Comparing generative and extractive approaches to information extraction from abstracts describing randomized clinical trials.

BACKGROUND: Systematic reviews of Randomized Controlled Trials (RCTs) are an important part of the evidence-based medicine paradigm. However, the creation of such systematic reviews by clinical experts is costly as well as time-consuming, and results can get quickly outdated after publication. Most RCTs are structured based on the Patient, Intervention, Comparison, Outcomes (PICO) framework and there exist many approaches which aim to extract PICO elements automatically. The automatic extraction of PICO information from RCTs has the potential to significantly speed up the creation process of systematic reviews and this way also benefit the field of evidence-based medicine. RESULTS: Previous work has addressed the extraction of PICO elements as the task of identifying relevant text spans or sentences, but without populating a structured representation of a trial. In contrast, in this work, we treat PICO elements as structured templates with slots to do justice to the complex nature of the information they represent. We present two different approaches to extract this structured information from the abstracts of RCTs. The first approach is an extractive approach based on our previous work that is extended to capture full document representations as well as by a clustering step to infer the number of instances of each template type. The second approach is a generative approach based on a seq2seq model that encodes the abstract describing the RCT and uses a decoder to infer a structured representation of a trial including its arms, treatments, endpoints and outcomes. Both approaches are evaluated with different base models on a manually annotated dataset consisting of RCT abstracts on an existing dataset comprising 211 annotated clinical trial abstracts for Type 2 Diabetes and Glaucoma. For both diseases, the extractive approach (with flan-t5-base) reached the best F 1 score, i.e. 0.547 ( ± 0.006 ) for type 2 diabetes and 0.636 ( ± 0.006 ) for glaucoma. Generally, the F 1 scores were higher for glaucoma than for type 2 diabetes and the standard deviation was higher for the generative approach. CONCLUSION: In our experiments, both approaches show promising performance extracting structured PICO information from RCTs, especially considering that most related work focuses on the far easier task of predicting less structured objects. In our experimental results, the extractive approach performs best in both cases, although the lead is greater for glaucoma than for type 2 diabetes. For future work, it remains to be investigated how the base model size affects the performance of both approaches in comparison. Although the extractive approach currently leaves more room for direct improvements, the generative approach might benefit from larger models.

Commonsense knowledge in cognitive robotics: a systematic literature review.

One of the big challenges in robotics is the generalization necessary for performing unknown tasks in unknown environments on unknown objects. For us humans, this challenge is simplified by the commonsense knowledge we can access. For cognitive robotics, representing and acquiring commonsense knowledge is a relevant problem, so we perform a systematic literature review to investigate the current state of commonsense knowledge exploitation in cognitive robotics. For this review, we combine a keyword search on six search engines with a snowballing search on six related reviews, resulting in 2,048 distinct publications. After applying pre-defined inclusion and exclusion criteria, we analyse the remaining 52 publications. Our focus lies on the use cases and domains for which commonsense knowledge is employed, the commonsense aspects that are considered, the datasets/resources used as sources for commonsense knowledge and the methods for evaluating these approaches. Additionally, we discovered a divide in terminology between research from the knowledge representation and reasoning and the cognitive robotics community. This divide is investigated by looking at the extensive review performed by Zech et al. (The International Journal of Robotics Research, 2019, 38, 518-562), with whom we have no overlapping publications despite the similar goals.

Patient listening on social media for patient-focused drug development: a synthesis of considerations from patients, industry and regulators.

Patients, life science industry and regulatory authorities are united in their goal to reduce the disease burden of patients by closing remaining unmet needs. Patients have, however, not always been systematically and consistently involved in the drug development process. Recognizing this gap, regulatory bodies worldwide have initiated patient-focused drug development (PFDD) initiatives to foster a more systematic involvement of patients in the drug development process and to ensure that outcomes measured in clinical trials are truly relevant to patients and represent significant improvements to their quality of life. As a source of real-world evidence (RWE), social media has been consistently shown to capture the first-hand, spontaneous and unfiltered disease and treatment experience of patients and is acknowledged as a valid method for generating patient experience data by the Food and Drug Administration (FDA). While social media listening (SML) methods are increasingly applied to many diseases and use cases, a significant piece of uncertainty remains on how evidence derived from social media can be used in the drug development process and how it can impact regulatory decision making, including legal and ethical aspects. In this policy paper, we review the perspectives of three key stakeholder groups on the role of SML in drug development, namely patients, life science companies and regulators. We also carry out a systematic review of current practices and use cases for SML and, in particular, highlight benefits and drawbacks for the use of SML as a way to identify unmet needs of patients. While we find that the stakeholders are strongly aligned regarding the potential of social media for PFDD, we identify key areas in which regulatory guidance is needed to reduce uncertainty regarding the impact of SML as a source of patient experience data that has impact on regulatory decision making.

