Search

Motivation Integrating multimodal data represents an effective approach to predicting biomedical characteristics, such as protein functions and disease outcomes. However, existing data integration approaches do not sufficiently address the heterogeneous semantics of multimodal data. In particular, early and intermediate approaches that rely on a uniform integrated representation reinforce the consensus among the modalities but may lose exclusive local information. The alternative late integration approach that can address this challenge has not been systematically studied for biomedical problems. Results We propose Ensemble Integration (EI) as a novel systematic implementation of the late integration approach. EI infers local predictive models from the individual data modalities using appropriate algorithms and uses heterogeneous ensemble algorithms to integrate these local models into a global predictive model. We also propose a novel interpretation method for EI models. We tested EI on the problems of predicting protein function from multimodal STRING data and mortality due to coronavirus disease 2019 (COVID-19) from multimodal data in electronic health records. We found that EI accomplished its goal of producing significantly more accurate predictions than each individual modality. It also performed better than several established early integration methods for each of these problems. The interpretation of a representative EI model for COVID-19 mortality prediction identified several disease-relevant features, such as laboratory test (blood urea nitrogen and calcium) and vital sign measurements (minimum oxygen saturation) and demographics (age). These results demonstrated the effectiveness of the EI framework for biomedical data integration and predictive modeling. Availability and implementation Code and data are available at https://github.com/GauravPandeyLab/ensemble_integration. Supplementary information Supplementary data are available at Bioinformatics Advances online.

We are pleased to introduce a first-of-its-kind algorithm that combines in-silico region detection and spatial gene-expression imputation. Spatial transcriptomics by 10X Visium (ST) is a new technology used to dissect gene and cell spatial organization. Analyzing this new type of data has two main challenges: automatically annotating the major tissue regions and excessive zero values of gene-expression due to high dropout rates. We developed a computational tool—MIST—that addresses both challenges by automatically identifying tissue regions and estimating missing gene-expression values for individual tissue regions. We validated MIST detected regions across multiple datasets using manual annotation on the histological staining images as references. We also demonstrated that MIST can accurately recover ST’s missing values through hold-out experiments. Furthermore, we showed that MIST could identify subtle intra-tissue heterogeneity and recover spatial gene-gene interaction signals. We therefore strongly encourage using MIST prior to downstream ST analysis because it provides unbiased region annotations and enables accurately de-noised spatial gene-expression profiles.

BackgroundPrediabetes and diabetes mellitus (preDM/DM) have become alarmingly prevalent among youth in recent years. However, simple questionnaire-based screening tools to reliably assess diabetes risk are only available for adults, not youth.MethodsAs a first step in developing such a tool, we used a large-scale dataset from the National Health and Nutritional Examination Survey (NHANES) to examine the performance of a published pediatric clinical screening guideline in identifying youth with preDM/DM based on American Diabetes Association diagnostic biomarkers. We assessed the agreement between the clinical guideline and biomarker criteria using established evaluation measures (sensitivity, specificity, positive/negative predictive value, F-measure for the positive/negative preDM/DM classes, and Kappa). We also compared the performance of the guideline to those of machine learning (ML) based preDM/DM classifiers derived from the NHANES dataset.ResultsApproximately 29% of the 2858 youth in our study population had preDM/DM based on biomarker criteria. The clinical guideline had a sensitivity of 43.1% and specificity of 67.6%, positive/negative predictive values of 35.2%/74.5%, positive/negative F-measures of 38.8%/70.9%, and Kappa of 0.1 (95%CI: 0.06-0.14). The performance of the guideline varied across demographic subgroups. Some ML-based classifiers performed comparably to or better than the screening guideline, especially in identifying preDM/DM youth (p=5.23×10−5).ConclusionsWe demonstrated that a recommended pediatric clinical screening guideline did not perform well in identifying preDM/DM status among youth. Additional work is needed to develop a simple yet accurate screener for youth diabetes risk, potentially by using advanced ML methods and a wider range of clinical and behavioral health data.Key MessagesAs a first step in developing a youth diabetes risk screening tool, we used a large-scale dataset from the National Health and Nutritional Examination Survey (NHANES) to examine the performance of a published pediatric clinical screening guideline in identifying youth with prediabetes/diabetes based on American Diabetes Association diagnostic biomarkers.In this cross-sectional study of youth, we found that the screening guideline correctly identified 43.1% of youth with prediabetes/diabetes, the performance of the guideline varied across demographic subgroups, and machine learning based classifiers performed comparably to or better than the screening guideline in identifying youth with prediabetes/diabetes.Additional work is needed to develop a simple yet accurate screener for youth diabetes risk, potentially by using advanced ML methods and a wider range of clinical and behavioral health data.

