Maria Kvist

Automatic recognition of disorders, findings, pharmaceuticals and body structures from clinical text

Automatic recognition of clinical entities in the narrative text of health records is useful for constructing applications for documentation of patient care, as well as for secondary usage in the form of medical knowledge extraction. There are a number of named entity recognition studies on English clinical text, but less work has been carried out on clinical text in other languages. This study was performed on Swedish health records, and focused on four entities that are highly relevant for constructing a patient overview and for medical hypothesis generation, namely the entities: Disorder, Finding, Pharmaceutical Drug and Body Structure. The study had two aims: to explore how well named entity recognition methods previously applied to English clinical text perform on similar texts written in Swedish; and to evaluate whether it is meaningful to divide the more general category Medical Problem, which has been used in a number of previous studies, into the two more granular entities, Disorder and Finding. Clinical notes from a Swedish internal medicine emergency unit were annotated for the four selected entity categories, and the inter-annotator agreement between two pairs of annotators was measured, resulting in an average F-score of 0.79 for Disorder, 0.66 for Finding, 0.90 for Pharmaceutical Drug and 0.80 for Body Structure. A subset of the developed corpus was thereafter used for finding suitable features for training a conditional random fields model. Finally, a new model was trained on this subset, using the best features and settings, and its ability to generalise to held-out data was evaluated. This final model obtained an F-score of 0.81 for Disorder, 0.69 for Finding, 0.88 for Pharmaceutical Drug, 0.85 for Body Structure and 0.78 for the combined category Disorder+Finding. The obtained results, which are in line with or slightly lower than those for similar studies on English clinical text, many of them conducted using a larger training data set, show that the approaches used for English are also suitable for Swedish clinical text. However, a small proportion of the errors made by the model are less likely to occur in English text, showing that results might be improved by further tailoring the system to clinical Swedish. The entity recognition results for the individual entities Disorder and Finding show that it is meaningful to separate the general category Medical Problem into these two more granular entity types, e.g. for knowledge mining of co-morbidity relations and disorder-finding relations.

Identifying adverse drug event information in clinical notes with distributional semantic representations of context

For the purpose of post-marketing drug safety surveillance, which has traditionally relied on the voluntary reporting of individual cases of adverse drug events (ADEs), other sources of information are now being explored, including electronic health records (EHRs), which give us access to enormous amounts of longitudinal observations of the treatment of patients and their drug use. Adverse drug events, which can be encoded in EHRs with certain diagnosis codes, are, however, heavily underreported. It is therefore important to develop capabilities to process, by means of computational methods, the more unstructured EHR data in the form of clinical notes, where clinicians may describe and reason around suspected ADEs. In this study, we report on the creation of an annotated corpus of Swedish health records for the purpose of learning to identify information pertaining to ADEs present in clinical notes. To this end, three key tasks are tackled: recognizing relevant named entities (disorders, symptoms, drugs), labeling attributes of the recognized entities (negation, speculation, temporality), and relationships between them (indication, adverse drug event). For each of the three tasks, leveraging models of distributional semantics - i.e., unsupervised methods that exploit co-occurrence information to model, typically in vector space, the meaning of words - and, in particular, combinations of such models, is shown to improve the predictive performance. The ability to make use of such unsupervised methods is critical when faced with large amounts of sparse and high-dimensional data, especially in domains where annotated resources are scarce.

Recent Advances in Clinical Natural Language Processing in Support of Semantic Analysis.

OBJECTIVES We present a review of recent advances in clinical Natural Language Processing (NLP), with a focus on semantic analysis and key subtasks that support such analysis. METHODS We conducted a literature review of clinical NLP research from 2008 to 2014, emphasizing recent publications (2012-2014), based on PubMed and ACL proceedings as well as relevant referenced publications from the included papers. RESULTS Significant articles published within this time-span were included and are discussed from the perspective of semantic analysis. Three key clinical NLP subtasks that enable such analysis were identified: 1) developing more efficient methods for corpus creation (annotation and de-identification), 2) generating building blocks for extracting meaning (morphological, syntactic, and semantic subtasks), and 3) leveraging NLP for clinical utility (NLP applications and infrastructure for clinical use cases). Finally, we provide a reflection upon most recent developments and potential areas of future NLP development and applications. CONCLUSIONS There has been an increase of advances within key NLP subtasks that support semantic analysis. Performance of NLP semantic analysis is, in many cases, close to that of agreement between humans. The creation and release of corpora annotated with complex semantic information models has greatly supported the development of new tools and approaches. Research on non-English languages is continuously growing. NLP methods have sometimes been successfully employed in real-world clinical tasks. However, there is still a gap between the development of advanced resources and their utilization in clinical settings. A plethora of new clinical use cases are emerging due to established health care initiatives and additional patient-generated sources through the extensive use of social media and other devices.

Diagnosis code assignment support using random indexing of patient records: a qualitative feasibility study

The prediction of diagnosis codes is typically based on free-text entries in clinical documents. Previous attempts to tackle this problem range from strictly rule-based systems to utilizing various classification algorithms, resulting in varying degrees of success. A novel approach is to build a word space model based on a corpus of coded patient records, associating co-occurrences of words and ICD-10 codes. Random Indexing is a computationally efficient implementation of the word space model and may prove an effective means of providing support for the assignment of diagnosis codes. The method is here qualitatively evaluated for its feasibility by a physician on clinical records from two Swedish clinics. The assigned codes were in this initial experiment found among the top 10 generated suggestions in 20% of the cases, but a partial match in 77% demonstrates the potential of the method.

