Machine Learning-Based Determination of Sampling Depth for Complex Environmental Systems: Case Study with Single-Cell Raman Spectroscopy Data in EBPR Systems.

Rapid progress in various advanced analytical methods, such as single-cell technologies, enable unprecedented and deeper understanding of microbial ecology beyond the resolution of conventional approaches. A major application challenge exists in the determination of sufficient sample size without sufficient prior knowledge of the community complexity and, the need to balance between statistical power and limited time or resources. This hinders the desired standardization and wider application of these technologies. Here, we proposed, tested and validated a computational sampling size assessment protocol taking advantage of a metric, named kernel divergence. This metric has two advantages: First, it directly compares data set-wise distributional differences with no requirements on human intervention or prior knowledge-based preclassification. Second, minimal assumptions in distribution and sample space are made in data processing to enhance its application domain. This enables test-verified appropriate handling of data sets with both linear and nonlinear relationships. The model was then validated in a case study with Single-cell Raman Spectroscopy (SCRS) phenotyping data sets from eight different enhanced biological phosphorus removal (EBPR) activated sludge communities located across North America. The model allows the determination of sufficient sampling size for any targeted or customized information capture capacity or resolution level. Promised by its flexibility and minimal restriction of input data types, the proposed method is expected to be a standardized approach for sampling size optimization, enabling more comparable and reproducible experiments and analysis on complex environmental samples. Finally, these advantages enable the extension of the capability to other single-cell technologies or environmental applications with data sets exhibiting continuous features.

Framework for denoising Monte Carlo photon transport simulations using deep learning.

SIGNIFICANCE: The Monte Carlo (MC) method is widely used as the gold-standard for modeling light propagation inside turbid media, such as human tissues, but combating its inherent stochastic noise requires one to simulate a large number photons, resulting in high computational burdens. AIM: We aim to develop an effective image denoising technique using deep learning (DL) to dramatically improve the low-photon MC simulation result quality, equivalently bringing further acceleration to the MC method. APPROACH: We developed a cascade-network combining DnCNN with UNet, while extending a range of established image denoising neural-network architectures, including DnCNN, UNet, DRUNet, and deep residual-learning for denoising MC renderings (ResMCNet), in handling three-dimensional MC data and compared their performances against model-based denoising algorithms. We also developed a simple yet effective approach to creating synthetic datasets that can be used to train DL-based MC denoisers. RESULTS: Overall, DL-based image denoising algorithms exhibit significantly higher image quality improvements over traditional model-based denoising algorithms. Among the tested DL denoisers, our cascade network yields a 14 to 19 dB improvement in signal-to-noise ratio, which is equivalent to simulating 25 × to 78 × more photons. Other DL-based methods yielded similar results, with our method performing noticeably better with low-photon inputs and ResMCNet along with DRUNet performing better with high-photon inputs. Our cascade network achieved the highest quality when denoising complex domains, including brain and mouse atlases. CONCLUSIONS: Incorporating state-of-the-art DL denoising techniques can equivalently reduce the computation time of MC simulations by one to two orders of magnitude. Our open-source MC denoising codes and data can be freely accessed at http://mcx.space/.

Machine learning-based biomarkers identification from toxicogenomics - Bridging to regulatory relevant phenotypic endpoints.

One of the major challenges in realization and implementations of the Tox21 vision is the urgent need to establish quantitative link between in-vitro assay molecular endpoint and in-vivo regulatory-relevant phenotypic toxicity endpoint. Current toxicomics approach still mostly rely on large number of redundant markers without pre-selection or ranking, therefore, selection of relevant biomarkers with minimal redundancy would reduce the number of markers to be monitored and reduce the cost, time, and complexity of the toxicity screening and risk monitoring. Here, we demonstrated that, using time series toxicomics in-vitro assay along with machine learning-based feature selection (maximum relevance and minimum redundancy (MRMR)) and classification method (support vector machine (SVM)), an "optimal" number of biomarkers with minimum redundancy can be identified for prediction of phenotypic toxicity endpoints with good accuracy. We included two case studies for in-vivo carcinogenicity and Ames genotoxicity prediction, using 20 selected chemicals including model genotoxic chemicals and negative controls, respectively. The results suggested that, employing the adverse outcome pathway (AOP) concept, molecular endpoints based on a relatively small number of properly selected biomarker-ensemble involved in the conserved DNA-damage and repair pathways among eukaryotes, were able to predict both Ames genotoxicity endpoints and in-vivo carcinogenicity in rats. A prediction accuracy of 76% with AUC = 0.81 was achieved while predicting in-vivo carcinogenicity with the top-ranked five biomarkers. For Ames genotoxicity prediction, the top-ranked five biomarkers were able to achieve prediction accuracy of 70% with AUC = 0.75. However, the specific biomarkers identified as the top-ranked five biomarkers are different for the two different phenotypic genotoxicity assays. The top-ranked biomarkers for the in-vivo carcinogenicity prediction mainly focused on double strand break repair and DNA recombination, whereas the selected top-ranked biomarkers for Ames genotoxicity prediction are associated with base- and nucleotide-excision repair The method developed in this study will help to fill in the knowledge gap in phenotypic anchoring and predictive toxicology, and contribute to the progress in the implementation of tox 21 vision for environmental and health applications.

