Health Information Science

Knowledge graphs can support different types of services and are a valuable source. Automatic methods have been widely used in many domains to construct the knowledge graphs. However, it is more complex and difficult in the medical domain. There are three reasons: (1) the complex and obscure nature of medical concepts and relations, (2) inconsistent standards and (3) heterogeneous multi-source medical data with low quality like EMRs (Electronic Medical Records). Therefore, the quality of knowledge requires a lot of manual efforts from experts in the process. In this paper, we introduce an overall framework called DocKG that provides insights on where and when to import manual efforts in the process to construct a health knowledge graph. In DocKG, four tools are provided to facilitate the doctors’ contribution, i.e. matching synonym, discovering and editing new concepts, annotating concepts and relations, together with establishing rule base. The application for cardiovascular diseases demonstrates that DocKG could improve the accuracy and efficiency of medical knowledge graph construction.

At home genetic testing is currently accepted by a lot of people in many countries, and statistical data show that, currently, at home genetic testing has more than 26, 000, 000 customers all over the world. Generally, the business process for at home genetic testing is: after a user’s order, a saliva collection kit will be sent to the user, the user should split saliva to a specific saliva collection tube and send the kit back to laboratory, then the laboratory will extract the DNA from the saliva, and sequence the DNA using next generation sequencing equipment or micro-array platform, the generated DNA sequencing data will be analyzed and genetically interpreted, finally, a genetic report will be sent to the user. To handle millions of samples in a year requires a scalable, robust, parallel, and easy to use business process management system to satisfy the external customer service and internal sample track and management requirement. In this paper, we first describe the detail business process of at home genetic testing, then based on our best practice, using spring cloud, spring boot, and microservices, we give the design and implementation of a business process management platform to support at home genetic testing business. The platform is flexible that supports both the business to business service as well as the business to customer service.

With the national emphasis on traditional Chinese medicine treatments and the development of the modern Internet, people are increasingly showing a strong interest in traditional Chinese medicine, leading to the transformation of traditional Chinese medicine enterprises. The optimisation and innovation of the traditional Chinese medicine pharmacy service has become a hot topic. Therefore, this study combines the advantages of traditional Chinese medicine with Internet technology to build a smart health-oriented traditional Chinese medicine pharmacy intelligent service platform. It integrates hospitals, pharmacies, drug decoction centres, distribution centres and other resources, and forms a traditional Chinese medicine decoction, distribution and traceability system. The system realises the informatisation, automation and standardisation of traditional Chinese medicine pharmacy services. In this study, the platform is implemented using the Internet of Things and the Internet in Nanjing Pharmaceutical Co., Ltd. to provide patients with standard modern drug decoction and distribution services, and to monitor and manage the decoction, distribution and traceability processes.

Medical data plays an important role in government regulation of resources, scientific research and precise treatment of medical staffs. Due to the different data management systems used by each hospital, it is difficult to exchange data among them, resulting in a waste of medical resources. In this paper, a medical data management model based on the blockchain is proposed, which takes advantage of the characteristics of the blockchain, such as decentralisation, tamper-proofing and realizability. A data-sharing reward mechanism was designed to maximise the benefits of both a medical data producer (MDP) and a medical data Miner (MDM) in the process of data sharing, while reducing the risk of leakage of a patient’s private information. Finally, a reward mechanism was analysed through experiments, which proved the validity and reliability of the medical data management model based on blockchain.

Correct code assignment of Medicare Severity Diagnosis-Related Group (MS-DRG) is critical for healthcare. However, there is a gap currently on automatically identifying all potentially miscoded cases and prioritizing manual reviews over these cases. This paper reports a new process using a data-driven machine learning approach to flag potentially misclassified cases for manual review. We investigated using a stack of regularized logistic/softmax regression, random forest, and support vector machine to suggest potential cases for manual review by care providers, provided details addressing the data imbalance, and explored using features from various source including diagnosis and procedure codes, length of stay and the access log data from the electronic health record system. This potentially improves the efficiency of the coding review by care providers, providing another line of defense against miscoding to enhance coding compliance, and reduce the negative effects of upcoding and downcoding. We tested the new method with four common pediatric conditions and demonstrated its feasibility.

