Abstract

Undisclosed allergic reactions of patients are a major risk when undertaking surgeries in hospitals. We present our early experience and preliminary findings for a Clinical Decision Support System (CDSS) being developed in a Norwegian Hospital Trust. The system incorporates unsupervised and supervised machine learning algorithms in combination with rule-based algorithms to identify and classify allergies of concern for anesthesia during surgery. Our approach is novel in that it utilizes unsupervised machine learning to analyze large corpora of narratives to automatically build a clinical language model containing words and phrases of which meanings and relative meanings are also learnt. It further implements a semi-automatic annotation scheme for efficient and interactive machine-learning, which to a large extent eliminates the substantial manual annotation (of clinical narratives) effort necessary for the training of supervised algorithms. Validation of system performance was performed through comparing allergies identified by the CDSS with a manual reference standard.

Recommended Citation

Berge, G., Granmo, O., & Tveit, T. (2017). Combining Unsupervised, Supervised, and Rule-based Algorithms for Text Mining of Electronic Health Records - A Clinical Decision Support System for Identifying and Classifying Allergies of Concern for Anesthesia During Surgery. In Paspallis, N., Raspopoulos, M. Barry, M. Lang, H. Linger, & C. Schneider (Eds.), Information Systems Development: Advances in Methods, Tools and Management (ISD2017 Proceedings). Larnaca, Cyprus: University of Central Lancashire Cyprus. ISBN: 978-9963-2288-3-6. http://aisel.aisnet.org/isd2014/proceedings2017/CogScience/2.

Paper Type

Event

Share

COinS
 

Combining Unsupervised, Supervised, and Rule-based Algorithms for Text Mining of Electronic Health Records - A Clinical Decision Support System for Identifying and Classifying Allergies of Concern for Anesthesia During Surgery

Undisclosed allergic reactions of patients are a major risk when undertaking surgeries in hospitals. We present our early experience and preliminary findings for a Clinical Decision Support System (CDSS) being developed in a Norwegian Hospital Trust. The system incorporates unsupervised and supervised machine learning algorithms in combination with rule-based algorithms to identify and classify allergies of concern for anesthesia during surgery. Our approach is novel in that it utilizes unsupervised machine learning to analyze large corpora of narratives to automatically build a clinical language model containing words and phrases of which meanings and relative meanings are also learnt. It further implements a semi-automatic annotation scheme for efficient and interactive machine-learning, which to a large extent eliminates the substantial manual annotation (of clinical narratives) effort necessary for the training of supervised algorithms. Validation of system performance was performed through comparing allergies identified by the CDSS with a manual reference standard.