Decision support system for the differentiation of schizophrenia and mood disorders using multiple deep learning models on wearable devices data.

In the modern world, with so much inherent stress, mental health disorders (MHDs) are becoming more common in every country around the globe, causing a significant burden on society and patients' families. MHDs come in many forms with various severities of symptoms and differing periods of suffering, and as a result it is difficult to differentiate between them and simple to confuse them with each other. Therefore, we propose a support system that employs deep learning (DL) with wearable device data to provide physicians with an objective reference resource by which to make differential diagnoses and plan treatment. We conducted experiments on open datasets containing activity motion signal data from wearable devices to identify schizophrenia and mood disorders (bipolar and unipolar), the datasets being named Psykose and Depresjon. The results showed that, in both workflow approaches, the proposed framework performed well in comparison with the traditional machine learning (ML) and DL methods. We concluded that applying DL models using activity motion signal data from wearable devices represents a prospective objective support system for MHD differentiation with a good performance.

Decline in hospitalization for childhood asthma in different air pollution regions in Taiwan, 2001-2012.

This study aimed to investigate the trends in childhood asthma hospitalization in regions with differing levels of air pollution in Taiwan, 2001-2012. Joinpoint regression was used to identify significant trend changes. The hospitalization rate varied according to gender, geographic region, and age. The incidence of childhood asthma hospitalization decreased from 127.99 to 76.67 (/100,000 population), with an average annual percentage change of around -4.1%; in the Yilan region, the average air pollution concentrations were 19.92 µg/m3, 39.47 µg/m3, 25.99 ppb, 2.19 ppb, and 11.23 ppb for PM2.5, PM10, O3, SO2, and NO2, respectively, which were lower than Taiwan's average values; however, the childhood asthma hospitalization rate was the highest (179.75/100,000 population). The national trend in childhood asthma hospitalization exhibited a significant decrease. The effects of air pollution on childhood asthma were greater in the higher-level air pollution regions, while less association was observed in the lower-level air pollution regions.

Deep learning approaches for sleep disorder prediction in an asthma cohort.

OBJECTIVE: Sleep is a natural activity of humans that affects physical and mental health; therefore, sleep disturbance may lead to fatigue and lower productivity. This study examined 1 million samples included in the Taiwan National Health Insurance Research Database (NHIRD) in order to predict sleep disorder in an asthma cohort from 2002-2010. METHODS: The disease histories of the asthma patients were transferred to sequences and matrices for the prediction of sleep disorder by applying machine learning (ML) algorithms, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Random Forest (RF), and deep learning (DL) models, including Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and Convolution Neural Network (CNN). RESULTS: Among 14,818 new asthma subjects in 2002, there were 4469 sleep disorder subjects from 2002 to 2010. The KNN, SVM, and RF algorithms were demonstrated to be successful sleep disorder prediction models, with accuracies of 0.798, 0.793, and 0.813, respectively (AUC: 0.737, 0.690, and 0.719, respectively). The results of the DL models showed the accuracies of the RNN, LSTM, GRU, and CNN to be 0.744, 0.815, 0.782, and 0.951, respectively (AUC: 0.658, 0.750, 0.732, and 0.934, respectively). CONCLUSIONS: The results showed that the CNN model had the best performance for sleep disorder prediction in the asthma cohort.

TNFPred: identifying tumor necrosis factors using hybrid features based on word embeddings.

BACKGROUND: Cytokines are a class of small proteins that act as chemical messengers and play a significant role in essential cellular processes including immunity regulation, hematopoiesis, and inflammation. As one important family of cytokines, tumor necrosis factors have association with the regulation of a various biological processes such as proliferation and differentiation of cells, apoptosis, lipid metabolism, and coagulation. The implication of these cytokines can also be seen in various diseases such as insulin resistance, autoimmune diseases, and cancer. Considering the interdependence between this kind of cytokine and others, classifying tumor necrosis factors from other cytokines is a challenge for biological scientists. METHODS: In this research, we employed a word embedding technique to create hybrid features which was proved to efficiently identify tumor necrosis factors given cytokine sequences. We segmented each protein sequence into protein words and created corresponding word embedding for each word. Then, word embedding-based vector for each sequence was created and input into machine learning classification models. When extracting feature sets, we not only diversified segmentation sizes of protein sequence but also conducted different combinations among split grams to find the best features which generated the optimal prediction. Furthermore, our methodology follows a well-defined procedure to build a reliable classification tool. RESULTS: With our proposed hybrid features, prediction models obtain more promising performance compared to seven prominent sequenced-based feature kinds. Results from 10 independent runs on the surveyed dataset show that on an average, our optimal models obtain an area under the curve of 0.984 and 0.998 on 5-fold cross-validation and independent test, respectively. CONCLUSIONS: These results show that biologists can use our model to identify tumor necrosis factors from other cytokines efficiently. Moreover, this study proves that natural language processing techniques can be applied reasonably to help biologists solve bioinformatics problems efficiently.

