Journal of biomedical informatics
-
Manually curating standardized phenotypic concepts such as Human Phenotype Ontology (HPO) terms from narrative text in electronic health records (EHRs) is time consuming and error prone. Natural language processing (NLP) techniques can facilitate automated phenotype extraction and thus improve the efficiency of curating clinical phenotypes from clinical texts. While individual NLP systems can perform well for a single cohort, an ensemble-based method might shed light on increasing the portability of NLP pipelines across different cohorts. ⋯ Our study demonstrates that ensembles of natural language processing can improve both generic phenotypic concept recognition and patient specific phenotypic concept identification over individual systems. Among the individual NLP systems, each individual system performed best when they were applied in the dataset that they were primary designed for. However, combining multiple NLP systems to create an ensemble can generally improve the performance. Specifically, the ensemble can increase the results reproducibility across different cohorts and tasks, and thus provide a more portable phenotyping solution compared to individual NLP systems.
-
Identifying medical persona from a social media post is critical for drug marketing, pharmacovigilance and patient recruitment. Medical persona classification aims to computationally model the medical persona associated with a social media post. We present a novel deep learning model for this task which consists of two parts: Convolutional Neural Networks (CNNs), which extract highly relevant features from the sentences of a social media post and average pooling, which aggregates the sentence embeddings to obtain task-specific document embedding. ⋯ We eliminate the need for manual labeling by employing a distant supervision based method to obtain labeled examples for training the models. We thoroughly analyze our model to discover cues that are indicative of a particular persona. Particularly, we use first derivative saliency to identify the salient words in a particular social media post.
-
Similarly to what already established and implemented in the United States, the concept of the Unique Device Identification (UDI) system has been introduced with the European Regulations for medical devices MDR (EU) 2017/745 and in-vitro diagnostic medical devices IVDR (EU) 2017/746 and it is on the way to become a worldwide standard. The aim of this work was to provide a possible approach for the implementation of UDI and traceability in Europe for standalone software medical devices according to lifecycle and quality system standards. ⋯ Among the various types of medical devices, software is an increasingly large reality with very specific characteristics that must be taken into consideration. All the relevant aspects for the implementation of the UDI should be taken into consideration to combine safety and feasibility in order to effectively pursue the traceability of these medical devices.
-
Over the last decades there has been an increasing interest in personalization: can we make sure that treatments are effective for individual patients? The quest for personalization affects biomedical informatics in two ways: first, we design systems-for example eHealth applications-that directly interact with patients and these systems might themselves one day be personalized. Hence, we seek effective methods to do so. ⋯ By clearly defining personalization and analyzing the effectiveness of different personalization methods this discussion highlights how we should embrace sequential experimentation-as opposed to the traditional randomized trial-if we want to personalize our informatics systems efficiently. Furthermore, we need to make sure that we capture the treatment assignment process in our health records: doing so will greatly increase the utility of the collected data for future personalization attempts.
-
One of the significant problems in the field of healthcare is the low survival rate of people who have experienced sudden cardiac arrest. Early prediction of cardiac arrest can provide the time required for intervening and preventing its onset in order to reduce mortality. Traditional statistical methods have been used to predict cardiac arrest. They have often analyzed group-level differences using a limited number of variables. On the other hand, machine learning approach, which is part of a growing trend of predictive medical analysis, has provided personalized predictive analyses on more complex data and produced remarkable results. ⋯ According to the results, machine learning techniques can improve the outcome of cardiac arrest prediction. However, future research should be carried out to evaluate the efficiency of rarely-used algorithms and to address the challenges of external validation, implementation and adoption of machine learning models in real clinical environments.