Outlier Analysis Research Articles

Perceived Trust, Convenience and Promotion For the Adoption of e-Wallet

The outbreak of Covid-19 had accelerated cashless and contactless digital payment like e-wallet. It was projected that the e-wallet market in Malaysia has the potential to grow to USD 20 billion by 2024. The Bank Negara Malaysia (BNM) had granted licence to 53 e-wallet providers and the competition remains fierce. This study is conducted to investigate the factors affecting the adoption of e-wallet by extending TAM model by perceived trust and promotion. The target population was those aged between 18 to 50 and 249 valid responses were collected by using purposive sampling technique. The demographic frequencies and outlier analysis were assessed using the Statistical Package for the Social Sciences (SPSS) while the measurement and structural models were examined using partial least squares-structural equation modelling (PLS-SEM). The results show that perceived trust, convenience, and promotion significantly affect the adoption of e-wallet. This study served as a guidance to the e-wallet service providers to gain competitive advantage in competitive e-wallet industry in terms of attracting new user and also retaining existing users.

Country-wide solar power load profile for Germany 2015 to 2019: The impact of system curtailments on prediction models

A large (43824 h) country-wide solar power load profile (LP), solar irradiance and meteorological dataset (ten variables) for Germany covering years 2015 to 2019 is compiled and forecast with eight machine learning and deep learning (ML/DL) algorithms. Analysis reveals that system curtailments are likely responsible for some outlying predictions. The adaptive boosting (ADA) and random forest (RF) algorithms outperform convolutional neural networks and other algorithms in supervised prediction and forecasting tasks with the dataset. Once tuned with 2015 to 2018 data, ADA and RF forecast 2019 hourly data with root mean squared error (RMSE) of < 0.03. Similar forecasting accuracy is achieved using smaller datasets of three months of historical data from 2015 to 2018 to forecast hourly LP in each month of 2019. Forecasts for July 2019 are associated with the highest errors (RMSE = 0.41). The transparent open box (TOB) algorithm, because it can reveal details of data matching contributions to each of its forecasts, is used to data mine the July 2019 forecasts and conduct in-depth outlier analysis. It reveals nine extreme outliers that each overestimate LP based on irradiance and meteorological inputs. Analysis suggests that substantial overestimates leading to high RMSE for July 2019 are likely due to system curtailments. It is beneficial to combine less transparent (e.g., ADA and RF) with more transparent (e.g. TOB) ML algorithms to accurately forecast and data mine large solar power data sets. It also indicates that solar LP forecasts cannot always rely on irradiance and meteorological variables in isolation. There is a need to be mindful of system constraints and market conditions when predicting LP on a country-wide basis.

OTL web interface: outlier testing in lifetime data

A web interface, named OTL, is developed to perform outlier analysis in exponential, Weibull, lognormal, and log-Laplace samples online over the Internet through Shiny and R. It is a tool to evaluate the exact or approximate distributions of block and consecutive test statistics for testing the presence of outliers that have been discussed over the years in the literature. As it allows for identifying outliers without using the classical extensive tables, OTL can be very useful to academics, policy-makers and executives to compare and understand the results of outlier testing procedures more efficiently in the case of lifetime data.

Electricity Spot Price Modeling and Forecasting in European Markets

In many competitive electricity markets around the world, the dynamic behavior of hourly electricity prices is subject to significant uncertainty and volatility due to electricity demand, availability of generation sources, fuel costs, and power plant availability. This work is devoted to describing and comparing the dynamics of electricity prices for some markets in Europe, selecting the five countries representing the largest economies in Western Europe (France, Germany, Italy, Spain, and the United Kingdom). Additionally, Denmark is included in the study to assess whether the size of the country is a determinant of price behavior. The six datasets of hourly price series, which exhibits a strong daily seasonality, are modelled using the most relevant well-known statistical models for time series analysis: ARIMA models and different versions of GARCH models. The comparison of the estimated models’ parameters, the analysis of outliers’ rate of appearance and the evaluation of out-of-sample one-day-ahead forecast let us draw some insightful similarities and dissimilarities between the analyzed countries.

Biological variation estimates obtained from Chinese subjects for 32 biochemical measurands in serum.

The European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) have established a program of work to make available, and to enable delivery of well characterized data describing the biological variation (BV) of clinically important measurands. Guided by the EFLM work the study presented here delivers BV estimates obtained from Chinese subjects for 32 measurands in serum. Samples were drawn from 48 healthy volunteers (26 males, 22 females; age range, 21-45years) for 5 consecutive weeks at Chinese laboratory. Sera were stored at-80°C before triplicate analysis of all samples on a Cobas 8000 modular analyzer series. Outlier and homogeneity analyses were performed, followed by CV-ANOVA, to determine BV estimates with confidence intervals. The within-subject biological variation (CVI) estimates for 30 of the 32 measurands studied, were lower than listed on the EFLM database; the exceptions were alanine aminotransferase (ALT), lipoprotein (a) (LP(a)). Most of the between-subject biological variation (CVG) estimates were lower than the EFLM database entries. This study delivers BV data for a Chinese population to supplement the EFLM BV database. Population differences may have an impact on applications of BV Data.