Liver cancer prediction in a viral hepatitis cohort: A deep learning approach.

Viral hepatitis is the primary cause of liver diseases, among which liver cancer is the leading cause of death from cancer. However, this cancer is often diagnosed in the later stages, which makes treatment difficult or even impossible. This study applied deep learning (DL) models for the early prediction of liver cancer in a hepatitis cohort. In this study, we surveyed 1 million random samples from the National Health Insurance Research Database (NHIRD) to analyze viral hepatitis patients from 2002 to 2010. Then, we used DL models to predict liver cancer cases based on the history of diseases of the hepatitis cohort. Our results revealed the annual prevalence of hepatitis in Taiwan increased from 2002 to 2010, with an average annual percentage change (AAPC) of 5.8% (95% CI: 4.2-7.4). However, young people (aged 16-30 years) exhibited a decreasing trend, with an AAPC of -5.6 (95% CI: -8.1 to -2.9). The results of applying DL models showed that the convolution neural network (CNN) model yielded the best performance in terms of predicting liver cancer cases, with an accuracy of 0.980 (AUC: 0.886). In conclusion, this study showed an increasing trend in the annual prevalence of hepatitis, but a decreasing trend in young people from 2002 to 2010 in Taiwan. The CNN model may be applied to predict liver cancer in a hepatitis cohort with high accuracy.

Using Language Representation Learning Approach to Efficiently Identify Protein Complex Categories in Electron Transport Chain.

We herein proposed a novel approach based on the language representation learning method to categorize electron complex proteins into 5 types. The idea is stemmed from the the shared characteristics of human language and protein sequence language, thus advanced natural language processing techniques were used for extracting useful features. Specifically, we employed transfer learning and word embedding techniques to analyze electron complex sequences and create efficient feature sets before using a support vector machine algorithm to classify them. During the 5-fold cross-validation processes, seven types of sequence-based features were analyzed to find the optimal features. On an average, our final classification models achieved the accuracy, specificity, sensitivity, and MCC of 96 %, 96.1 %, 95.3 %, and 0.86, respectively on cross-validation data. For the independent test data, those corresponding performance scores are 95.3 %, 92.6 %, 94 %, and 0.87. We concluded that using feature extracted using these representation learning methods, the prediction performance of simple machine learning algorithm is on par with existing deep neural network method on the task of categorizing electron complexes while enjoying a much faster way for feature generation. Furthermore, the results also showed that the combination of features learned from the representation learning methods and sequence motif counts helps yield better performance.

Cost Effectiveness Analysis and Payment Policy Recommendation-Population-Based Survey with Big Data Methodology for Readmission Prevention of Patients with Paroxysmal Supraventricular Tachycardia treated with Radiofrequency Catheter Ablation.

Recurrence of paroxysmal supraventricular tachycardia (PSVT) has been reported to be lower in patients treated with radiofrequency catheter ablation (RFCA) than in those who are not. Few population-based surveys have stated the cost-effectiveness related to this treatment. We, therefore, performed a nationwide retrospective study using National Health Insurance Research Database (NHIRD) data from 2001-2012 in Taiwan. The incidence of PSVT-related admissions was computed from patients' first admission for a primary PSVT diagnosis. There were 21,086 patients hospitalized due to first-time PSVT, of whom 13,075 underwent RFCA, with 374 recurrences (2.86%). In contrast, 1751 (21.86%) of the remaining 8011 patients who did not receive RFCA, most of whom had financial concerns, experienced PSVT recurrence. The relative PSVT recurrence risk in those who did not receive RFCA was 7.6 times (95%CI: 6.67-8.33) that of those who did undergo RFCA. In conclusion, the PSVT recurrence rate was much higher in patients who did not receive RFCA at their first admission. Furthermore, RFCA proved cost-effective, with the ratio of the incremental cost-effectiveness ratio (ICER) and gross domestic product (GDP) being only 1.15. To prevent readmission and avoid incremental cost, the authority could provide a financial supplement for every patient so that the procedure is performed, reducing the PSVT-recurrence life-years (disease-specific DALY).