A Secure and Reusable Software Architecture for Supporting Online Data Harmonization.

Retrospective data harmonization across multiple research cohorts and studies is frequently done to increase statistical power, provide comparison analysis, and create a richer data source for data mining. However, when combining disparate data sources, harmonization projects face data management and analysis challenges. These include differences in the data dictionaries and variable definitions, privacy concerns surrounding health data representing sensitive populations, and lack of properly defined data models. With the availability of mature open-source web-based database technologies, developing a complete software architecture to overcome the challenges associated with the harmonization process can alleviate many roadblocks. By leveraging state-of-the-art software engineering and database principles, we can ensure data quality and enable cross-center online access and collaboration. This paper outlines a complete software architecture developed and customized using the Django web framework, leveraged to harmonize sensitive data collected from three NIH-support birth cohorts. We describe our framework and show how we successfully overcame challenges faced when harmonizing data from these cohorts. We discuss our efforts in data cleaning, data sharing, data transformation, data visualization, and analytics, while reflecting on what we have learned to date from these harmonized datasets.

Effects of trust-based decision making in disrupted supply chains.

The United States has experienced prolonged severe shortages of vital medications over the past two decades. The causes underlying the severity and prolongation of these shortages are complex, in part due to the complexity of the underlying supply chain networks, which involve supplier-buyer interactions across multiple entities with competitive and cooperative goals. This leads to interesting challenges in maintaining consistent interactions and trust among the entities. Furthermore, disruptions in supply chains influence trust by inducing over-reactive behaviors across the network, thereby impacting the ability to consistently meet the resulting fluctuating demand. To explore these issues, we model a pharmaceutical supply chain with boundedly rational artificial decision makers capable of reasoning about the motivations and behaviors of others. We use multiagent simulations where each agent represents a key decision maker in a pharmaceutical supply chain. The agents possess a Theory-of-Mind capability to reason about the beliefs, and past and future behaviors of other agents, which allows them to assess other agents' trustworthiness. Further, each agent has beliefs about others' perceptions of its own trustworthiness that, in turn, impact its behavior. Our experiments reveal several counter-intuitive results showing how small, local disruptions can have cascading global consequences that persist over time. For example, a buyer, to protect itself from disruptions, may dynamically shift to ordering from suppliers with a higher perceived trustworthiness, while the supplier may prefer buyers with more stable ordering behavior. This asymmetry can put the trust-sensitive buyer at a disadvantage during shortages. Further, we demonstrate how the timing and scale of disruptions interact with a buyer's sensitivity to trustworthiness. This interaction can engender different behaviors and impact the overall supply chain performance, either prolonging and exacerbating even small local disruptions, or mitigating a disruption's effects. Additionally, we discuss the implications of these results for supply chain operations.

Effect of hydrogeological and anthropogenic factors on the spatial and temporal distribution of CVOCs in the karst system of northern Puerto Rico.