To enable efficient computations on rapidly growing big data, a variety of high-performance computing (HPC) platforms, such as traditional multi-processor systems, Hadoop and cloud computing systems, have been developed. On the analytics side of big data, several innovative machine learning methods have been developed to enable the extraction of accurate and actionable knowledge from large datasets. In particular, heterogeneous ensemble algorithms, which are designed to aggregate an unrestricted variety and number of analytical models, have performed well for a variety of prediction problems. However, the performance of these algorithms in terms of computational metrics, such as time requirement, disk space consumption and memory usage, on these HPC platforms has not been systematically examined yet. Here, we address this gap in knowledge by implementing these algorithms and systematically assessing their computational performance on traditional HPC and Hadoop platforms. Our results show that these implementations used the resources, especially disk space and memory, consistent with the respective designs of the platforms. Furthermore, due to the iterative nature of the heterogeneous ensemble computations, the traditional HPC system executed them faster than Hadoop, since an in-memory design is better suited for them than a disk-based one. Overall, our study sheds new light on the computational performance of ensemble algorithms and software frameworks on two prominent HPC platforms, and offers a systematic methodology for conducting similar assessments for other data analytics methods as well. Basic source code of our heterogeneous ensemble implementations, as well as the HPC performance assessments, are available at https://github.com/GauravPandeyLab/HPC-Ensemble.

Steel slag, a byproduct of the steel making process, has been adopted as a material to reduce non-point phosphorus (P) losses from agricultural land. Although substantial studies have been conducted on characterizing P removed by steel slag, few data are available on the removal of P under different conditions of P input, slag mass, and retention time (RT). The objective of this study was to investigate P removal efficiency as impacted by slag mass and RT at different physical locations through a horizontal steel slag column. Downstream slag segments were more efficient at removing P than upstream segments because they were exposed to more favorable conditions for calcium phosphate precipitation, specifically higher Ca2+ concentrations and pH. These results showed that P is removed in a moving front as Ca2+ and slag pH buffer capacity are consumed. In agreement with the calcium phosphate precipitation mechanism shown in previous studies, an increase in RT increased P removal, resulting in an estimated removal capacity of 61 mg kg&minus, 1 at a RT of 30 min. Results emphasized the importance of designing field scale structures with sufficient RT to accommodate the formation of calcium phosphate.

Soil and water losses in agriculture are major environmental problems worldwide, especially on the Loess Plateau, China. Summer fallow management may help to control soil erosion and conserve water. This study investigated the effects of wheat stubble on runoff, infiltration, and soil loss in laboratory plots under simulated rainfall. The treatments comprised wheat stubble cover (WS) and traditional plowing (TP) in runoff plots (4.0 m × 1.0 m) with three slope gradients (5, 10, and 15°) under simulated rainfall at 80 mm h−1 for 1 h. The runoff volume from WS plots was significantly less than that from TP. The runoff reduction with WS ranged from 91.92 to 92.83 % compared with TP. The runoff rates varied with the runoff volume in the same manner. The infiltration amount was higher with WS (94.8–96.2 % of rainwater infiltrated) than TP (35.4–57.1 %). The sediment concentration was significantly lower with WS than TP. Compared with TP (304.31–731.23 g m−2), the sediment losses were reduced dramatically in WS (2.41–3.78 g m−2) and the sediment loss slightly increased with slope; however, it was greatly increased as slope increased in TP. These results revealed that the stubble cover was the main factor reducing runoff and sediment losses and improving infiltration and that stubble showed a great potential to control erosion and conserve soil and water resources during the summer fallow period in the Loess Plateau region.

