Advances in knowledge discovery and data mining : ... Pacific-Asia Conference, PAKDD ..., proceedings. Pacific-Asia Conference on Knowledge Discovery and Data Mining最新文献
Pub Date : 2024-05-01DOI: 10.1007/978-981-97-2238-9_25
Gabriel Frattallone-Llado, Juyong Kim, Cheng Cheng, Diego Salazar, Smitha Edakalavan, Jeremy C Weiss
Textual data often describe events in time but frequently contain little information about their specific timing, whereas complementary structured data streams may have precise timestamps but may omit important contextual information. We investigate the problem in healthcare, where we produce clinician annotations of discharge summaries, with access to either unimodal (text) or multimodal (text and tabular) data, (i) to determine event interval timings and (ii) to train multimodal language models to locate those events in time. We find our annotation procedures, dashboard tools, and annotations result in high-quality timestamps. Specifically, the multimodal approach produces more precise timestamping, with uncertainties of the lower bound, upper bounds, and duration reduced by 42% (95% CI 34-51%), 36% (95% CI 28-44%), and 13% (95% CI 10-17%), respectively. In the classification version of our task, we find that, trained on our annotations, our multimodal BERT model outperforms unimodal BERT model and Llama-2 encoder-decoder models with improvements in F1 scores for upper (10% and 61%, respectively) and lower bounds (8% and 56%, respectively). The code for the annotation tool and the BERT model is available (link).
文本数据通常按时间描述事件，但往往很少包含有关其具体时间的信息，而互补的结构化数据流可能有精确的时间戳，但可能遗漏重要的上下文信息。我们研究了医疗保健领域的这一问题，我们通过访问单模态（文本）或多模态（文本和表格）数据，对出院摘要进行临床医生注释，(i) 确定事件间隔时间，(ii) 训练多模态语言模型，以便及时定位这些事件。我们发现，我们的注释程序、仪表板工具和注释可生成高质量的时间戳。具体来说，多模态方法产生了更精确的时间戳，下限、上限和持续时间的不确定性分别降低了 42% (95% CI 34-51%)、36% (95% CI 28-44%) 和 13% (95% CI 10-17%)。在分类版本的任务中，我们发现，在我们的注释上进行训练后，我们的多模态 BERT 模型优于单模态 BERT 模型和 Llama-2 编码器-解码器模型，在上限（分别为 10%和 61%）和下限（分别为 8%和 56%）方面的 F1 分数有所提高。注释工具和 BERT 模型的代码已发布（链接）。
{"title":"Using Multimodal Data to Improve Precision of Inpatient Event Timelines.","authors":"Gabriel Frattallone-Llado, Juyong Kim, Cheng Cheng, Diego Salazar, Smitha Edakalavan, Jeremy C Weiss","doi":"10.1007/978-981-97-2238-9_25","DOIUrl":"10.1007/978-981-97-2238-9_25","url":null,"abstract":"<p><p>Textual data often describe events in time but frequently contain little information about their specific timing, whereas complementary structured data streams may have precise timestamps but may omit important contextual information. We investigate the problem in healthcare, where we produce clinician annotations of discharge summaries, with access to either unimodal (text) or multimodal (text and tabular) data, (i) to determine event interval timings and (ii) to train multimodal language models to locate those events in time. We find our annotation procedures, dashboard tools, and annotations result in high-quality timestamps. Specifically, the multimodal approach produces more precise timestamping, with uncertainties of the lower bound, upper bounds, and duration reduced by 42% (95% CI 34-51%), 36% (95% CI 28-44%), and 13% (95% CI 10-17%), respectively. In the classification version of our task, we find that, trained on our annotations, our multimodal BERT model outperforms unimodal BERT model and Llama-2 encoder-decoder models with improvements in F1 scores for upper (10% and 61%, respectively) and lower bounds (8% and 56%, respectively). The code for the annotation tool and the BERT model is available (link).</p>","PeriodicalId":517371,"journal":{"name":"Advances in knowledge discovery and data mining : ... Pacific-Asia Conference, PAKDD ..., proceedings. Pacific-Asia Conference on Knowledge Discovery and Data Mining","volume":"14648 ","pages":"322-334"},"PeriodicalIF":0.0,"publicationDate":"2024-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11228894/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141565577","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-01Epub Date: 2023-05-27DOI: 10.1007/978-3-031-33374-3_34
Zhiqi Bu, Zongyu Dai, Yiliang Zhang, Qi Long
Multiple imputation (MI) has been widely applied to missing value problems in biomedical, social and econometric research, in order to avoid improper inference in the downstream data analysis. In the presence of high-dimensional data, imputation models that include feature selection, especially regularized regression (such as Lasso, adaptive Lasso, and Elastic Net), are common choices to prevent the model from underdetermination. However, conducting MI with feature selection is difficult: existing methods are often computationally inefficient and poor in performance. We propose MISNN, a novel and efficient algorithm that incorporates feature selection for MI. Leveraging the approximation power of neural networks, MISNN is a general and flexible framework, compatible with any feature selection method, any neural network architecture, high/low-dimensional data and general missing patterns. Through empirical experiments, MISNN has demonstrated great advantages over state-of-the-art imputation methods (e.g. Bayesian Lasso and matrix completion), in terms of imputation accuracy, statistical consistency and computation speed.
{"title":"MISNN: Multiple Imputation via Semi-parametric Neural Networks.","authors":"Zhiqi Bu, Zongyu Dai, Yiliang Zhang, Qi Long","doi":"10.1007/978-3-031-33374-3_34","DOIUrl":"10.1007/978-3-031-33374-3_34","url":null,"abstract":"<p><p>Multiple imputation (MI) has been widely applied to missing value problems in biomedical, social and econometric research, in order to avoid improper inference in the downstream data analysis. In the presence of high-dimensional data, imputation models that include feature selection, especially <math><mrow><msub><mo>ℓ</mo><mn>1</mn></msub></mrow></math> regularized regression (such as Lasso, adaptive Lasso, and Elastic Net), are common choices to prevent the model from underdetermination. However, conducting MI with feature selection is difficult: existing methods are often computationally inefficient and poor in performance. We propose MISNN, a novel and efficient algorithm that incorporates feature selection for MI. Leveraging the approximation power of neural networks, MISNN is a general and flexible framework, compatible with any feature selection method, any neural network architecture, high/low-dimensional data and general missing patterns. Through empirical experiments, MISNN has demonstrated great advantages over state-of-the-art imputation methods (e.g. Bayesian Lasso and matrix completion), in terms of imputation accuracy, statistical consistency and computation speed.</p>","PeriodicalId":517371,"journal":{"name":"Advances in knowledge discovery and data mining : ... Pacific-Asia Conference, PAKDD ..., proceedings. Pacific-Asia Conference on Knowledge Discovery and Data Mining","volume":"13935 ","pages":"430-442"},"PeriodicalIF":0.0,"publicationDate":"2023-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10869892/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139901038","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Advances in knowledge discovery and data mining : ... Pacific-Asia Conference, PAKDD ..., proceedings. Pacific-Asia Conference on Knowledge Discovery and Data Mining
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号