Using word embedding technique to efficiently represent protein sequences for identifying substrate specificities of transporters.

Membrane transport proteins and their substrate specificities play crucial roles in various cellular functions. Identifying the substrate specificities of membrane transport proteins is closely related to protein-target interaction prediction, drug design, membrane recruitment, and dysregulation analysis, thus being an important problem for bioinformatics researchers. In this study, we applied word embedding approach, the main cause for natural language processing breakout in recent years, to protein sequences of transporters. We defined each protein sequence based on the word embeddings and frequencies of its biological words. The protein features were then fed into machine learning models for prediction. We also varied the lengths of protein sequence's constituent biological words to find the optimal length which generated the most discriminative feature set. Compared to four other feature types created from protein sequences, our proposed features can help prediction models yield superior performance. Our best models reach an average area under the curve of 0.96 and 0.99, respectively on the 5-fold cross validation and the independent test. With this result, our study can help biologists identify transporters based on substrate specificities as well as provides a basis for further research that enriches a field of applying natural language processing techniques in bioinformatics.

The Short-Term Effects of Ambient Air Pollutants on Childhood Asthma Hospitalization in Taiwan: A National Study.

This investigation determined the effects of air pollution on childhood asthma hospitalization in regions with differing air pollution levels in Taiwan over a long time period. Data of childhood hospital admissions for asthma in patients aged 0⁻18 years and air quality in eight regions for the period 2001⁻2012 in Taiwan were collected. Poisson generalized linear regression analysis was employed to identify the relative risks of hospitalization due to asthma in children associated with exposure to varying levels of air pollutants with a change in the interquartile range after adjusting for temperature and relative humidity. Particulate matter ≤2.5 µm (PM2.5), particulate matter ≤10 µm (PM10), ozone (O3), sulfur dioxide (SO2), and nitrogen dioxide (NO2), were positively associated with childhood asthma hospitalization, while O3 was negatively associated with childhood asthma hospitalization. SO2 was identified as the most significant risk factor. The relative risks for asthma hospitalization associated with air pollutants were higher among children aged 0⁻5 years than aged 6⁻18 years and were higher among males than females. The effects of air pollution on childhood asthma were greater in the higher-level air pollution regions, while no association was observed in the lower-level air pollution regions. These findings may prove important for policymakers involved in implementing policies to reduce air pollution.

A survey of ambulatory-treated asthma and correlation with weather and air pollution conditions within Taiwan during 2001-2010.

Objective: This study of asthma was performed to evaluate annual trends in emergency department (ED) for 10 years. Weather and air pollution factors affecting asthma were also studied in order to identify the important factors and alert the public in advance. Methods: A survey of ambulatory-treated asthma patients was performed and the correlations with weather and air pollution factors examined in a cohort of one million patients in 2010. The fixed-cohort study analyzed trends, medical costs, and annual prevalence grouped by age and gender. Results: The number of asthma patients visiting EDs and non-emergency (non-ED) clinics significantly increased, with average annual percentage changes (AAPCs) of 2.3 and 4.6%, respectively. The average direct medical cost for EDs was increased significantly as compared with that of non-ED visits. Classification of asthma visits by hospital level indicated that local hospitals and others exhibited a significantly increasing trend (AAPC =15.3% [95% CI: 14.3-16.2]). The annual prevalence of asthma in males, females, and children was significantly increased (AAPCs of 1.5, 1.8, and 3.9%, respectively). Asthma patient hospitalizations were significantly correlated with temperature, humidity, and air pollution factors. Conclusions: The number of non-ED visits due to asthma increased, and the average direct medical cost for ED admissions also increased. Asthma patients tended to visit local hospitals primarily. Asthma visits by children increased, but a decrease was observed in adults. The number of hospitalized asthma patients was negatively correlated with temperature and humidity but positively correlated with the levels of PM2.5, PM10, and NO2.

Investigating the effect of daily sleep on memory capacity in college students.

BACKGROUND: Sleep is a natural periodic state of rest for body and mind and daily sleep affects physical and mental health. However, it is essential to address intensity of sleep characteristics affecting the memory capacity of humans positively or negatively. OBJECTIVE: Using wearable devices to observe and assess the effect of daily sleep on memory capacity of college students. METHODS: This study assessed the daily sleep characteristics and memory capacity of 39 college students who used wrist-worn devices. The spatial span test (SST) was used to evaluate the memory capacity. RESULTS: The study indicated a negative correlation between memory capacity and awake count on the test date and during the week before the test date (r=-0.153 (95% CI: -0.032, -0.282), r=-0.391 (95% CI: -0.520, -0.235), respectively). However, the minutes asleep on the test date and during the week before the test date positively affected memory capacity (r= 0.127 (95% CI: 0.220, 0.025), r= 0.370 (95% CI: 0.208, 0.500), respectively). In addition, spending ⩾ 6 hours and 42 minutes asleep on the test date or ⩾ 6 hours and 37 minutes asleep per day on average during the week before the test date resulted in a better memory capacity. CONCLUSIONS: A lower awake count led to a higher memory capacity in college students, as did more minutes asleep.