This study evaluates factors affecting the spatial and temporal distribution of chlorinated volatile organic contaminants (CVOCs) in the highly productive aquifers of the karst region in northern Puerto Rico (KR-NPR). Historical records from 1982 to 2016 are analyzed using spatial and statistical methods to evaluate hydrogeological and anthropogenic factors affecting the presence and concentrations of multiple CVOCs in the KR-NPR. Results show extensive spatial and temporal distributions of CVOCs, as single entities and as mixtures. It is found that at least one type of CVOC is present above detection limits in 64% of the samples and 77% of the sampling sites during the study period. CVOC distribution in the KR-NPR is contaminant-dependent, with some species being strongly influenced by the source of contamination and hydrogeological characteristics of the system. Persistent presence of CVOCs in the KR-NPR system, even after contaminated sites have been subjected to active remediation, reflect the high capacity of the system to store and slowly release contaminants over long periods of time. This study shows that karst aquifers are highly vulnerable to contamination and can serve as a long-term route of contaminants to potential points of exposure.

The influence of hydrogeological and anthropogenic variables on phthalate contamination in eogenetic karst groundwater systems.

This study investigates the occurrence of six phthalates and distribution of the three most-detected phthalates in the karst region of northern Puerto Rico (KRNPR) using data from historical records and current field measurements. Statistical data analyses, including ANOVA, Chi-Square, and logistic regression models are used to examine the major factors affecting the presence and concentrations of phthalates in the KRNPR. The most detected phthalates include DEHP, DBP, and DEP. At least one phthalate specie is detected above DL in 7% of the samples and 24% of the sampling sites. Concentrations of total phthalates average 5.08 ±â€¯1.37 µg L-1, and range from 0.093 to 58.4 µg L-1. The analysis shows extensive spatial and temporal presence of phthalates resulting from dispersed phthalate sources throughout the karst aquifers. Hydrogeological factors are significantly more important in predicting the presence and concentrations of phthalates in eogenetic karst aquifers than anthropogenic factors. Among the hydrogeological factors, time of detection and hydraulic conductivities larger than 300 m d-1 are the most influential factors. Persistent presence through time reflects continuous sources of phthalates entering the aquifers and a high capacity of the karst aquifers to store and slowly release contaminants for long periods of time. The influence of hydraulic conductivity reveals the importance of contaminant fate and transport mechanisms from contamination sources. This study improves the understanding of factors affecting the spatial variability and fate of phthalates in karst aquifers, and allows us to better predict their occurrence based on these factors.

Scalable and massively parallel Monte Carlo photon transport simulations for heterogeneous computing platforms.

We present a highly scalable Monte Carlo (MC) three-dimensional photon transport simulation platform designed for heterogeneous computing systems. Through the development of a massively parallel MC algorithm using the Open Computing Language framework, this research extends our existing graphics processing unit (GPU)-accelerated MC technique to a highly scalable vendor-independent heterogeneous computing environment, achieving significantly improved performance and software portability. A number of parallel computing techniques are investigated to achieve portable performance over a wide range of computing hardware. Furthermore, multiple thread-level and device-level load-balancing strategies are developed to obtain efficient simulations using multiple central processing units and GPUs.

Patterns of temporal scaling of groundwater level fluctuation.

We studied the fractal scaling behavior of groundwater level fluctuation for various types of aquifers in Puerto Rico using the methods of (1) detrended fluctuation analysis (DFA) to examine the monofractality and (2) wavelet transform maximum modulus (WTMM) to analyze the multifractality. The DFA results show that fractals exist in groundwater fluctuations of all the aquifers with scaling patterns that are anti-persistent (1 < ß < 1.5; 1.32 ± 0.12, 18 wells) or persistent (ß > 1.5; 1.62 ± 0.07, 4 wells). The multi-fractal analysis confirmed the need to characterize these highly complex processes with multifractality, which originated from the stochastic distribution of the irregularly-shaped fluctuations. The singularity spectra of the fluctuation processes in each well were site specific. We found a general elevational effect with smaller fractal scaling coefficients in the shallower wells, except for the Northern Karst Aquifer Upper System. High spatial variability of fractal scaling of groundwater level fluctuations in the karst aquifer is due to the coupled effects of anthropogenic perturbations, precipitation, elevation and particularly the high heterogeneous hydrogeological conditions.

Big Data Analysis on Puerto Rico Testsite for Exploring Contamination Threats.

In this paper, we present the use of Principal Component Analysis and customized software, to accelerate the spectral analysis of biological samples. The work is part of the mission of the National Institute of Environmental Health Sciences sponsored Puerto Rico Testsite for Exploring Contamination Threats Center, establishing linkages between environmental pollutants and preterm birth. This paper provides an overview of the data repository developed for the Center, and presents a use case analysis of biological sample data maintained in the database system.