Automatic knowledge graph population with model-complete text comprehension for pre-clinical outcomes in the field of spinal cord injury.

The paradigm of evidence-based medicine requires that medical decisions are made on the basis of the best available knowledge published in the literature. Existing evidence is often summarized in the form of systematic reviews and/or meta-reviews and is rarely available in a structured form. Manual compilation and aggregation is costly, and conducting a systematic review represents a high effort. The need to aggregate evidence arises not only in the context of clinical trials, but is also important in the context of pre-clinical animal studies. In this context, evidence extraction is important to support translation of the most promising pre-clinical therapies into clinical trials or to optimize clinical trial design. Aiming at developing methods that facilitate the task of aggregating evidence published in pre-clinical studies, in this paper a new system is presented that automatically extracts structured knowledge from such publications and stores it in a so-called domain knowledge graph. The approach follows the paradigm of model-complete text comprehension by relying on guidance from a domain ontology creating a deep relational data-structure that reflects the main concepts, protocol, and key findings of studies. Focusing on the domain of spinal cord injuries, a single outcome of a pre-clinical study is described by up to 103 outcome parameters. Since the problem of extracting all these variables together is intractable, we propose a hierarchical architecture that incrementally predicts semantic sub-structures according to a given data model in a bottom-up fashion. At the heart of our approach is a statistical inference method that relies on conditional random fields to infer the most likely instance of the domain model given the text of a scientific publication as input. This approach allows modeling dependencies between the different variables describing a study in a semi-joint fashion. We present a comprehensive evaluation of our system to understand the extent to which our system can capture a study in the depth required to enable the generation of new knowledge. We conclude the article with a brief description of some applications of the populated knowledge graph and show the potential implications of our work for supporting evidence-based medicine.

An annotated corpus of clinical trial publications supporting schema-based relational information extraction.

BACKGROUND: The evidence-based medicine paradigm requires the ability to aggregate and compare outcomes of interventions across different trials. This can be facilitated and partially automatized by information extraction systems. In order to support the development of systems that can extract information from published clinical trials at a fine-grained and comprehensive level to populate a knowledge base, we present a richly annotated corpus at two levels. At the first level, entities that describe components of the PICO elements (e.g., population's age and pre-conditions, dosage of a treatment, etc.) are annotated. The second level comprises schema-level (i.e., slot-filling templates) annotations corresponding to complex PICO elements and other concepts related to a clinical trial (e.g. the relation between an intervention and an arm, the relation between an outcome and an intervention, etc.). RESULTS: The final corpus includes 211 annotated clinical trial abstracts with substantial agreement between annotators at the entity and scheme level. The mean Kappa value for the glaucoma and T2DM corpora was 0.74 and 0.68, respectively, for single entities. The micro-averaged F1 score to measure inter-annotator agreement for complex entities (i.e. slot-filling templates) was 0.81.The BERT-base baseline method for entity recognition achieved average micro- F1 scores of 0.76 for glaucoma and 0.77 for diabetes with exact matching. CONCLUSIONS: In this work, we have created a corpus that goes beyond the existing clinical trial corpora, since it is annotated in a schematic way that represents the classes and properties defined in an ontology. Although the corpus is small, it has fine-grained annotations and could be used to fine-tune pre-trained machine learning models and transformers to the specific task of extracting information about clinical trial abstracts.For future work, we will use the corpus for training information extraction systems that extract single entities, and predict template slot-fillers (i.e., class data/object properties) to populate a knowledge base that relies on the C-TrO ontology for the description of clinical trials. The resulting corpus and the code to measure inter-annotation agreement and the baseline method are publicly available at https://zenodo.org/record/6365890.

Synthesizing evidence from clinical trials with dynamic interactive argument trees.