A Study of the Effects of Daily Physical Activity on Memory and Attention Capacities in College Students.

This study evaluated the relationship between daily physical activity (DPA) and memory capacity, as well as the association between daily activity and attention capacity, in college students in Taiwan. Participants (mean age = 20.79) wore wearable trackers for 106 days in order to collect DPA. These data were analyzed in association with their memory and attention capacities, as assessed using the spatial span test (SST) and the trail making test (TMT). The study showed significant negative correlations between memory capacity, time spent on the attention test (TSAT), calories burnt, and very active time duration (VATD) on the day before testing (r = -0.272, r = -0.176, r = 0.289, r = 0.254, resp.) and during the week prior to testing (r = -0.364, r = -0.395, r = 0.268, r = 0.241, resp.). The calories burnt and the VATD per day thresholds, which at best discriminated between normal-to-good and low attention capacity, were ≥2283 calories day-1, ≥20 minutes day-1 of very high activity (VHA) on the day before testing, or ≥13,640 calories week-1, ≥76 minutes week-1 of VHA during the week prior to testing. Findings indicated the short-term effects that VATD and calories burnt on the day before or during the week before testing significantly and negatively associated with memory and attention capacities of college students.

Application of a Time-Stratified Case-Crossover Design to Explore the Effects of Air Pollution and Season on Childhood Asthma Hospitalization in Cities of Differing Urban Patterns: Big Data Analytics of Government Open Data.

Few studies have assessed the lagged effects of levels of different urban city air pollutants and seasons on asthma hospitalization in children. This study used big data analysis to explore the effects of daily changes in air pollution and season on childhood asthma hospitalization from 2001 to 2010 in Taipei and Kaohsiung City, Taiwan. A time-stratified case-crossover study and conditional logistic regression analysis were employed to identify associations between the risk of hospitalization due to asthma in children and the levels of air pollutants (PM2.5, PM10, O3, SO2, and NO2) in the days preceding hospitalization. During the study period, 2900 children in Taipei and 1337 in Kaohsiung aged ≤15 years were hospitalized due to asthma for the first time. The results indicated that the levels of air pollutants were significantly associated with the risk of asthma hospitalization in children, and seasonal effects were observed. High levels of air pollution in Kaohsiung had greater effects than in Taipei after adjusting for seasonal variation. The most important factor was O3 in spring in Taipei. In children aged 0-6 years, asthma was associated with O3 in Taipei and SO2 in Kaohsiung, after controlling for the daily mean temperature and relative humidity.

Retrospective one-million-subject fixed-cohort survey of utilization of emergency departments due to traumatic causes in Taiwan, 2001-2010.

BACKGROUND: Epidemiological study was needed to evaluate trends in emergency department (ED) utilization that could be taken into account when making policy decisions regarding the delivery and distribution of medical resources. METHODS: A retrospective fixed-cohort study of emergency medical utilization from 2001 to 2010 was performed based on one-million people sampled in 2010 in Taiwan. Focusing on traumatic cases, the annual incidences in various groups split according to sex and age were calculated, and further information regarding location of trauma and type of trauma was obtained. RESULTS: In 2010, significantly greater proportions of male and younger subjects were visitors to EDs with a traumatic injury. During 2001-2010, the number of both traumatic cases and non-traumatic cases presenting at EDs significantly increased (average annual percentage change, AAPC 4.7 and 3.6, respectively) and a significantly greater direct medical cost associated with traumatic cases than non-traumatic cases was noted. Focusing on traumatic cases, most of these cases were directed to highest-level hospitals, accounting for 73.5-78.8 % of all traumatic cases, with a significant AAPC of 5.6. The traumatic ED visit annual incidence in males was 58.63 in 2001, which significantly increased to 69.35 per 1000 persons in 2010 (AAPC 1.5); and in females was 38.96 in 2001, which significantly increased to 50.73 per 1000 persons in 2010 (AAPC 2.5). Most of the traumatic cases treated in EDs were minor injuries, such as contusion with the skin intact, open wound of the upper limbs, open wound of the head, neck, or trunk, and other superficial injury (accounting for about 60 % of all cases). The traumatic categories of sprains/strains of joints and adjacent muscles, fractures of upper limbs, fractures of lower limbs, and fractures of the spine/trunk required greater medical resources and significantly positive AAPC values (4.3, 4.0, 4.5 and 6.8, respectively). CONCLUSIONS: Increased ED utilization due to traumatic causes, as assessed by the annual number of cases and incidence, average direct medical cost and highest-level hospital utilization, was observed from 2001 to 2010. Orthopedic-related injuries, including soft tissue trauma of extremities and various fractures, were the categories with the greatest increase in incidence.

