ABSTRACT
Objectives: We aimed to automate routine extraction of clinically relevant unstructured information from uro-oncological histopathology reports by applying rule-based and machine learning (ML)/deep learning (DL) methods to develop an oncology focused natural language processing (NLP) algorithm. Methods: Our algorithm employs a combination of a rule-based approach and support vector machines/neural networks (BioBert/Clinical BERT), and is optimised for accuracy. We randomly extracted 5772 uro-oncological histology reports from 2008 to 2018 from electronic health records (EHRs) and split the data into training and validation datasets in an 80:20 ratio. The training dataset was annotated by medical professionals and reviewed by cancer registrars. The validation dataset was annotated by cancer registrars and defined as the gold standard with which the algorithm outcomes were compared. The accuracy of NLP-parsed data was matched against these human annotation results. We defined an accuracy rate of >95% as "acceptable" by professional human extraction, as per our cancer registry definition. Results: There were 11 extraction variables in 268 free-text reports. We achieved an accuracy rate of between 61.2% and 99.0% using our algorithm. Of the 11 data fields, a total of 8 data fields met the acceptable accuracy standard, while another 3 data fields had an accuracy rate between 61.2% and 89.7%. Noticeably, the rule-based approach was shown to be more effective and robust in extracting variables of interest. On the other hand, ML/DL models had poorer predictive performances due to highly imbalanced data distribution and variable writing styles between different reports and data used for domain-specific pre-trained models. Conclusion: We designed an NLP algorithm that can automate clinical information extraction accurately from histopathology reports with an overall average micro accuracy of 93.3%.
ABSTRACT
INTRODUCTION: International guidelines recommend that interruptions to chest compressions are minimised during defibrillation. As a result, some resuscitation educators have adopted a more structured approach to defibrillation. One such approach is the 'C.O.A.C.H.E.D.' cognitive aid (Continue compressions, Oxygen away, All others away, Charging, Hands off, Evaluate, Defibrillate or Disarm). To date, there are no studies assessing the use of this cognitive aid. METHODS: This study utilised an Emergency Department in situ simulated model of cardiac arrest. The defibrillator used was a proprietary R-Series (Zoll, PA, USA) connected to a CS1201 rhythm generator (Symbio, Beaverton, OR, USA). The study cohorts were interdisciplinary advanced life support (ALS) providers. Paired providers were enrolled in a mechanical CPR (M-CPR) training programme with no feedback related to defibrillation performance. As part of this 6-month programme, serial defibrillation performance was assessed. The outcome measures were the length of 'peri-shock' pause and 'safety' of defibrillation practice. Comparative statistical analysis using the Mann-Whitney U-test was made between groups of providers with 'correct use or near correct' or 'entirely incorrect or absent' use of the cognitive aid. RESULTS: The C.O.A.C.H.E.D. cognitive aid was applied correctly in 92 of 109 defibrillations. Providers with correct cognitive aid use had a median length of peri-shock pause time of 6.0s (IQR 5.0-7.0). Providers with 'entirely incorrect or absent' cognitive aid use had a peri-shock pause time of 8.0s (IQRF 6.6-10.0) (p≤0.001). No unsafe defibrillation practices were observed. CONCLUSION: In this observational study of defibrillation performance, the use of the C.O.A.C.H.E.D. cognitive aid was associated with a significant decrease in the length of peri-shock pause. Therefore, we conclude that the use of a cognitive aid is appropriate for teaching and performing defibrillation.
