Showing papers by "Sarvnaz Karimi published in 2021"
TL;DR: A Show, Tell and Summarise model is proposed that first generates findings from chest X-rays and then summarises them to provide impression section, overcoming the limitation of previous research in radiology report generation.
Abstract: Radiology plays a vital role in health care by viewing the human body for diagnosis, monitoring, and treatment of medical problems. In radiology practice, radiologists routinely examine medical images such as chest X-rays and describe their findings in the form of radiology reports. However, this task of reading medical images and summarising its insights is time consuming, tedious, and error-prone, which often represents a bottleneck in the clinical diagnosis process. A computer-aided diagnosis system which can automatically generate radiology reports from medical images can be of great significance in reducing workload, reducing diagnostic errors, speeding up clinical workflow, and helping to alleviate any shortage of radiologists. Existing research in radiology report generation focuses on generating the concatenation of the findings and impression sections. Also, existing work ignores important differences between normal and abnormal radiology reports. The text of normal and abnormal reports differs in style and it is difficult for a single model to learn both the text style and learn to transition from findings to impression. To alleviate these challenges, we propose a Show, Tell and Summarise model that first generates findings from chest X-rays and then summarises them to provide impression section. The proposed work generates the findings and impression sections separately, overcoming the limitation of previous research. Also, we use separate models for generating normal and abnormal radiology reports which provide true insight of model’s performance. Experimental results on the publicly available IU-CXR dataset show the effectiveness of our proposed model. Finally, we highlight limitations in the radiology report generation research and present recommendations for future work.
7 citations
TL;DR: In the 2019 N2C2/Open Health Natural Language Processing Shared Task, the median F1 score of all 17 participating teams was 76.59% as discussed by the authors, which was a statistically significant improvement over the baseline (P < 0.001).
Abstract: BACKGROUND The prognosis, diagnosis, and treatment of many genetic disorders and familial diseases significantly improve if the family history (FH) of a patient is known. Such information is often written in the free text of clinical notes. OBJECTIVE The aim of this study is to develop automated methods that enable access to FH data through natural language processing. METHODS We performed information extraction by using transformers to extract disease mentions from notes. We also experimented with rule-based methods for extracting family member (FM) information from text and coreference resolution techniques. We evaluated different transfer learning strategies to improve the annotation of diseases. We provided a thorough error analysis of the contributing factors that affect such information extraction systems. RESULTS Our experiments showed that the combination of domain-adaptive pretraining and intermediate-task pretraining achieved an F1 score of 81.63% for the extraction of diseases and FMs from notes when it was tested on a public shared task data set from the National Natural Language Processing Clinical Challenges (N2C2), providing a statistically significant improvement over the baseline (P<.001). In comparison, in the 2019 N2C2/Open Health Natural Language Processing Shared Task, the median F1 score of all 17 participating teams was 76.59%. CONCLUSIONS Our approach, which leverages a state-of-the-art named entity recognition model for disease mention detection coupled with a hybrid method for FM mention detection, achieved an effectiveness that was close to that of the top 3 systems participating in the 2019 N2C2 FH extraction challenge, with only the top system convincingly outperforming our approach in terms of precision.
5 citations
11 Jul 2021
TL;DR: In this article, the authors present a search system that given a patient profile searches over clinical trials for potential matches, enabling the users to leverage the powerful querying language that comes with Apache Lucene query syntax in combination with state-of-the-art Divergence From Randomness retrieval coupled with a BERT-based neural ranking component.
Abstract: With the advances in precision medicine, identifying clinical trials relevant to a specific patient profile becomes more challenging. Often very specific molecular-level patient features need to be matched for the trial to be deemed relevant. Clinical trials contain strict inclusion and exclusion criteria, often written in free-text. Patients profiles are also semi-structured, with some important information hidden in clinical notes. We present a search system that given a patient profile searches over clinical trials for potential matches. It enables the users to leverage the powerful querying language that comes with Apache Lucene query syntax in combination with state-of-the-art Divergence From Randomness retrieval coupled with a BERT-based neural ranking component. This system aims to assist in clinical decision making.
4 citations
26 Oct 2021
TL;DR: This paper proposed a two-step method to incorporate treatment into the query formulation and ranking, which achieved state-of-the-art performance for TREC Precision Medicine 2020, which used a zero-shot setup to incorporate the novel focus on treatments which did not exist in any previous TREC tracks.
Abstract: High-quality evidence from the biomedical literature is crucial for decision making of oncologists who treat cancer patients. Search for evidence on a specific treatment for a patient is the challenge set by the precision medicine track of TREC in 2020. To address this challenge, we propose a two-step method to incorporate treatment into the query formulation and ranking. Training of such ranking function uses a zero-shot setup to incorporate the novel focus on treatments which did not exist in any of the previous TREC tracks. Our treatment-aware neural reranking approach, FAT, achieves state-of-the-art effectiveness for TREC Precision Medicine 2020. Our analysis indicates that the BERT-based rerankers automatically learn to score documents through identifying concepts relevant to precision medicine, similar to hand-crafted heuristics successful in the earlier studies.
1 citations
01 Nov 2021
TL;DR: The authors proposed to learn a similarity metric for opinion similarity via fine-tuning the Sentence-BERT sentence embedding network based on review text and weak supervision by review ratings.
Abstract: For many NLP applications of online reviews, comparison of two opinion-bearing sentences is key. We argue that, while general purpose text similarity metrics have been applied for this purpose, there has been limited exploration of their applicability to opinion texts. We address this gap in the literature, studying: (1) how humans judge the similarity of pairs of opinion-bearing sentences; and, (2) the degree to which existing text similarity metrics, particularly embedding-based ones, correspond to human judgments. We crowdsourced annotations for opinion sentence pairs and our main findings are: (1) annotators tend to agree on whether or not opinion sentences are similar or different; and (2) embedding-based metrics capture human judgments of “opinion similarity” but not “opinion difference”. Based on our analysis, we identify areas where the current metrics should be improved. We further propose to learn a similarity metric for opinion similarity via fine-tuning the Sentence-BERT sentence-embedding network based on review text and weak supervision by review ratings. Experiments show that our learned metric outperforms existing text similarity metrics and especially show significantly higher correlations with human annotations for differing opinions.