Analysis of free text in electronic health records for identification of cancer patient trajectories

Abstract

With an aging patient population and increasing complexity in patient disease trajectories, physicians are often met with complex patient histories from which clinical decisions must be made. Due to the increasing rate of adverse events and hospitals facing financial penalties for readmission, there has never been a greater need to enforce evidence-led medical decision-making using available health care data. In the present work, we studied a cohort of 7,741 patients, of whom 4,080 were diagnosed with cancer, surgically treated at a University Hospital in the years 2004–2012. We have developed a methodology that allows disease trajectories of the cancer patients to be estimated from free text in electronic health records (EHRs). By using these disease trajectories, we predict 80% of patient events ahead in time. By control of confounders from 8326 quantified events, we identified 557 events that constitute high subsequent risks (risk > 20%), including six events for cancer and seven events for metastasis. We believe that the presented methodology and findings could be used to improve clinical decision support and personalize trajectories, thereby decreasing adverse events and optimizing cancer treatment.

Visually guided classification trees for analyzing chronic patients

“…Statistical learning methods provide a way to create models (linear or nonlinear) to find patterns in data, and have shown potential to extract new knowledge from clinical datasets [32,33]. In the health care domain, the use of linear regression has been commonly used since it provides insight about the relevance of the features [34].…”

Design and Implementation of a Novel Electronic Health Record Tool to Enhance the Care of Individuals with Cystic Fibrosis: The Cystic Fibrosis Note Template

“…Although dictation into the EHR is faster than typing for many clinicians, data entered in this manner cannot easily be migrated to disease-specific registries, perhaps with the exception of using natural language processing (NLP) algorithms for data extraction. [50][51][52] To the best of our knowledge, NLP has not yet been successfully used to accomplish this for the CFFPR but has shown promise in other areas such as rheumatology. 50 The checkboxes, macros, and form fields in the CF NoteWriter template capture discrete data elements while concurrently generating the office note; however, some clinicians might desire greater flexibility in their approach to note writing than the tool currently provides.…”

Using case-level context to classify cancer pathology reports

“…Within the same sequence, EHRs are generally related to each other in some manner; for example, the diagnosis of a disease in one EHR may indicate additional tests for that disease in following EHRs, and later EHRs may document the treatment or progression of the disease. In some applications, for any given clinical report, it is helpful or necessary to extract aggregate information using other reports in the sequence [4][5][6]. An important example is cancer pathology reports-individual cancer pathology reports may need to be tagged with aggregate labels that describe the cancer case as a whole, and these aggregate labels require collective analysis of all pathology reports belonging to a given cancer case.…”