Innovative Asthma Ascertainment by Using Natural Language Processing Algorithm Enabling Automated Chart Review

Technology #2016-271

Mayo Clinic investigators have developed a novel natural language processing (NLP) algorithm designed to automate chart review in order to ascertain asthma status and date of onset by using free text from a patient’s medical record. An initial validation of this algorithm in a patient cohort showed a sensitivity and specificity of 89% and 97% (respectively) when compared to asthma status ascertained through manual chart review. Whereas, in the same cohort, when an ICD code alone was used to determine asthma status, the sensitivity (versus manual chart review) was only 31%, suggesting criterion validity. This tool can serve to identify undiagnosed children and adults with recurring symptoms, providing early access to preventative measures. Data generated through the use of this tool can also be utilized in large-scale population management and as a means to measure quality of care.