Automatically extracting cancer disease characteristics from pathology reports into a Disease Knowledge Representation Model.

We introduce an extensible and modifiable knowledge representation model to represent cancer disease characteristics in a comparable and consistent fashion. We describe a system, MedTAS/P which automatically instantiates the knowledge representation model from free-text pathology reports. MedTAS/P is based on an open-source framework and its components use natural language processing principles, machine learning and rules to discover and populate elements of the model. To validate the model and measure the accuracy of MedTAS/P, we developed a gold-standard corpus of manually annotated colon cancer pathology reports. MedTAS/P achieves F1-scores of 0.97-1.0 for instantiating classes in the knowledge representation model such as histologies or anatomical sites, and F1-scores of 0.82-0.93 for primary tumors or lymph nodes, which require the extractions of relations. An F1-score of 0.65 is reported for metastatic tumors, a lower score predominantly due to a very small number of instances in the training and test sets.

Analysis of clinical uncertainties by health professionals and patients: an example from mental health.

BACKGROUND: The first step in practising Evidence Based Medicine (EBM) has been described as translating clinical uncertainty into a structured and focused clinical question that can be used to search the literature to ascertain or refute that uncertainty. In this study we focus on questions about treatments for schizophrenia posed by mental health professionals and patients to gain a deeper understanding about types of questions asked naturally, and whether they can be reformulated into structured and focused clinical questions. METHODS: From a survey of uncertainties about the treatment of schizophrenia we describe, categorise and analyse the type of questions asked by mental health professionals and patients about treatment uncertainties for schizophrenia. We explore the value of mapping from an unstructured to a structured framework, test inter-rater reliability for this task, develop a linguistic taxonomy, and cross tabulate that taxonomy with elements of a well structured clinical question. RESULTS: Few of the 78 Patients and 161 clinicians spontaneously asked well structured queries about treatment uncertainties for schizophrenia. Uncertainties were most commonly about drug treatments (45.3% of clinicians and 41% of patients), psychological therapies (19.9% of clinicians and 9% of patients) or were unclassifiable.(11.8% of clinicians and 16.7% of patients). Few naturally asked questions could be classified using the well structured and focused clinical question format (i.e. PICO format). A simple linguistic taxonomy better described the types of questions people naturally ask. CONCLUSION: People do not spontaneously ask well structured clinical questions. Other taxonomies may better capture the nature of questions. However, access to EBM resources is greatly facilitated by framing enquiries in the language of EBM, such as posing queries in PICO format. People do not naturally do this. It may be preferable to identify a way of searching the literature that more closely matches the way people naturally ask questions if access to information about treatments are to be made more broadly available.