Soil erosion, particularly during summer fallow period, in sloped farmland has been identified as a critical threat to sustainable agricultural development in the Loess Plateau of China. However, the effects of varying tillage practices during the summer fallow period have not been fully investigated. The objective of this study was to determine the effects of tillage practices on runoff and sediment loss under plot (4.0 m in length and 1.0 m in width) rainfall simulations. Runoff initiation time, runoff amount, sediment loss and cumulative infiltration amount were recorded. One rainfall intensity (90 mm h −1 , 40 min), three slopes (5°, 10°, and 15°) and four prevalent tillage practices (Artificial Digging, AD; Artificial Hoeing, AH; Contour Plow, CP; and Traditional Plow, TP) were studied. Two indices, runoff reduction benefits (RRB) and sediment reduction benefits (SRB), were selected to evaluate the effect of the practice on controlling runoff and sediment. Compared to TP treatment, the AD, AH and CP were more effective in reducing the magnitude of runoff amount and sediment loss as well as increasing the rain water infiltration amount. CP with a 15° slope was an exception in which the sediment loss increased. In general, the SRB was greater than the RRB for a given tillage practice under the same slope, suggesting that the benefits from sediment reduction were more effective than the benefits from runoff reduction. RRB and SRB were the largest for AD, followed by AH and CP, irrespective of slopes. The capacity of the three tillage practices to reduce runoff and sediment decreased as the slope increased. In particular, AD at a 5° slope had the best performance in reducing runoff and sediment loss; however, SRB and RRB were −57% and 1%, respectively, for a CP plot with a 15° slope. This finding suggested that CP had a higher sediment loss and a similar runoff amount compared to those of TP. These results of plot simulated rainfall will contribute to our knowledge about the effects of different tillage practices on soil erosion from sloped farmland.

Heterogeneous ensembles are an effective approach in scenarios where the ideal data type and/or individual predictor are unclear for a given problem. These ensembles have shown promise for protein function prediction (PFP), but their ability to improve PFP at a large scale is unclear. The overall goal of this study is to critically assess this ability of a variety of heterogeneous ensemble methods across a multitude of functional terms, proteins and organisms. Our results show that these methods, especially Stacking using Logistic Regression, indeed produce more accurate predictions for a variety of Gene Ontology terms differing in size and specificity. To enable the application of these methods to other related problems, we have publicly shared the HPC-enabled code underlying this work as LargeGOPred (https://github.com/GauravPandeyLab/LargeGOPred). This work was supported in part by National Institutes of Health [R01GM114434] and by an IBM faculty award to GP. It was also partially supported by the Office of the Director of National Intelligence (ODNI), Intelligence Advanced Research Projects Activity (IARPA), via the Army Research Office (ARO) under Cooperative Agreement Number [W911NF-17-2-0105].

Accumulating evidence implicates gut microbiota as promising targets for the treatment of type 2 diabetes mellitus (T2DM). With a randomized clinical trial, we tested the hypothesis that alteration of gut microbiota may be involved in the alleviation of T2DM with hyperlipidemia by metformin and a specifically designed herbal formula (AMC). Four hundred fifty patients with T2DM and hyperlipidemia were randomly assigned to either the metformin- or AMC-treated group. After 12 weeks of treatment, 100 patients were randomly selected from each group and assessed for clinical improvement. The effects of the two drugs on the intestinal microbiota were evaluated by analyzing the V3 and V4 regions of the 16S rRNA gene by Illumina sequencing and multivariate statistical methods. Both metformin and AMC significantly alleviated hyperglycemia and hyperlipidemia and shifted gut microbiota structure in diabetic patients. They significantly increased a coabundant group represented by Blautia spp., which significantly correlated with the improvements in glucose and lipid homeostasis. However, AMC showed better efficacies in improving homeostasis model assessment of insulin resistance (HOMA-IR) and plasma triglyceride and also exerted a larger effect on gut microbiota. Furthermore, only AMC increased the coabundant group represented by Faecalibacterium spp., which was previously reported to be associated with the alleviation of T2DM in a randomized clinical trial. Metformin and the Chinese herbal formula may ameliorate type 2 diabetes with hyperlipidemia via enriching beneficial bacteria, such as Blautia and Faecalibacterium spp. IMPORTANCE Metabolic diseases such as T2DM and obesity have become a worldwide public health threat. Accumulating evidence indicates that gut microbiota can causatively arouse metabolic diseases, and thus the gut microbiota serves as a promising target for disease control. In this study, we evaluated the role of gut microbiota during improvements in hyperglycemia and hyperlipidemia by two drugs: metformin and a specifically designed Chinese herbal formula (AMC) for diabetic patients with hyperlipidemia. Both drugs significantly ameliorated blood glucose and lipid levels and shifted the gut microbiota. Blautia spp. were identified as being associated with improvements in glucose and lipid homeostasis for both drugs. AMC exerted larger effects on the gut microbiota together with better efficacies in improving HOMA-IR and plasma triglyceride levels, which were associated with the enrichment of Faecalibacterium spp. In brief, these data suggest that gut microbiota might be involved in the alleviation of diabetes with hyperlipidemia by metformin and the AMC herbal formula.

