JMIR Publications

ABSTRACT

Background:

Recent advances in Natural Language Processing (NLP) have boosted the interest of the medical community in their application to health care in general, and particularly to stroke, a medical emergency with a huge impact. In this context with a rapid pace of evolution, it is necessary to know and understand the experience already gathered by the medical and scientific community.

Objective:

Explore the studies made in the last 10 years using NLP to assist the management of stroke emergencies, in order to gain insights on state-of-the art, and the main contexts of application and software tools that are used.

Methods:

Data were extracted from Scopus and Medline through PubMed, using the keywords “Natural Language Processing” and “stroke”. Primary questions related to the phases, contexts, and types of textual data used in the studies, and secondary questions were also asked about the numerical and statistical methods, and software used to process the data. The extracted data were structured in tables, and their relative frequencies were calculated, as well as the relationships between categories through Multiple Correspondence Analysis.

Results:

Twenty-nine papers were included, most of them cohort studies on ischemic stroke, published in the last two years. The majority focused on the use of NLP to assist in the diagnostic phase, followed by the prognosis of outcomes, using text data from diagnostic reports, and in many cases annotations on medical images, too. The most frequent approach was based on general Machine Learning techniques, applied to the outcome of relatively simple NLP methods, with the support of ontologies and standard vocabularies. Although smaller in number, there is an increasing body of studies using Deep Learning architectures on numerical and vectorized representations of the texts, obtained with more sophisticated NLP tools.

Conclusions:

Studies focused on NLP applied to stroke show specific trends that can be compared to the more general application of Artificial Intelligence to stroke. The purpose of using NLP is often to improve processes in a clinical context, rather than assist in the rehabilitation process. The state of the art in NLP is represented by Deep Learning architectures, among which BERT has been found to be specially used in the medical field in general, and for stroke in particular, with an increasing focus on the processing of annotations in medical images.