A Metadata Extraction Approach for Clinical Case Reports to Enable Advanced Understanding of Biomedical Concepts

This method can help answer questions about diseases and pathologies by providing a structured view on disease symptoms, diagnostic and therapeutic procedures, and patient outcomes from clinical case reports. The main advantage of this technique is that it can be applied to any patient cohort in the biomedical literature. Making epidemiological studies of rare or emerging conditions more realizable.

This method makes subsequent text analysis workflows more intuitive and easier to implement. By making text data machine readable with rich metadata, we can establish clinical case reports as a findable, accessible, interoperable and reusable or fair data resource. Central to this method is the metadata template used to extract all of the clinical information contained within the selected clinical case reports.

Following the workflow as illustrated will allow the assembly of a clinical case report corpus. To enhance the find ability, accessibility, interoperability and re usability of the medical content from a clinical case report. Include contextual details in each field rather than providing terms only from a controlled vocabulary.

Omit commonly repeated words and phrases. In the annotation template, provide specific terms identified within a document usually in its header as key terms separated with a semicolon. Provide demographic values.

Specifically any text statements describing a patient's background including sex and/or gender, age ethnicity or nationality. Provide geographic locations mentioned within the clinical narrative other than specific institution addresses. Provide lifestyle values including any text statements describing frequent patient activities or behaviors relevant to their general health.

In practice, this frequently involves smoking or alcohol consumption habits but may also include sun exposure, diet or frequency of specific types of physical activity. Provide medical history values referring to family history including any text statements describing clinical observations of, and events experienced by siblings, parents and other family members. As well as genetic conditions and negative observations.

Provide values referring to the social history including any text statements describing the patient background not covered in the demography or lifestyle sections. The statements may include occupational history and social habits. Provide values referring to the patients medical and surgical history including any text statements describing medical observations, treatments or other events that took place prior to the beginning of the clinical presentation.

Specify one or more of the following 16 disease system categories based on those used in the international statistical classification of diseases and related health problems, revision 10 as indicated in the table. Provide details of all of the signs and symptoms including any text statements describing any medical observations of signs or symptoms beginning at the initial presentation about the onset, duration, severity or resolution of the symptoms as provided. Provide details of any comorbidities including any terms or phrases describing distinct diseases present at the time of the initial clinical presentation.

Provide details of all of the diagnostic techniques and procedures including the names of medical procedures completed for diagnostic purposes. Such as examinations, tests and imagining. As well as the conditions under which the tests were performed and the relevant, anatomical locations.

But excluding any test results. Provide details of the diagnosis. Including any text statements describing the diagnoses of the disease.

Even if the final diagnosis is ambiguous. Provide all of the laboratory values and test results. Including the names of the diagnostic tests, their values and the conditions under which the tests were performed.

The use of both numerical and qualitative values is acceptable. Provide details of the pathology. Including any text statements describing all of the results of the pathology and histology studies.

Such as gross pathology, immunology or microscopy studies. Some of the terms may overlap with those used in the diagnostic techniques and procedures section. Provide all of the pharmacological therapies.

Including any text statements describing the drug therapies used in the course of the treatment and when and how the therapies were stopped. Using general terms such as antibiotics or specific drug names. Provide all interventional procedures.

Including any text statements describing therapeutic procedures used in the course of treatment and when and how the procedures were stopped as well as any invasive procedures. The implantation of medical devices or procedures completed to facilitate other therapies. Provide the patient outcome.

Including any text statements describing the health of the patient as of the end of the clinical presentation described in the report. Including any follow up tests. Then, provide counts of all of the diagnostic images, figures, videos or animations and tables.

Including all counts of visual media included in the report in the indicated format. To distinguish between the figures and images, count the images as any products of clinical diagnostics. Including photographs, micro graphs, electrocardiogram rhythm images or other products of diagnostic imaging.

While the figures should be considered all other images. Generally including data plots and illustrations. In this table examples of two sets of completed clinical case report metadata annotations are shown.

The first example includes mock data to illustrate the ideal format of each value. While the second example contains values extracted from an actual published clinical case report on a rare condition;Acrodermatitis Enteropathica. This method facilitates both human and automated approaches to the annotation of free text clinical documents.

By establishing a structured metadata standard template that ensures the information contained within is both accessible and useful. The metadata generated by this method supports downstream text analysis such as named entity recognition. Enabling researchers to extract information on signs and symptoms, diagnostic techniques and procedures, and laboratory values across a patient cohort represented in the biomedical literature.