Recent studies have introduced attention models for medical visual question answering (MVQA). In medical research, not only is the modeling of “visual attention” crucial, but the modeling of “question attention” is equally significant. To facilitate bidirectional reasoning in the attention processes involving medical images and questions, a new MVQA architecture, named MCAN, has been proposed. This architecture incorporated a cross-modal co-attention network, FCAF, which identifies key words in questions and principal parts in images. Through a meta-learning channel attention module (MLCA), weights were adaptively assigned to each word and region, reflecting the model’s focus on specific words and regions during reasoning. Additionally, this study specially designed and developed a medical domain-specific word embedding model, Med-GloVe, to further enhance the model’s accuracy and practical value. Experimental results indicated that MCAN proposed in this study improved the accuracy by 7.7% on free-form questions in the Path-VQA dataset, and by 4.4% on closed-form questions in the VQA-RAD dataset, which effectively improves the accuracy of the medical vision question answer.
With the increasing popularity of smart phones, the electronic test of clinical trials has become a common means of investigation research. The APP of REDCap system can quickly construct a multi-center questionnaire system and obtain a large quantity of reliable and complete questionnaire data, shortening the cost and cycle of research. This paper primarily introduces how to conduct research on electronic questionnaire based on mobile APPs of REDCap system.
ObjectiveTo systematically review the efficacy of Tai Chi on patients with heart failure.MethodsDatabases including CNKI, VIP, WanFang Data, Web of Science, PubMed, EMbase and The Cochrane Library (Issue 8, 2016) were searched from inception to August, 2016 to collect randomized controlled trials (RCTs) of Tai Chi for heart failure patients. Two reviewers independently screened literature, extracted data, and assessed the risk of bias of included studies. Then meta-analysis was performed using RevMan 5.3 software.ResultsA total of 10 RCTs involving 689 patients were included. The results of meta-analysis showed that, compared with the control group, the heart failure patients in Tai Chi group had better score of minnesotaliving with heart failure questionnaire (MLHFQ) (MD=–9.37, 95%CI –13.09 to –5.65, P<0.000 01), longer six minute walk test (6MWT) (MD=40.37, 95%CI 9.48 to 71.27, P=0.01), higher left ventricular ejectionfractions (LVEF) (MD=7.89, 95%CI 3.01 to 12.77, P=0.002) and lower level of BNP (brain natriuretic peptide) (MD=–10.75, 95%CI –13.20 to –8.30, P<0.000 01); however, as to the maximal oxygen consumption (VO2max) (MD=0.29, 95%CI –1.223 to 1.81, P=0.71), systolic pressure (SBP) (MD=–2.81, 95%CI –8.52 to 2.90, P=0.33) and diastolic pressure (DBP) (MD=0.37, 95%CI –3.73 to 4.48, P=0.86), there were no significant differences between both groups.ConclusionThe current evidence shows that Tai Chi is feasible for patients with heart failure as it has positive effects on life quality, physiological functions. Due to the limited quality and quantity of included studies, the above conclusion should be validated by more high quality studies.
ObjectiveThe current medical questionnaire resources are mainly processed and organized at the document level, which hampers user access and reuse at the questionnaire item level. This study aims to propose a multi-class classification of items in medical questionnaires in low-resource scenarios, and to support fine-grained organization and provision of medical questionnaires resources. MethodsWe introduced a novel, BERT-based, prompt learning approach for multi-class classification of items in medical questionnaires. First, we curated a small corpus of lung cancer medical assessment items by collecting relevant clinical assessment questionnaires, extracting function and domain classifications, and manually annotating the items with "function-domain" combination labels. We then employed prompt learning by feeding the customized template into BERT. The masked positions were predicted and filled, followed by mapping the populated text to labels. This process enables the multi-class classification of item texts in medical questionnaires. ResultsThe constructed corpus comprised 347 clinical assessment items for lung cancer, across nine "function-domain" labels. The experimental results indicated that the proposed method achieved an average accuracy of 93% on our self-constructed dataset, outperforming the runner-up GAN-BERT by approximately 6%. ConclusionThe proposed method can maintain robust performance while minimizing the cost of building medical questionnaire item corpora, illustrating its promotion value of research and practice in medical questionnaire classification.
