Utilizing uncoded consultation notes from electronic medical records for predictive modeling of colorectal cancer

ObjectiveMachine learning techniques can be used to extract predictive models for diseases from electronic medical records (EMRs). However, the nature of EMRs makes it difficult to apply off-the-shelf machine learning techniques while still exploiting the rich content of the EMRs. In this paper, we explore the usage of a range of natural language processing (NLP) techniques to extract valuable predictors from uncoded consultation notes and study whether they can help to improve predictive performance.MethodsWe study a number of existing techniques for the extraction of predictors from the consultation notes, namely a bag of words based approach and topic modeling. In addition, we develop a dedicated technique to match the uncoded consultation notes with a medical ontology. We apply these techniques as an extension to an existing pipeline to extract predictors from EMRs. We evaluate them in the context of predictive modeling for colorectal cancer (CRC), a disease known to be difficult to diagnose before performing an endoscopy.ResultsOur results show that we are able to extract useful information from the consultation notes. The predictive performance of the ontology-based extraction method moves significantly beyond the benchmark of age and gender alone (area under the receiver operating characteristic curve (AUC) of 0.870 versus 0.831). We also observe more accurate predictive models by adding features derived from processing the consultation notes compared to solely using coded data (AUC of 0.896 versus 0.882) although the difference is not significant. The extracted features from the notes are shown be equally predictive (i.e. there is no significant difference in performance) compared to the coded data of the consultations.ConclusionIt is possible to extract useful predictors from uncoded consultation notes that improve predictive performance. Techniques linking text to concepts in medical ontologies to derive these predictors are shown to perform best for predicting CRC in our EMR dataset.     

Implementing transitional care in skilled nursing facilities: Evaluation of a learning collaborative

Proctor's Framework for Implementation Research describes the role of implementation strategies and outcomes in the pathway from evidence-based interventions to service and client outcomes. This report describes the evaluation of a learning collaborative to implement a transitional care intervention in skilled nursing facilities (SNF). The collaborative protocol included implementation strategies to promote uptake of a transitional care intervention in SNFs. Using RE-AIM to evaluate outcomes, the main findings were intervention reach to 550 SNF patients, adoption in three of four SNFs that expressed interest in participation, and high fidelity to the implementation strategies. Fidelity to the transitional care intervention was moderate to high; SNF staff provided the five key components of the transitional care intervention for 64–93% of eligible patients. The evaluation was completed during the COVID-19 pandemic, which suggests the protocol was valued by staff and feasible to use amid serious internal and external challenges.     

Assessing individual environmental exposure derived from the spatiotemporal behavior context and its impacts on mental health

Research on the associations between environmental exposures and mental health has attracted considerable attention. Most studies to date have mainly estimated environmental health effects based on static geographic contexts (e.g., residential neighborhoods, administrative units), ignoring the dynamic nature of individual spatiotemporal behavior, which may lead to unreliable results. To address this limitation, this study collects survey data from 1003 adults in Guangzhou, China. Then, it delineates dynamic geographic context to capture individual daily activity and travel and assesses individual exposure to environmental factors derived from the home buffer (HB) and the time-weighted activity and travel buffer (TATB). Finally, multiple linear regression models are used in this paper to examine and compare the relationships between individual environmental exposure and mental health based on the HB and TATB. The results of this study indicate that there are great differences in individual environmental exposure levels based on the HB and TATB. The explanatory power of the environmental factors obtained from the TATB on mental health is greater than that derived from the HB. Specifically, exposures to some environmental factors (i.e., green space coverage, blue space coverage, fitness facility density, and recreational facility density) derived from the TATB have mental health-promoting effects, while exposures to the other environmental factors (i.e., public transit station density) have mental health-constraining effects. These findings enrich our knowledge of spatiotemporal behavior and the effects of the dynamic contextual environment on mental health, as well as provide valuable implications for urban planning and public health service.     