Improving Readability of Swedish Electronic Health Records through Lexical Simplification: First Results

This paper describes part of an ongoing effort to improve the readability of Swedish electronic health records (EHRs). An EHR contains systematic documentation of a single patient’s medical history across time, entered by healthcare professionals with the purpose of enabling safe and informed care. Linguistically, medical records exemplify a highly specialised domain, which can be superficially characterised as having telegraphic sentences involving displaced or missing words, abundant abbreviations, spelling variations including misspellings, and terminology. We report results on lexical simplification of Swedish EHRs, by which we mean detecting the unknown, out-ofdictionary words and trying to resolve them either as compounded known words, abbreviations or misspellings.

Medical text simplification using synonym replacement: Adapting assessment of word difficulty to a compounding language

Medical texts can be difficult to understand for laymen, due to a frequent occurrence of specialised medical terms. Replacing these difficult terms with easier synonyms can, however, lead to improved readability. In this study, we have adapted a method for assessing difficulty of words to make it more suitable to medical Swedish. The difficulty of a word was assessed not only by measuring the frequency of the word in a general corpus, but also by measuring the frequency of substrings of words, thereby adapting the method to the compounding nature of Swedish. All words having a MeSH synonym that was assessed as easier, were replaced in a corpus of medical text. According to the readability measure LIX, the replacement resulted in a slightly more difficult text, while the readability increased according to the OVIX measure and to a preliminary reader study.

Cue-based assertion classification for Swedish clinical text-Developing a lexicon for pyConTextSwe

OBJECTIVE The ability of a cue-based system to accurately assert whether a disorder is affirmed, negated, or uncertain is dependent, in part, on its cue lexicon. In this paper, we continue our study of porting an assertion system (pyConTextNLP) from English to Swedish (pyConTextSwe) by creating an optimized assertion lexicon for clinical Swedish. METHODS AND MATERIAL We integrated cues from four external lexicons, along with generated inflections and combinations. We used subsets of a clinical corpus in Swedish. We applied four assertion classes (definite existence, probable existence, probable negated existence and definite negated existence) and two binary classes (existence yes/no and uncertainty yes/no) to pyConTextSwe. We compared pyConTextSwes performance with and without the added cues on a development set, and improved the lexicon further after an error analysis. On a separate evaluation set, we calculated the systems final performance. RESULTS Following integration steps, we added 454 cues to pyConTextSwe. The optimized lexicon developed after an error analysis resulted in statistically significant improvements on the development set (83% F-score, overall). The systems final F-scores on an evaluation set were 81% (overall). For the individual assertion classes, F-score results were 88% (definite existence), 81% (probable existence), 55% (probable negated existence), and 63% (definite negated existence). For the binary classifications existence yes/no and uncertainty yes/no, final system performance was 97%/87% and 78%/86% F-score, respectively. CONCLUSIONS We have successfully ported pyConTextNLP to Swedish (pyConTextSwe). We have created an extensive and useful assertion lexicon for Swedish clinical text, which could form a valuable resource for similar studies, and which is publicly available.

Detection of healthcare-associated urinary tract infection in Swedish electronic health records.

The prevalence of healthcare-associated infections (HAI) stresses the need for automatic surveillance in order to follow the effect of preventive measures. A number of detection systems have been set up for several languages, but none is known for Swedish hospitals. We plan a series of infection type specific programs for detection of HAI in electronic health records at a Swedish university hospital. Also, we aim at detecting HAI for patients entering hospital with HAI from previous care, a task that is not often addressed. This first study aims at surveillance of healthcare-associated urinary tract infections. The created rule-based system depends on acquiring the essential clinical information, and a combination of data and text mining is used. The wide range of diverse clinics with different traditions of documentation poses difficulties for detection. Results from evaluation on 1,867 care episodes from Oncology and Surgery show high precision (0.98), specificity (0.99) and negative predictive value (0.99), but an intermediate recall (0.60). An error analysis of the evaluation is presented and discussed.

SCAN: A Swedish Clinical Abbreviation Normalizer

Abbreviations pose a challenge for information extraction systems. In clinical text, abbreviations are abundant, as this type of documentation is written under time-pressure. We report work on characterizing abbreviations in Swedish clinical text and the development of SCAN: a Swedish Clinical Abbreviation Normalizer, which is built for the purpose of improving information access systems in the clinical domain. The clinical domain includes several subdomains with differing vocabularies depending on the nature of the specialist work, and adaption of NLP-tools may consequently be necessary. We extend and adapt SCAN, and evaluate on two different clinical subdomains: emergency department (ED) and radiology (X-ray). Overall final results are 85% (ED) and 83% (X-ray) F1-measure on the task of abbreviation identification. We also evaluate coverage of abbreviation expansion candidates in existing lexical resources, and create two new, freely available, lexicons with abbreviations and their possible expansions for the two clinical subdomains.

Professional language in Swedish clinical text: Linguistic characterization and comparative studies

This study investigates the linguistic characteristics of Swedish clinical text in radiology reports and doctors daily notes from electronic health records (EHRs) in comparison to general Swedish and biomedical journal text. We quantify linguistic features through a comparative register analysis to determine how the free text of EHRs differ from general and biomedical Swedish text in terms of lexical complexity, word and sentence composition, and common sentence structures. The linguistic features are extracted using state-of-the-art computational tools: a tokenizer, a part-of-speech tagger, and scripts for statistical analysis. Results show that technical terms and abbreviations are more frequent in clinical text, and lexical variance is low. Moreover, clinical text frequently omit subjects, verbs, and function words resulting in shorter sentences. Clinical text not only differs from general Swedish, but also internally, across its sub-domains, e.g. sentences lacking verbs are significantly more frequent in radiology reports. These results provide a foundation for future development of automatic methods for EHR simplification or clarification.