Epidemiological survey of orthopedic joint dislocations based on nationwide insurance data in Taiwan, 2000-2005.

BACKGROUND: The epidemiology of acute orthopedic dislocations is poorly understood. A nationwide database provides a valuable resource for examining this issue in the Taiwanese population. METHODS: A 6-year retrospective cohort study of 1,000,000 randomly-sampled beneficiaries from the year 2005 was used as the original population. Based on the hospitalized and ambulatory data, the concomitant ICD9-CM diagnosis codes and treatment codes were evaluated and classified into 8 and 3 major categories, respectively. The cases matching both inclusive criteria of dislocation-related diagnosis codes and treatment codes were defined as incident cases. RESULTS: During 2000-2005, the estimated annual incidence (per 100,000 population) of total orthopedic dislocations in Taiwan was 42.1 (95%CI: 38.1-46.1). The major cause of these orthopedic dislocations was traffic accidents (57.4%), followed by accident falls (27.5%). The annual incidence dislocation by location was shoulder, 15.3; elbow, 7.7; wrist, 3.5; finger, 4.6; hip, 5.2; knee, 1.4; ankle, 2.0; and foot, 2.4. Approximately 16% of shoulder dislocations occurred with other concomitant fractures, compared with 17%, 53%, 16%, 76% and 52%, respectively, of dislocated elbow, wrist, hip, knee, and ankle cases. Including both simple and complex dislocated cases, the mean medical cost was US$612 for treatment of a shoulder dislocation, $504 for the elbow, $1,232 for the wrist, $1,103 for the hip, $1,888 for the knee, and $1,248 for the ankle. CONCLUSIONS: In Taiwan, three-quarters of all orthopedic dislocations were of the upper limbs. The most common complex fracture-dislocation was of the knee, followed by the wrist and the ankle. Those usually needed a treatment combined with open reduction of fractures and resulted in a higher direct medical expenditure.

Treatment incidence of and medical utilization for hospitalized subjects with pathologic fractures in Taiwan-Survey of the 2008 National Health Insurance data.

BACKGROUND: Almost all studies of pathologic fractures have been conducted based on patients with tumours and hospital-based data; however, in the present study, a nationwide epidemiological survey of pathologic fractures in Taiwan was performed and the medical utilization was calculated. METHODS: All claimants of Taiwan's National Health Insurance (NHI) Program in 2008 were included in the target population of this descriptive cross-sectional study. The registration and inpatient expenditure claims data by admission of all hospitalized subjects of the target population were examined and the concomitant International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes were evaluated and classified into seven major categories of fracture. RESULTS: A total of 5,244 incident cases of pathologic fracture were identified from the 2008 hospitalized patient claims data. The incidence of pathologic fracture of the humerus, distal radius/ulna, vertebrae, femoral neck, other part of the femur, and tibia/fibula was 0.67, 0.08, 10.58, 1.11, 0.56, and 0.11 per 100,000 people, respectively, and patients with those fractures were hospitalized for 43.9 ± 42.9, 31.1 ± 32.9, 29. 4 ± 34.4, 43.3 ± 41.2, 42.4 ± 38.1, and 42.0 ± 32.8 days, respectively, incurring an average medical cost of US$11,049 ± 12,730, US$9,181 ± 12,115, US$6,250 ± 8,021, US$9,619 ± 8,906, US$10,646 ± 11,024, and US$9,403 ± 9,882, respectively. The percentage of patients undergoing bone surgery for pathologic fracture of the humerus, radius/ulna, vertebrae, femoral neck, other part of the femur, and tibia/fibula was 31.2%, 44.4%, 11.3%, 46.5%, 48.4%, and 52.5% respectively. CONCLUSIONS: Comparing Taiwan to other countries, this study observed for Taiwan higher medical utilization and less-aggressive surgical intervention for patients hospitalized with pathologic fractures.