Applying Intelligent Natural Language Processing (iNLP) to Public Health Case Detection and Notifications

Abstract: Limited funds, growing population needs, and the movement towards electronic health records and health information exchanges will require greater automation of public health operations. The requirement to shorten timeframes and improve the quality of reporting to local public health jurisdictions will place strains on limited public health resources. Novel methods are needed to reduce costs and shorten the timeframes needed to initiate public health responses. iNLP techniques are an emerging, powerful tool set to push forward Public Health informatics practice. A working prototype has been developed to show how advanced semantic relationships, when embedded within iNLP technology, can improve the overall detection of cases which should be of interest to public health entities. A by-product of this technology is a higher sensitivity through synonymy, a higher specificity through concept level understanding of the source records, and precision through distinguishing positive from negative from uncertain assertions in the identification of cases which should be of interest to public health. This should lead to more timely notification of local public health jurisdictions with improved public health interventions, and ultimately more timely notification of national public health authorities. National public health awareness can help to recognize cross jurisdictional spread of infections and can facilitate response by alerting as yet unaffected neighboring communities.