Objective To investigate the free influenza vaccination of health care workers in major departments and explore the possible influencing factors of influenza vaccination of staff. Methods In November 2021, a questionnaire survey was conducted among health care workers who received free influenza vaccination in 19 major departments of West China Hospital of Sichuan University, and the un-vaccinated workers’ information was obtained from the registration system of staff information. Multiple logistic regression model was used to analyze the possible influencing factors of free influenza vaccination. Results The coverage rate of centralized free influenza vaccination of staff in major departments was 32.7% (1101/3369). Multiple logistic regression analysis showed that workers who were female [odds ratio (OR)=1.853, 95% confidence interval (CI) (1.481, 2.318), P<0.001], with an educational background of high school or below [OR=4.304, 95%CI (2.484, 7.455), P<0.001], engaged in nursing work [OR=2.341, 95%CI (1.701, 3.221), P<0.001], and with 11 or more years of working experience [OR=2.410, 95%CI (1.657, 3.505), P<0.001] were more likely to inject influenza vaccine, and workers who had a bachelor’s degree were less likely to inject influenza vaccine. Conclusions The rate of free influenza vaccination among medical staff is low. In order to mobilize the enthusiasm of influenza vaccination among medical staff, it is necessary to analyze the characteristics of the population and take targeted measures to improve the level of vaccination among medical staff.
Automatic classification of medical questions is of great significance in improving the quality and efficiency of online medical services, and belongs to the task of intent recognition. Joint entity recognition and intent recognition perform better than single task models. Currently, most publicly available medical text intent recognition datasets lack entity annotation, and manual annotation of these entities requires a lot of time and manpower. To solve this problem, this paper proposes a medical text classification model, bidirectional encoder representation based on transformer-recurrent convolutional neural network-entity-label-semantics (BRELS), which integrates medical entity label semantics. This model firstly utilizes an adaptive fusion mechanism to absorb prior knowledge of medical entity labels, achieving local feature enhancement. Then in global feature extraction, a lightweight recurrent convolutional neural network (LRCNN) is used to suppress parameter growth while preserving the original semantics of the text. The ablation and comparison experiments are conducted on three public medical text intent recognition datasets to validate the performance of the model. The results show that F1 score reaches 87.34%, 81.71%, and 77.74% on each dataset, respectively. The results show that the BRELS model can effectively identify and understand medical terminology, thereby effectively identifying users’ intentions, which can improve the quality and efficiency of online medical services.
Objective To compare the clinical characteristics of chronic cough, and to establish the Modified Cough Assessment Test and the simple decision tree to improve the efficacy of etiologic diagnosis. Methods Patients with chronic cough consulted in Tongji Hospital between October 2021 and August 2023 were enrolled in our study. The patients with identified single cause were divided into 3 groups accordingly: corticosteroid-responsive cough (CRC), upper airway cough syndrome (UACS) and gastroesophageal reflux-related cough (GERC). And the characteristics of chronic cough in different causes were assessed and compared by cough questionnaires. Independent predictors of various causes were identified by multivariate logistic regression analysis and used to establish the Modified Cough Assessment Test (MCET) and to construct the simple decision tree. Results A total of 358 patients were enrolled, including 201 with CRC (56.1%), 125 with UACS (34.9%) and 32 with GERC (8.94%). "Cough with wheezing or chest tightness" (OR=3.222, 95%CI 2.144 - 4.843, P<0.001), "Cough with daytime heaviness and nighttime lightness" (OR=1.755, 95%CI 1.264 - 2.435, P<0.001), and "Cough with acid reflux, heartburn or indigestion" (OR=15.580, 95%CI 5.894 - 41.184, P<0.001) were independent factors for each group, respectively. The area under ROC curve for classification of CRC, UACS and GERC were 0.871, 0.840 and 0.988 for MCET, which were better than those of Leicester Cough Questionnaire (LCQ) (0.792, 0.766 and 0.913) and Cough Evaluation Test (CET) (0.649, 0.691 and 0.580). The accuracy of the simple decision tree for the differential diagnosis of chronic cough was 77.4%. Conclusion The simple decision tree based on the Modified Cough Evaluation Test is a simple and effective method of etiologic diagnosis of chronic cough, which can be used as a tool to improve the efficacy of clinical diagnosis in outpatient settings.