There are various of skin pigmented lesions with high risk. Melanoma is one of the most dangerous forms of skin cancer. It is one of the important research directions of medical artificial intelligence to carry out classification research of skin pigmented lesions based on deep learning. It can assist doctors to make clinical diagnosis and make patients receive treatment as soon as possible to improve survival rate. Aiming at the similar and imbalanced dermoscopic image data of pigmented lesions, this paper proposes a deep residual network improved by Squeeze-and-Excitation module, and dynamic update class-weight, in batches, with model ensemble adjustment strategies to change the attention of imbalanced data. The results show that the above method can increase the average precision by 9.1%, the average recall by 15.3%, and the average F1-score by 12.2%, compared with the multi-class classification using the deep residual network. Thus, the above method is a better classification model and weight adjustment strategy.

With the development of high-throughput sequencing technology, it brings a large volume of data of transcriptome. Long non-protein-coding RNAs (lncRNAs) identification is pervasive in transcriptome studies in their important roles in biological process. This paper proposed a computational method for identifying lncRNAs based on machine learning. The method first selects feature using k-mer for traversing the transcript sequence to obtain a large class of features, integrated GC content and sequence length. Then it uses variance test to select three kinds of features by grid searching and reduce the data dimension and support vector machine pressure to establish a recognition model, the final model has a certain stability and robustness. The method obtain 95.7% accuracy, 0.99 AUC for test dataset. Therefore, it could be promising for identifying lncRNA.

Quality and safety evaluation of Chinese herbal medicine has always been a difficult and key issue in the research and application of Chinese herbal medicine, which restricts the modernization and internationalization of Chinese herbal medicine. In this paper, we present a quality and safety credit evaluation index of Chinese herbal medicines production, and the numerical value was used to reflect the quality and safety credit status of Chinese herbal medicine enterprises. TOPSIS method is introduced into the field of quality and safety evaluation of CHM. Considering the multi-level characteristics of the evaluation index system, the weight of the index is calculated by analytic hierarchy process (AHP), and the ranking of quality and safety credit of CHM in the region is obtained. The quality and safety credit evaluation index of Chinese herbal medicine was constructed. It is an important basic application platform for establishing a sound quality supervision system for Chinese herbal medicine.

Predicting ICU mortality and finding key risk factors make sense for both doctors and patients. Although there has been a number of research pertaining to ICU mortality prediction systems and algorithms, plenty of room still exists for improvement in practical prediction results and identification of important risk factors. In this study, we use C5 decision tree model to predict mortality of ICU patients and identify key risk factors. Totally 4367 records of ICU patients from a local grade-A tertiary hospital were selected for motality prediction, including 244 dead records with demographic information and physiological parameters. In order to solve the problem of inconsistent data sampling frequency, we extracted 96 statistical indicators based on the original records, such as the kurtosis value of red blood cells (HXB_kurt), the skewness coefficient of red blood cells (HXB_skew). Totally 41 indicators as the final input of the prediction model were extracted through feature extraction method. The experimental results show that C5 decision tree model outperform C&RT, CHDID, KNN, Logistic, SVM and Random Forest in five different performance indicators. Moreover, worst-case status and state of changes in respiratory, body temperature, care level, diastolic blood pressure and age were found to be the key risk factors.

