A post-processing algorithm for building longitudinal medication dose data from extracted medication information using natural language processing from electronic health records

Abstract

ObjectiveWe developed a post-processing algorithm to convert raw natural language processing output from electronic health records into a usable format for analysis. This algorithm was specifically developed for creating datasets that can be used for medication-based studies. Materials and MethodsThe algorithm was developed using output from two natural language processing systems, MedXN and medExtractR. We extracted medication information from deidentified clinical notes from Vanderbilt's electronic health record system for two medications, tacrolimus and lamotrigine, which have widely different prescribing patterns. The algorithm consists of two parts. Part I parses the raw output and connects entities together and Part II removes redundancies and calculates dose intake and daily dose. We evaluated both parts of the algorithm by comparing to gold standards that were generated using approximately 300 records from 10 subjects for both medications and both NLP systems. ResultsBoth parts of the algorithm performed well. For MedXN, the F-measures for Part I were at or above 0.94 and for Part II they were at or above 0.98. For medExtractR the Fmeasures for Part I were at or above 0.98 and for Part II they were at or above 0.91. DiscussionOur post-processing algorithm is useful for drug-based studies because it converts NLP output to analyzable data. It performed well, although it cannot handle highly complicated cases, which usually occurred when a NLP incorrectly extracted dose information. Future work will focus on identifying the most likely correct dose when conflicting doses are extracted on the same day. Part I Output:ID6_2015-12-28_note1.txt Lamictal 80 mcg 1 ID6_2015-12-28_note1.txt Lamictal 4.5 mcg

Development of a System for Postmarketing Population Pharmacokinetic and Pharmacodynamic Studies Using Real‐World Data From Electronic Health Records

“…As it was challenging to process the raw extracted data, especially for drugs prescribed multiple times a day, we developed a rigorous postprocessing algorithm that was implemented in Pro‐Med‐NLP. The details of this algorithm can be found in McNeer et al . Briefly, step 1 (Parsing) transforms each NLP output to a standardized form through its parse function written specific to each NLP system.…”

“…As it was challenging to process the raw extracted data, especially for drugs prescribed multiple times a day, we developed a rigorous postprocessing algorithm that was implemented in Pro-Med-NLP. The details of this algorithm can be found in McNeer et al 19 Briefly, step 1 (Parsing) transforms each NLP output to a standardized form through its parse function written specific to each NLP system. After this step, the next steps are common across all NLP systems, allowing greater generalizability, which include: step 2 (Pairing) uses regular expressions to match drug names of interest (e.g., "lamot|lamictal|ltg" for lamotrigine), and is able to handle special cases, such as missing frequency; and step 3 (Building) removes redundancies, and calculates dose intake and daily dose (Figure 3).…”

Sensitivity of Estimated Tacrolimus Population Pharmacokinetic Profile to Inaccurate Assumptions about Dose Timing and Absorption: An Investigation in Real-World and Simulated Data

“…We recently developed a natural language processing (NLP) system, medExtractR , 18 to extract medication information from free-text clinical notes as part of a system to enable the use of EHRs in retrospective studies of drugs. 19,20 The system, once finalized, should relieve the primary burden in data generation and manual extraction of medication data. In addition to drug dosing information, medExtractR is designed to extract explicit last dosing times (timing of the dose prior to a recorded blood concentration) if present in the notes.…”