Spatiotemporal changes of CVOC concentrations in karst aquifers: analysis of three decades of data from Puerto Rico.

We studied the spatial and temporal distribution patterns of Chlorinated Volatile Organic Compounds (CVOCs) in the karst aquifers in northern Puerto Rico (1982-2013). Seventeen CVOCs were widely detected across the study area, with the most detected and persistent contaminated CVOCs including trichloroethylene (TCE), tetrachloroethylene (PCE), carbon tetrachloride (CT), chloroform (TCM), and methylene chloride (DCM). Historically, 471 (76%) and 319 (52%) of the 615 sampling sites have CVOC concentrations above the detection limit and maximum contamination level (MCL), respectively. The spatiotemporal patterns of the CVOC concentrations showed two clusters of contaminated areas, one near the Superfund site "Upjohn" and another near "Vega Alta Public Supply Wells." Despite a decreasing trend in concentrations, there is a general northward movement and spreading of contaminants even beyond the extent of known sources of the Superfund and landfill sites. Our analyses suggest that, besides the source conditions, karst characteristics (high heterogeneity, complex hydraulic and biochemical environment) are linked to the long-term spatiotemporal patterns of CVOCs in groundwater.

Accelerating mesh-based Monte Carlo method on modern CPU architectures.

In this report, we discuss the use of contemporary ray-tracing techniques to accelerate 3D mesh-based Monte Carlo photon transport simulations. Single Instruction Multiple Data (SIMD) based computation and branch-less design are exploited to accelerate ray-tetrahedron intersection tests and yield a 2-fold speed-up for ray-tracing calculations on a multi-core CPU. As part of this work, we have also studied SIMD-accelerated random number generators and math functions. The combination of these techniques achieved an overall improvement of 22% in simulation speed as compared to using a non-SIMD implementation. We applied this new method to analyze a complex numerical phantom and both the phantom data and the improved code are available as open-source software at http://mcx.sourceforge.net/mmc/.

Towards the development of an error checker for radiotherapy treatment plans: a preliminary study.

Major accidents can happen during radiotherapy, with an extremely severe consequence to both patients and clinical professionals. We propose to use machine learning and data mining techniques to help detect large human errors in a radiotherapy treatment plan, as a complement to human inspection. One such technique is computer clustering. The basic idea of using clustering algorithms for outlier detection is to first cluster (based on the treatment parameters) a large number of patient treatment plans. Then, when checking a new treatment plan, the parameters of the plan will be tested to see whether or not they belong to the established clusters. If not, they will be considered as 'outliers' and therefore highlighted to catch the attention of the human chart checkers. As a preliminary study, we applied the K-means clustering algorithm to a simple patient model, i.e., 'four-field' box prostate treatment. One thousand plans were used to build the clusters while another 650 plans were used to test the proposed method. It was found that there are eight distinct clusters. At the error levels of +/-100% of the original values of the monitor unit, the detection rate is about 100%. At +/-50% error level, the detection rate is about 80%. The false positive rate is about 10%. When purposely changing the beam energy to a value different from that in the treatment plan, the detection rate is 100% for posterior, right-lateral and left-lateral fields, and about 77% for the anterior field. This preliminary work has shown promise for developing the proposed automatic outlier detection software, although more efforts will still be required.

A finite state model for respiratory motion analysis in image guided radiation therapy.

Effective image guided radiation treatment of a moving tumour requires adequate information on respiratory motion characteristics. For margin expansion, beam tracking and respiratory gating, the tumour motion must be quantified for pretreatment planning and monitored on-line. We propose a finite state model for respiratory motion analysis that captures our natural understanding of breathing stages. In this model, a regular breathing cycle is represented by three line segments, exhale, end-of-exhale and inhale, while abnormal breathing is represented by an irregular breathing state. In addition, we describe an on-line implementation of this model in one dimension. We found this model can accurately characterize a wide variety of patient breathing patterns. This model was used to describe the respiratory motion for 23 patients with peak-to-peak motion greater than 7 mm. The average root mean square error over all patients was less than 1 mm and no patient has an error worse than 1.5 mm. Our model provides a convenient tool to quantify respiratory motion characteristics, such as patterns of frequency changes and amplitude changes, and can be applied to internal or external motion, including internal tumour position, abdominal surface, diaphragm, spirometry and other surrogates.