Stroke patients are often associated with lower levels of heart rate variability, suggesting that autonomic dysfunction is very common in stroke patients. Recent studies have shown that heart rate variability (HRV) is an early predictor of prognosis in patients with acute stroke, but the relationship between HRV and functional status in chronic rehabilitation patients is not clear. The purpose of this study was to investigate the clinical value of heart rate variability parameters in predicting motor function assessment in convalescent stroke patients. Methods: Sixty-four patients with strokes admitted to Beijing Bo’ai Hospital from October 2015 to October 2016 were enrolled. Dynamic electrocardiogram was used to continuously record the data of 24-h monitoring and analyze the heart rate variability, including time domain parameters [standard deviation of all NN intervals (SDNN, where NN intervals refer to the RR intervals of sinus beats), standard deviation of the 5-mins average NN intervals (SDANN), percentage of successive NN intervals greater than 50 ms (PNN50), root mean square of differences between adjacent RR intervals (RMSSD)], frequency domain parameters [high frequency component (HF), low frequency component (LF), very low frequency component (VLF), ratio of low frequency to high frequency component (LF/HF)] and heart rate variability triangular index. And by using the Barthel Index for Activities of Daily Living (ADLs) and the Fugl-Meyer Motor Assessment (FMA) simultaneously, the patients’ functional status was assessed. Results: The correlation analysis with related factors controlled showed that the HRV parameters were significantly correlated with the recovery of motor function [time domain indicators RR triangular index (r = 0.252; P = 0.05) and frequency domain indicator VLF (r = 0.302; P = 0.018)] and there was no relationship with HRV parameter and the improvement in activities of daily living of stroke patients in the chronic rehabilitation period. The monitoring of HRV-related parameters of stroke patients in their chronic rehabilitation period has a certain correlation with their motor function outcomes and daily living ability. Non-invasive monitoring of HRV may be an alternative method to judge the prognosis of stroke. In the future, further research is needed to verify the relevance of HRV to clinical outcomes.

Personalized knowledge recommendation is an effective measure to provide individual information services in the field of brain science. It is essential that a complete understanding of authors’ interests and accurate recommendation are carried out to achieve this goal. In this paper, a collaborative recommendation method based on co-authorship is proposed to make. In our approach, analysis of collaborators’ interests and the calculation of collaborative value are used for recommendations. Finally, the experiments using real documents associated with brain science are given and provide supports for collaborative document recommendation in the field of brain science.

Sleep is vital to the health of the human being. Accurate sleep stage scoring is an important prerequisite for diagnosing sleep health problems. The sleep electroencephalogram (EEG) waveform shows diverse variations under the physical conditions of subjects. To help neurologists better analyze sleep data in a fairly short time, we decide to develop a novel method to extract features from EEG signals. Traditional sleep stage scoring methods typically extract the one-dimensional (1D) features of single-channel EEG signals. This paper is the very first time to represent the single-channel EEG signals as two-dimensional (2D) frequency domain representation. Comparing with similar currently existing methods, a deep learning model trained by frequency domain representation can extract frequency morphological features over EEG signal patterns. We conduct experiments on the real EEG signals dataset, which is obtained from PhysioBank Community. The experiment results show that our method significantly improved the performance of the classifier.

Physiological datasets such as Electroencephalography (EEG) data offer an insight into some of the less well understood aspects of human physiology. This paper investigates simple methods to develop models of high level behavior from low level electrode readings. These methods include using neuron activity based pruning and large time slices of the data. Both approaches lead to solutions whose performance and transparency are superior to existing methods.

Premature ventricular contraction (PVC) is one of the most common arrhythmia diseases. The traditional diagnosis of PVC by visual inspection of PVC beats in electrocardiogram (ECG) is a time-consuming process. Hence, there has been an increasing interest in the study of automatic identification of PVC using ECGs in recent years. In this paper, a novel automatic PVC identification method is proposed. We first design a new approach to detect peak points of QRS complex. Then nine features are extracted from ECG according to the detected peak points, which are used to measure the morphological characteristics of PVC beats from different points of view. Finally, the key features are selected and fed into back propagation neural network (BPNN) to differentiate PVC ECGs from normal ECGs. Simulation results on the China Physiological Signal Challenge 2018 (CPSC2018) Database verify the feasibility and efficiency of the proposed method. The average accuracy attains 97.46%, as well as the average false detection rate and omission ratio are 3.41% and 1.37% respectively, which implies that the proposed method does a good job in identifying PVC automatically.

