Neural translation and automated recognition of ICD-10 medical entities from natural language (Preprint)
BACKGROUND The recognition of medical entities from natural language is an ubiquitous problem in the medical field, with applications ranging from medical act coding to the analysis of electronic health data for public health. It is however a complex task usually requiring human expert intervention, thus making it expansive and time consuming. The recent advances in artificial intelligence, specifically the raise of deep learning methods, has enabled computers to make efficient decisions on a number of complex problems, with the notable example of neural sequence models and their powerful applications in natural language processing. They however require a considerable amount of data to learn from, which is typically their main limiting factor. However, the CépiDc stores an exhaustive database of death certificates at the French national scale, amounting to several millions of natural language examples provided with their associated human coded medical entities available to the machine learning practitioner. OBJECTIVE This article investigates the applications of deep neural sequence models to the medical entity recognition from natural language problem. METHODS The investigated dataset is based on every French death certificate from 2011 to 2016, containing information such as the subject’s age, gender, and the chain of events leading to his or her death both in French and encoded as ICD-10 medical entities, for a total of around 3 million observations. The task of automatically recognizing ICD-10 medical entities from the French natural language based chain of event is then formulated as a type of predictive modelling problem known as a sequence-to-sequence modelling problem. A deep neural network based model known as the Transformer is then slightly adapted and fit to the dataset. Its performance is then assessed on an exterior dataset and compared to the current state of the art. Confidence intervals for derived measurements are derived via bootstrap. RESULTS The proposed approach resulted in a test F-measure of .952 [.946, .957], which constitutes a significant improvement on the current state of the art and its previously reported 82.5 F-measure assessed on a comparable dataset. Such an improvement opens a whole field of new applications, from nosologist level automated coding to temporal harmonization of death statistics. CONCLUSIONS This article shows that deep artificial neural network can directly learn from voluminous datasets complex relationships between natural language and medical entities, without any explicit prior knowledge. Although not entirely free from mistakes, the derived model constitutes a powerful tool for automated coding of medical entities from medical language with promising potential applications.