BACKGROUND: Evidence-based medicine propagates that medical/clinical decisions are made by taking into account high-quality evidence, most notably in the form of randomized clinical trials. Evidence-based decision-making requires aggregating the evidence available in multiple trials to reach -by means of systematic reviews- a conclusive recommendation on which treatment is best suited for a given patient population. However, it is challenging to produce systematic reviews to keep up with the ever-growing number of published clinical trials. Therefore, new computational approaches are necessary to support the creation of systematic reviews that include the most up-to-date evidence.We propose a method to synthesize the evidence available in clinical trials in an ad-hoc and on-demand manner by automatically arranging such evidence in the form of a hierarchical argument that recommends a therapy as being superior to some other therapy along a number of key dimensions corresponding to the clinical endpoints of interest. The method has also been implemented as a web tool that allows users to explore the effects of excluding different points of evidence, and indicating relative preferences on the endpoints. RESULTS: Through two use cases, our method was shown to be able to generate conclusions similar to the ones of published systematic reviews. To evaluate our method implemented as a web tool, we carried out a survey and usability analysis with medical professionals. The results show that the tool was perceived as being valuable, acknowledging its potential to inform clinical decision-making and to complement the information from existing medical guidelines. CONCLUSIONS: The method presented is a simple but yet effective argumentation-based method that contributes to support the synthesis of clinical trial evidence. A current limitation of the method is that it relies on a manually populated knowledge base. This problem could be alleviated by deploying natural language processing methods to extract the relevant information from publications.

Lexical alignment in triadic communication.

Lexical alignment refers to the adoption of one's interlocutor's lexical items. Accounts of the mechanisms underlying such lexical alignment differ (among other aspects) in the role assigned to addressee-centered behavior. In this study, we used a triadic communicative situation to test which factors may modulate the extent to which participants' lexical alignment reflects addressee-centered behavior. Pairs of naïve participants played a picture matching game and received information about the order in which pictures were to be matched from a voice over headphones. On critical trials, participants did or did not hear a name for the picture to be matched next over headphones. Importantly, when the voice over headphones provided a name, it did not match the name that the interlocutor had previously used to describe the object. Participants overwhelmingly used the word that the voice over headphones provided. This result points to non-addressee-centered behavior and is discussed in terms of disrupting alignment with the interlocutor as well as in terms of establishing alignment with the voice over headphones. In addition, the type of picture (line drawing vs. tangram shape) independently modulated lexical alignment, such that participants showed more lexical alignment to their interlocutor for (more ambiguous) tangram shapes compared to line drawings. Overall, the results point to a rather large role for non-addressee-centered behavior during lexical alignment.

A computational model for the item-based induction of construction networks.

According to usage-based approaches to language acquisition, linguistic knowledge is represented in the form of constructions--form-meaning pairings--at multiple levels of abstraction and complexity. The emergence of syntactic knowledge is assumed to be a result of the gradual abstraction of lexically specific and item-based linguistic knowledge. In this article, we explore how the gradual emergence of a network consisting of constructions at varying degrees of complexity can be modeled computationally. Linguistic knowledge is learned by observing natural language utterances in an ambiguous context. To determine meanings of constructions starting from ambiguous contexts, we rely on the principle of cross-situational learning. While this mechanism has been implemented in several computational models, these models typically focus on learning mappings between words and referents. In contrast, in our model, we show how cross-situational learning can be applied consistently to learn correspondences between form and meaning beyond such simple correspondences.

Online semi-supervised growing neural gas.

In this paper we introduce online semi-supervised growing neural gas (OSSGNG), a novel online semi-supervised classification approach based on growing neural gas (GNG). Existing semi-supervised classification approaches based on GNG require that the training data is explicitly stored as the labeling is performed a posteriori after the training phase. As main contribution, we present an approach that relies on online labeling and prediction functions to process labeled and unlabeled data uniformly and in an online fashion, without the need to store any of the training examples explicitly. We show that using on-the-fly labeling strategies does not significantly deteriorate the performance of classifiers based on GNG, while circumventing the need to explicitly store training examples. Armed with this result, we then present a semi-supervised extension of GNG (OSSGNG) that relies on the above mentioned online labeling functions to label unlabeled examples and incorporate them into the model on-the-fly. As an important result, we show that OSSGNG performs as good as previous semi-supervised extensions of GNG which rely on offline labeling strategies. We also show that OSSGNG compares favorably to other state-of-the-art semi-supervised learning approaches on standard benchmarking datasets.