Automated detection of first-degree atrioventricular block (I-AVB) using electrocardiogram (ECG) has been paid more and more attraction since it is very helpful for the timely and efficient diagnosis and treatment of AVB-related heart diseases. In this paper, a novel automated I-AVB detection method FPR\(_{dur}\)-SVM is proposed, where the I-AVB feature FPR\(_{dur}\) is extracted from ECGs and then fed into the support vector machine (SVM) to differentiating I-AVB ECG from normal ECG. Performances of the proposed method FPR\(_{dur}\)-SVM are verified on the China Physiological Signal Challenge 2018 Database (CPSC2018). Simulation results show that the accuracy, sensitivity and specificity are reached 98.5%, 98.7% and 98.3%.

The automatic detection of bundle branch block (BBB) using electrocardiogram (ECG) has been attracting more and more attention, which is recognized to be helpful in the diagnosis and treatment of BBB-related heart diseases. In this paper, a novel automatic BBB detection method is developed. We first propose a new R peak detection algorithm which is able to detect both single R peak and multiple R peaks in one ECG beat. Then the number of R peaks and the length of RR interval are calculated to be the extracted features. Finally, linear classification is implemented to differentiate BBB ECG from normal ECG. Simulation results on CPSC2018 Database show that the average accuracy, sensitivity and specificity attain 96.45%, 95.81% and 96.80% respectively, demonstrating that the presented method of automatic BBB detection works well in distinguishing the normal ECG signals and the BBB ECG signals. This research thus provides insights for the automatic detection of BBB.

With the application of information technology, the scale of various digital image resources is increasing. How to make good use of these digital resources has become a hotspot of hospital management. Because of the use of artificial intelligence technology, the results of automatic recognition and annotation of hospital image resources can be displayed in two-dimensional form. But two-dimensional display, without stereoscopic visual effect, doctors are not easy to observe. In addition, in the real medical environment, it is necessary to combine the virtual information with the real scene, and add some additional information in the real environment to help doctors obtain more useful information. Augmented reality (AR) is a new visualization technology, which can meet the actual business needs mentioned above. In this paper, we focus on brain image, and use PACS as the channel of image acquisition and communication to study the feature extraction of brain image and three-dimensional reconstruction of brain image diseases. Based on the image file, the features of brain image are extracted, and the function module of augmented reality is designed. Unity is used to realize the three-dimensional reconstruction and visualization of brain image features through programming.

Various imaging modalities (CT, MRI, PET, etc.) encompass abundant information which is different and complementary to each other. It is reasonable to combine images from multiple modalities to make a more accurate assessment. Multimodal medical imaging has shown notable achievements in improving clinical accuracy. Deep learning has achieved great success in image recognition, and also shown huge potential for multimodal medical imaging analysis. This paper gives a review of deep learning in multimodal medical imaging analysis, aiming to provide a starting point for people interested in this field, and highlight gaps and challenges of this topic. Based on the introduction of basic ideas of deep learning and medical imaging, the state-of-the-art multimodal medical image analysis is given, with emphasis on the fusion technique and feature extraction deep models. Multimodal medical image applications, especially cross-modality related, are also summarized.

Diabetic retinopathy (DR) is one of the major diseases causing blindness, and microaneurysms in the fundus are the first reliable lesions in its early stage. This paper proposes an object detection method for microaneurysms based on R-CNN, which consists of five steps: image preprocessing, candidate region generation, feature extraction, classification and non-maximal suppression. First, a fundus image preprocessing method and a candidate region generation algorithm for microaneurysms are proposed. Then, the VGG16 network is trained using the transferred fine-tuning model to extract features from candidate samples. Finally, real aneurysms are selected from candidate regions by a classifier. The experimental results in the internationally published database e-ophtha show that the proposed method outperforms other known related methods on the FROC indicator.