This paper constructs an export model of heterogeneous enterprises with financing constraints and put forward the hypothesis of financing constraints and export. Based on the panel data of China's industry enterprises from 2005 to 2014, this paper establishes a model and estimates the hypothesis. The main conclusions are as follows: the larger the financing constraint, the less export; Compared with domestic financing constraints, export enterprises are more constrained by external financing constraints; In the process of export, the state-owned enterprises are less constrained than the private ones, and the private ones are heavily affected by the financing constraints. Among them, the influence of exogenous financing constraints is the biggest.

Based on the existing literature, this paper analyzes the mechanism of financing constraints affect going global and puts forward corresponding theoretical hypotheses. By using the panel data of Yunnan Industrial Enterprises from 2005 to 2014, this paper uses the export as a measure of going global and constructs a measurement model based on the total amount of enterprise profit, interest expense and control variable as explanatory variables, and conducts an empirical test. The research shows that the larger the financial constraints, the fewer exports; compared with the endogenous financing constraints, export-oriented enterprises are more constrained by exogenous financing.

Purpose To measure the displacement of the tumor of the gross tumor volume (GTV) of thoracic esophageal cancer in the calm states of end-inspiration and end-expiration for determining the internal margin of the GTV (IGTV). Methods Twenty-two patients with thoracic esophageal cancer who were unable to undergo surgery were identified in our hospital. The patients received radiotherapy. By using 16-slice spiral computed tomography (CT), we acquired the calm states of end-inspiration and end-expiration. The displacement and volume changes in tumor target volume were measured, and the changes were analyzed to determine if these were associated with the tidal volume and the location and length of the target volume V. In the end, we analyzed the displacement of tumor target volume and calculated the internal margin of the GTV by empirical formula. Results The average tidal volume was 463.6 ml. The average GTV at end-inspiration was 33.3 ml and at end-expiration was 33.35 ml. Three was not any significant between two groups (T = −0.034, P > 0.05). The IGTV (X-axis direction) was 3.09 mm for the right sector and 4.08 mm for the left border; the IGTV (Z-axis direction) was 3.96 mm for the anterior border and 2.83 mm for the posterior border; and the IGTV (Y-axis direction) was 7.31 mm for the upper boundary (head direction) and 10.16 mm for the lower boundary (feet direction). The motion of the GTV showed no significant correlation with the tidal volume of patients and the length of the tumor, but in relation to the tumor location, the displacement of the lower thoracic and the middle thoracic target volumes occurred in the direction of the anterior and right, which were not significantly different (T = 0.859, 0.229, P > 0.05) The significant differences were observed for the other directions (P Conclusions Because of respiratory and organ movements, the displacement of the tumor target volume was different in all directions. Therefore, we recommend that expansion of the planning target volume during clinical radiation treatment needs to include the displacement of the tumor target volume caused by respiratory and organ movements during each radiotherapy session.

Sixteen of 24 infants with vomiting were examined by real-time sonography, which established the diagnosis of hypertrophic pyloric stenosis (HPS); 12 underwent surgery. Of the three parameters measured, pyloric muscle wall thickness was the most reliable and accurate for the diagnosis of HPS. We suggest that a muscle wall thickness of 4 mm or more and stenosis index greater than 50% are standard findings in patients with HPS. Serial real-time sonography following pyloromyotomy showed that the pyloric tumor changed significantly during the 1st postoperative week and returned to normal size by the end of the 4th week.