Objective To understand the current status of the preferences and opinions on the investigator-initiated trails (IIT) of the neurosurgeons participating in INTERACT3 in China, as well as the design preference for IIT projects, and to provide a basis for the design and organization of multi-center clinical studies in the future. Methods Neurosurgeons with different seniority and professional titles from 89 domestic research institutions participating in the INTERACT3 project were collected from September to October 2023. The questionnaires were collected by questionnaire star. Results A total of 56 valid questionnaires were collected from 29 units. Among the 56 respondents, 52 neurosurgeons (92.86%) were from teaching hospitals and 45 (80.36%) were from grade A tertiary hospitals. 30 neurosurgeons (53.57%) had experience in conducting various clinical studies, and 55 neurosurgeons (98.21%) had experience in participating in various clinical studies. The main purposes of presiding over or participating in clinical research focused on “accumulating relevant experience and preparing for future projects” and “standardizing clinical diagnosis and treatment”, which were 89.29% and 83.93%. Respectively, regarding the way the case report form completing, respondents preferred to use electronic data collection systems (83.93%). Conclusions The purpose of the neurosurgeons interviewed to host or participate in clinical research is mainly to assist clinical and scientific research. Economic reasons have little impact on whether to participate in clinical research. The rationality and ease of operation of the trail design are the keys to attracting respondents to participate in clinical researches, and the level of remuneration has little impact on the decision-making of the respondents. The safety of clinical studies and the difficulty of enrolling subjects are the key factors that hinder respondents’ participation in clinical studies.
Medical visual question answering (MVQA) plays a crucial role in the fields of computer-aided diagnosis and telemedicine. Due to the limited size and uneven annotation quality of the MVQA datasets, most existing methods rely on additional datasets for pre-training and use discriminant formulas to predict answers from a predefined set of labels. This approach makes the model prone to overfitting in low resource domains. To cope with the above problems, we propose an image-aware generative MVQA method based on image caption prompts. Firstly, we combine a dual visual feature extractor with a progressive bilinear attention interaction module to extract multi-level image features. Secondly, we propose an image caption prompt method to guide the model to better understand the image information. Finally, the image-aware generative model is used to generate answers. Experimental results show that our proposed method outperforms existing models on the MVQA task, realizing efficient visual feature extraction, as well as flexible and accurate answer outputs with small computational costs in low-resource domains. It is of great significance for achieving personalized precision medicine, reducing medical burden, and improving medical diagnosis efficiency.
Objective To investigate the opinions of operating room nurse (ORN) on enhanced recovery after surgery (ERAS). Methods A questionnaire survey was performed among 215 ORNs in West China Hospital. There were 10 males and 205 females at age of 33.4±8.84 years. Results A total of 154 ORNs (71.6%) thought that we already had very good ERAS theory but we still needed more practice. Thirty-four ORNs(15.8%) thought that the application of ERAS was poor in our clinic comparing to other countries.A percentage of 84.2% (181/215) ORNs thought the criteria to judge whether the ERAS succeed or not should be average days of hospitalization, patients' feeling, and experience and social satisfactions. Besides, 78.1% (168/215) ORNs selected team building as the key point of ERAS success. There were 91.2% (196/215) ORNs who believed expert consensus and ERAS guide should be worked out and propagandized through academic forum or conference in order to popularize the ERAS. Conclusion The theory of ERAS has already been accepted by almost all the clinicians and team building is the best way to make ERAS work well.