Graph convolution network with similarity awareness and adaptive calibration for disease-induced deterioration prediction

Graph convolution networks (GCN) have been successfully applied in disease prediction tasks as they capture interactions (i.e., edges and edge weights on the graph) between individual elements. The interactions in existing works are constructed by fusing similarity between imaging information and distance between non-imaging information, whereas disregarding the disease status of those individuals in the training set. Besides, the similarity is being evaluated by computing the correlation distance between feature vectors, which limits prediction performance, especially for predicting significant memory concern (SMC) and mild cognitive impairment (MCI). In this paper, we propose three mechanisms to improve GCN, namely similarity-aware adaptive calibrated GCN (SAC-GCN), for predicting SMC and MCI. First, we design a similarity-aware graph using different receptive fields to consider disease status. The labelled subjects on the graph are only connected with those labelled subjects with the same status. Second, we propose an adaptive mechanism to evaluate similarity. Specifically, we construct initial GCN with evaluating similarity by using traditional correlation distance, then pre-train the initial GCN by using training samples and use it to score all subjects. Then, the difference between these scores replaces correlation distance to update similarity. Last, we devise a calibration mechanism to fuse functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI) information into edges. The proposed method is tested on the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. Experimental results demonstrate that our proposed method is useful to predict disease-induced deterioration and superior to other related algorithms, with a mean classification accuracy of 86.83% in our prediction tasks.     

全表型组关联研究方法学研究进展

全表型组关联研究(PheWAS)是一种反向遗传学分析方法, 旨在研究哪些表型可能与给定的遗传变异相关联。随着生物医疗数据库和电子病历信息的开放获取, PheWAS已逐渐成为探索暴露因素与多种健康结局之间关联的有效方法。这种方法具有同时探索某一种暴露与多种疾病表型之间的统计学关联的独特优势, 从而有助于揭示多重因果关联以及各疾病间共同的致病机制。然而, PheWAS目前也面临诸多挑战。该方法本身存在一定的局限性, 包括工具变量的选择是否具有代表性以及繁重的多重校正负担。此外, 如何应用生物学知识阐释研究结果是PheWAS的另一重点问题。本文将围绕PheWAS方法学进行概述, 以期为后续更好地开展PheWAS提供思路和建议。     

Asthma and landscape fire smoke: A Thoracic Society of Australia and New Zealand position statement

Landscape fires are increasing in frequency and severity globally. In Australia, extreme bushfires cause a large and increasing health and socioeconomic burden for communities and governments. People with asthma are particularly vulnerable to the effects of landscape fire smoke (LFS) exposure. Here, we present a position statement from the Thoracic Society of Australia and New Zealand. Within this statement we provide a review of the impact of LFS on adults and children with asthma, highlighting the greater impact of LFS on vulnerable groups, particularly older people, pregnant women and Aboriginal and Torres Strait Islander peoples. We also highlight the development of asthma on the background of risk factors (smoking, occupation and atopy). Within this document we present advice for asthma management, smoke mitigation strategies and access to air quality information, that should be implemented during periods of LFS. We promote clinician awareness, and the implementation of public health messaging and preparation, especially for people with asthma.     

ResGANet: Residual group attention network for medical image classification and segmentation

In recent years, deep learning technology has shown superior performance in different fields of medical image analysis. Some deep learning architectures have been proposed and used for computational pathology classification, segmentation, and detection tasks. Due to their simple, modular structure, most downstream applications still use ResNet and its variants as the backbone network. This paper proposes a modular group attention block that can capture feature dependencies in medical images in two independent dimensions: channel and space. By stacking these group attention blocks in ResNet-style, we obtain a new ResNet variant called ResGANet. The stacked ResGANet architecture has 1.51–3.47 times fewer parameters than the original ResNet and can be directly used for downstream medical image segmentation tasks. Many experiments show that the proposed ResGANet is superior to state-of-the-art backbone models in medical image classification tasks. Applying it to different segmentation networks can improve the baseline model in medical image segmentation tasks without changing the network architecture. We hope that this work provides a promising method for enhancing the feature representation of convolutional neural networks (CNNs) in the future.