A natural language processing algorithm to measure quality prostate cancer care.

Abstract

232 Background: Electronic health records (EHRs) are a widely adopted but underutilized source of data for systematic assessment of healthcare quality. Barriers for use of this data source include its vast complexity, lack of structure, and the lack of use of standardized vocabulary and terminology by clinicians. This project aims to develop generalizable algorithms to extract useful knowledge regarding prostate cancer quality metrics from EHRs. Methods: We used EHR ICD-9/10 codes to identify prostate cancer patients receiving care at our academic medical center. Patients were confirmed in the California Cancer Registry (CCR), which provided data on tumor characteristics, treatment data, treatment outcomes and survival. We focused on three potential pretreatment process quality measures, which included documentation within 6 months prior to initial treatment of prostate-specific antigen (PSA), digital rectal exam (DRE) performance, and Gleason score. Each quality metric was defined using target terms and concepts to extract from the EHRs. Terms were mapped to a standardized medical vocabulary or ontology, enabling us to represent the metric elements by a concept domain and its permissible values. The structured representation of the quality metric included rules that accounted for the temporal order of the metric components. Our algorithms used natural language processing for free text annotation and negation, to ensure terms such as ‘DRE deferred’ are appropriately categorized. Results: We identified 2,123 patients receiving prostate cancer treatment between 2008-2016, of whom 1413 (67%) were matched in the CCR. We compared accuracy of our data mining algorithm, a random sample of manual chart review, and the CCR. (See Table.) Conclusions: EHR systems can be used to assess and report quality metrics systematically, efficiently, and with high accuracy. The development of such systems can improve and reduce the burden of quality reporting and potentially reduce costs of measuring quality metrics through automation. [Table: see text]

Leveraging Digital Data to Inform and Improve Quality Cancer Care

“…We have demonstrated the feasibility of our data-mining workflow to extract accurate, clinically meaningful information from EHR. [25][26][27][28] The key for data extraction is to transform patient encounters into a retrospective longitudinal record for each patient and identify cohorts of interest, known as clinical phenotyping, 17 using structured and unstructured data. The custom extractors we develop range in complexity based on the types of data and analytic methods required to identify and pull each variable at high fidelity.…”

A frame semantic overview of NLP-based information extraction for cancer-related EHR notes

“…Frame Elements References CODE TERMINOLOGY: clinical terminology used for the procedure mention [52] CODE VALUE: value of the terminology code [52], [1] INSTITUTION: institution where the procedure was performed [52] NEGATION: whether the mention is negated [52], [77] MENTION: words related to any procedure term (e.g. flex sig, guaiac card) [75], [52], [15], [84], [77], [43] MARGIN: usually the rim of normal tissue taken removed during or after procedure (surgical margin) [52] ANATOMICAL SITE: part of body procedure targets (e.g., breast) [79], [52] TEMPORAL INFORMATION: time and date descriptors (e.g., "colonoscopy in 2005", "flexible sigmoidoscopy 5 years ago), date of completion [52], [15], [1], [10] STATUS: procedure or treatment status (e.g., refused, declined, scheduled, planned, completed, reported vs not reported) [15], [42], [1], [82] MODIFIER: negation and other modifiers that change the status of procedure (e.g., "no", "never") [15] TUMOR DESCRIPTION ANATOMICAL SITE: anatomic locations (e.g., "segment 5" or "left lobe") with attributes (Liver, Non Liver), target location (e.g., liver and segment #7) as well as non-target location (e.g., breast) [13], [8], [52], [85], [42], [25], [9], [23], [26], [41] LATERALITY: side of a paired organ associated with origin of the primary tumor [25] TYPE: primary/metastatic [25] STATUS: benign or malignancy status along with diagno...…”

“…[79] TEST RESULT: result of the test (e.g. positive vs negative) [14], [82], [94], [83], [77] THERAPEUTIC PROCEDURE TYPE: treatment type (e.g., RFA and TACE) [42], [60], [50] LINE OF THERAPY: initial treatment is referred to as first-line treatment or first-line therapy, however, a secondline treatment may be suggested later [50] THERAPY DOSE: the total about of treatment (e.g., radiation) the patient is exposed to (e.g. radiation therapy dose) [10] TOXICITIES: various toxicities related to cancer treatment therapy along with negation and certainty [10] CANCER FINDING PROCEDURE: name of the associated cancer procedure (e.g., Breast Core biopsy), can be a separate frame (CANCER PROCEDURE) [7], [79], [71], [81], [11] FINDING TYPE: type of the finding (e.g., negative, normal, positive, possible, [80] Frame Frame Elements References probably, history, mild, stable, improved, or recommendation) FINDING MODIFIER: modifying words from within the report (e.g., type: mild, modifiers: tortuous) [80] LATERALITY: location or sidedness of the finding (e.g., "left", "right," "both", or "bilateral") [79] BODY PARTS: body organs on which the finding is reported [11] PATHOLOGY FINDING POSITIVE LYMPH NODES NUMBER: number of positive lymph nodes mentioned in the finding [71], [48], [65] POSITIVE LYMPH NODES STATUS: presence or absence of positive lymph nodes [64] LYMPH NODES REMOVED: number of lymph nodes removed [48] NUCLEAR GRADE: describes how closely the nuclei of cancer cells look like the nuclei of normal cells [71] PLOIDY: refers to amount of DNA the cancer cells contain [71] QUALITATIVE S-PHASE: indicator of tumor growth rate [71] BIOMARKER: name of the biomarker [12], [40] BIOMARKER TEST RESULTS / MUTATION STATUS: positive or negative for biomarkers...…”

“…Frame Elements References CODE TERMINOLOGY: clinical terminology used for the procedure mention [52] CODE VALUE: value of the terminology code [52], [1] INSTITUTION: institution where the procedure was performed [52] NEGATION: whether the mention is negated [52], [77] MENTION: words related to any procedure term (e.g. flex sig, guaiac card)…”

Is it possible to automatically assess pretreatment digital rectal examination documentation using natural language processing? A single-centre retrospective study

“…Since biomedical domains exhibit high degree of terminological variation, NLP’s precision becomes even more valuable based on its ability to automatically recognise all variants of domain language 15. There are a few studies16–20 using NLP techniques to extract concepts related to prostate cancer from EHRs for decision support, but the accuracy of DRE extraction was reported only in two studies 16 21. The features related to DRE reporting was also missing in those studies.…”