarding anonymised health data use.

RESULTS

There were 350 participants in total. Central concerns for health data use included unauthorised data re-use, the potential for discrimination and data sharing without patient benefit. 94% of respondents were happy for their data to be used for NHS research, 85% for academic research and 68% by health companies, but less than 50% for non-healthcare companies and opinions varied with demography and participant group. Questionnaires showed that knowledge of the NDOO was low, with 32% of all respondents, 53% of all NHS staff and 29% of all patients aware of the NDOO. Recommendations to guide unconsented secondary health data use included that health data use should benefit patients; data sharing decisions should involve patients/public. That data should remain in close proximity to health services with the principles of data minimisation applied. Further, that there should be transparency in secondary health data use, including publicly available lists of projects, summaries and benefits. Finally, organisations involved in data access decisions should participate in programmes to increase knowledge of the NDOO, to ensure public members were making informed choices about their own data.

CONCLUSION

The majority of participants in this study reported that the use of healthcare data for secondary purposes was acceptable when accessed by NHS. Academic and health-focused companies. However, awareness was limited, including of the NDOO. Further development of publicly-agreed recommendations for secondary health data use may improve both awareness and confidence in secondary health data use.

Infrastructure and operating processes of PIONEER, the HDR-UK Data Hub in Acute Care and the workings of the Data Trust Committee: a protocol paper

INTRODUCTION

Health Data Research UK designated seven UK-based Hubs to facilitate health data use for research. PIONEER is the Hub in Acute Care. PIONEER delivered workshops where patients/public citizens agreed key principles to guide access to unconsented, anonymised, routinely collected health data. These were used to inform the protocol.

METHODS

This paper describes the PIONEER infrastructure and data access processes. PIONEER is a research database and analytical environment that links routinely collected health data across community, ambulance and hospital healthcare providers. PIONEER aims ultimately to improve patient health and care, by making health data discoverable and accessible for research by National Health Service, academic and commercial organisations. The PIONEER protocol incorporates principles identified in the public/patient workshops. This includes all data access requests being reviewed by the Data Trust Committee, a group of public citizens who advise on whether requests should be supported prior to licensed access.

ETHICS AND DISSEMINATION

East Midlands-Derby REC (20/EM/0158): Confidentiality Advisory Group (20/CAG/0084). www.PIONEERdatahub.co.uk.