Alzheimer’s disease has become a worldwide problem. Cognitive training can effectively slow the progression of Alzheimer’s disease and improve the quality of life of patients with Alzheimer’s disease. Spatial orientation is an important aspect of cognitive training. Due to the immersive and interactive features of VR (Virtual Reality) technology, VR technology has been gradually applied to cognitive training systems. This paper designs and implements a VR technology-based intelligent cognitive rehabilitation system for Alzheimer’s disease for assessing and training the spatial orientation of patients with Alzheimer’s disease. First, pre-assess the physiological status and operational ability of patients with Alzheimer’s disease. Then, build a realistic environment, guide through endogenous orientation and auditory orientation, and complete orientation training with different difficulty levels. Finally, the system provides relevant data for correlation analysis, combined with computer technology, to help doctors understand the patient’s condition to complete further treatment.

With the development of social economy and the aggravation of environmental pressure, autism has been gradually recognized and its incidence has become higher and higher. At present, the research on autism at home and abroad is still in its infancy, and the research methods and tools are relatively few, especially the research on the acquisition and display of autism-related knowledge is relatively scarce. Doctors and various rehabilitation institutions as well as the vast number of autistic patients have an urgent need to acquire knowledge about treatment and rehabilitation of autism. In the information age, the Internet provides people with convenient access to information and knowledge. The purpose of this study is to use Internet technology to collect information and knowledge about autism. The autism database is designed and the autism information system based on knowledge map is constructed. Autism information system can provide doctors and patients with efficient access to autism information and knowledge, help them to acquire the required knowledge at any time and anywhere, and promote the early diagnosis and treatment of autism.

Gut microbiota that lives in the human gastrointestinal tract impacts on the mental illness through the neurotransmitter-mediated pathway. It’s well known that the imbalance of neurotransmitter leads to mental problems. The association between gut microbiota and neurotransmitter needs to be explored in depth. In this paper, we aim at identifying the quality evidence of neurotransmitter modulation by gut microbiota. We use evidence-based medical analysis to characterize the relevant articles to five levels in terms of the strength and reliability of evidence. Thirty-four articles are identified to evaluate their evidence. Gut microbiota not only produces neurotransmitters directly but also modulates neurotransmitters level via metabolism pathways. Also, the growth of some gut microbiota can be counter-regulated by neurotransmitters. This paper provides a comprehensive picture of the association between gut microbiota and neurotransmitter, which give researchers an insight into neurotransmitter modulation by gut microbiota.

Most people with Autism Spectrum Disorder (ASD) experience atypical sensory modality and need help to self-regulate their sensory responses. Results of a pilot study are presented here where temperature, noise types and noise levels are used as independent variables. Attention-based tests (ABTs), Electrodermal Activity (EDA) and Electroencephalography (EEG) sensors are used as dependent variables to quantify the effects of temperature and noise. Based on the outcome of the analyses, it is feasible to use off-the-shelf sensors to recognize physiological changes, indicating a possibility to develop sensory management recommendation interventions to support people with ASD.

The personality description has been considered to be an important part of the electronic medical records (EMRs) for depression in-patients. However, the personality description of a patient is usually provided by a family member of the patient when he or she is admitted to the hospital. Because of non-professional background of those family members, personality descriptions in EMRS have various problems such as non-standard description, confusion between personality and behavior, and others. In this paper, we propose an approach to dealing with non-standard description of personality for depression patients by introducing a semantic relevance measure of personality. Furthermore, we make a mapping of those personality description items into the personality items in the well-known personality model of the Sixteen Personality Factor Questionnaire (16PF). We find some interesting observations on the connection between existing personality description in EMRs and the personality items in the 16PF model, and suggest possible improvement of the personality description in EMRs for depression patients.

With the development of molecular biology techniques, genomics is broadly introduced to expound the individual difference and molecular mechanism. Effects of genetic diversity on sports performance have been more and more found by scientists. For instance, studies on association of gene polymorphisms and training response mainly intend to discover the effects of different genotypes on the effectiveness of aerobic exercise training to increase aerobic physical fitness. Genes related to training response primarily include PPARD, PPARGC1A, ACTN3, ACE, HBB, TFAM, NFR2, AR and etc. Gene polymorphisms of PPARD and PPARGC1A are shown to be associated with post-training individual anaerobic threshold. ACE is one of the earliest and most studied gene in genes related to endurance performance that ACE is a carboxypeptidase, the key enzyme in Renin-angiotensin system, it was found in many tissues in the body including skeletal muscle, associated to functions of degrading bradykinin and transferring angiotensin I to angiotensin II. Therefore, genetic testing can help people to know the precise information on how body uniquely responds to exercise. Based on the deeper understanding of a person’s genetics and physiology information, in this paper, we provide a platform with sports and nutrition actions that are tailored specifically for different people to optimize their sports performance and effects as well as improve their health.

Traditional Chinese Medicine (TCM) for weight loss is a personalized medical treatment which is widely used and effective at present. The evaluation of TCM intervention technology for different obesity grades can provide a basis for optimizing treatment schedule. This study retrieved and selected 640 literatures about the treatment of simple obesity with Body Mass Index (BMI) information from 1980 to 2016. Through literature research, expert consultation, mathematical statistics, and machine learning, from the perspective of single intervention technology and intervention technology combination, the evaluation index system of single intervention technology based on literature and the evaluation method of intervention technology combination were established. Empirical study takes overweight patients (BMI 25–26) in obesity grade as an example. Single intervention technology evaluation found that the comprehensive top-ranking TCM intervention technology is “acupuncture, catgut embedding, electro-acupuncture and ear acupoint”. Intervention technology combination evaluation found that the most commonly used TCM intervention techniques were “electro-acupuncture and acupuncture”, “ear acupoint and acupuncture”, “ear acupoint and electro-acupuncture”, “ear acupoint, electro-acupuncture and ear acupuncture”, and “acupuncture and catgut embedding”. The conclusion was in line with clinical practice, thus confirming the rationality of the method. The establishment of this method provides a powerful reference for the clinical selection of appropriate TCM intervention technologies for obesity.

The bibliometrics method is used to analyze the literature related to the treatment of simple obesity with Traditional Chinese Medicine (TCM), to understand the current situation and the development tendency of obesity intervention technology. From a variety of literature databases, the papers on the treatment of simple obesity by TCM from 1980 to 2016 are selected, and 640 papers are finally included. The papers are analyzed from six dimensions: time distribution, type of publication, distribution of authors and units, fund projects, titles and key words. According to the study, since 2001, the number of articles published has been steadily increasing; Beijing, Guangzhou and other first-tier cities have paid more attention to obesity, and some core research teams have formed; high-frequency keywords of intervention technology mainly include “acupuncture”, “electro-acupuncture”, “acupoint catgut embedding”, “ear acupoint” and “cupping”. The results show that the research of TCM intervention technology for obesity is in the rapid development stage with broad application prospects. This paper can provide reference for clinical practice of TCM treatment of obesity.

The purpose of health consumer–targeted digital information channels is to enhance the knowledge levels of consumers and patients, and thus, increase the intensity of care. Medical information content as a knowledge entity forms the essence, the substance of these systems, and all the other development work should be constructed around the content. Despite broad efforts, recent evidence shows that refinement is needed in the design of information systems in this core area, as well as in its assessment. In addition to tools for evaluating the quality of information, there is a continuing need for design strategies that produce better information quality. Values and their constant evaluation also have an important role in producing high-quality information.