Transcriptome deregulation of peripheral monocytes and whole blood in GBA-related Parkinson’s disease

BackgroundGenetic mutations in beta-glucocerebrosidase (GBA) represent the major genetic risk factor for Parkinson’s disease (PD). GBA participates in both the endo-lysosomal pathway and the immune response, two important mechanisms involved in the pathogenesis of PD. However, modifiers of GBA penetrance have not yet been fully elucidated.MethodsWe characterized the transcriptomic profiles of circulating monocytes in a population of patients with PD and healthy controls (CTRL) with and without GBA variants (n = 23 PD/GBA, 13 CTRL/GBA, 56 PD, 66 CTRL) and whole blood (n = 616 PD, 362 CTRL, 127 PD/GBA, 165 CTRL/GBA). Differential expression analysis, pathway enrichment analysis, and outlier detection were performed. Ultrastructural characterization of isolated CD14+ monocytes in the four groups was also performed through electron microscopy.ResultsWe observed hundreds of differentially expressed genes and dysregulated pathways when comparing manifesting and non-manifesting GBA mutation carriers. Specifically, when compared to idiopathic PD, PD/GBA showed dysregulation in genes involved in alpha-synuclein degradation, aging and amyloid processing. Gene-based outlier analysis confirmed the involvement of lysosomal, membrane trafficking, and mitochondrial processing in manifesting compared to non-manifesting GBA-carriers, as also observed at the ultrastructural levels. Transcriptomic results were only partially replicated in an independent cohort of whole blood samples, suggesting cell-type specific changes.ConclusionsOverall, our transcriptomic analysis of primary monocytes identified gene targets and biological processes that can help in understanding the pathogenic mechanisms associated with GBA mutations in the context of PD.

Remote wireless monitoring of localized corrosion using compact solitary waves

This article presents a method to monitor corrosion remotely, based on highly nonlinear solitary waves, which are compact and nondispersive. In the study presented in this article, two types of solitary wave transducers were used to monitor accelerated localized corrosion on a steel plate. The first type consists of a chain of spherical particles surmounted by a commercial solenoid wired to, and controlled by, a data acquisition system used to lift and release the first particle of the chain and induce the mechanical impacts and stress waves in the chain. The chain included a piezoelectric wafer disk, also wired to the same data acquisition system, to sense, digitize, and store the propagating waves for post-processing. The second type of transducer was identical to the first one but the data acquisition system was replaced by a wireless node that communicated with a mobile device using a Bluetooth connection. Eight transducers were used to monitor the plate for over a week to detect the onset and progression of localized corrosion. Corrosion detection was performed by extracting a few features from the time waveforms and feeding these features to an outlier analysis algorithm based on the Mahalanobis distance. The results of the experiment showed the effectiveness of the proposed monitoring approach at detecting defects close to the transducers and confirm previous numerical predictions by the authors. The experiments also provided evidence that the performance of the wireless transducers is nearly identical to the performance of their wired counterparts, paving the way to a new paradigm for the structural health monitoring of remote structural components in harsh environments.

Exclusive use of hygienic menstrual absorbents among rural adolescent women in India: A geospatial analysis

ObjectivesPoor menstrual hygiene can contribute significantly to female morbidity, yet the use of hygienic absorbents during menstruation is considerably low among rural adolescent women. This study examines the spatial patterns of hygienic absorbent use during menstruation among rural adolescent women to identify the districts of India where it is significantly clustered. Further, it examines the factors responsible for the spatial disparity in the exclusive use of hygienic absorbents. MethodsWe applied Global Moran's I to estimate the degree of spatial heterogeneity. In addition, we ran Cluster and Outlier Analysis (Anselin Local Moran's I) to locate clusters and outliers in the exclusive use of hygienic absorbents across the districts. Ordinary least squares, spatial lag, and spatial error models were used to identify the determinants of exclusive use of hygienic absorbents. ResultsThe Global Moran's I value for the outcome variable was 0.51, indicating a positive spatial autocorrelation in the exclusive use of hygienic absorbents. Central Indian districts had statistically significant cold spots, whereas statistically significant hot spots were discovered in south Indian districts. Our study identified that women's education, religion, social group, household wealth, mass media exposure, and women's employment status were significant determinants of the exclusive use of hygienic absorbents across Indian districts. ConclusionThere is a substantial spatial disparity in the exclusive use of hygienic absorbents among rural adolescent women in India. The use of hygienic absorbents by rural adolescent women could be encouraged through mass media campaigns, rural healthcare personnel, and free or subsidized delivery of hygienic absorbents in the identified cold spots.

Circulating Concentrations of Nutrition-Related Factors Are Not Causally Associated With Atrial Fibrillation: A Mendelian Randomization Study.

Observational studies reported conflicting results regarding the association between circulating concentrations of nutrition-related factors and atrial fibrillation (AF). The aim of this study was to evaluate the potential causal effect of 8 circulating nutrition-related factors (vitamin B12, vitamin E, folate, retinol, β-carotene, iron, zinc, and copper) on AF risk using mendelian randomization (MR). Summary-level data for the nutrition-related factors and AF were obtained from genome-wide association studies conducted among individuals of European ancestry. The genome-wide association study on AF included 60,620 cases and 970,216 controls. A 2-sample MR design was applied for evaluating the causal association. In the primary MR analyses, the inverse variance-weighted method did not identify any causal effect of circulating concentrations of vitamin B12 [β = 0.000, standard error (SE) = 0.021, P = 0.994], vitamin E (β = 0.080, SE = 0.152, P = 0.600), retinol (β = 0.098, SE = 0.397, P = 0.806), folate (β = -0.006, SE = 0.052, P = 0.901), β-carotene (β = 0.014, SE = 0.025, P = 0.560), iron (β = -0.009, SE = 0.072, P = 0.905), zinc (β = 0.038, SE = 0.032, P = 0.239), and copper (β = -0.012, SE = 0.023, P = 0.589) on AF. The MR-Egger and MR pleiotropy residual sum and outlier (MR-PRESSO) analyses did not suggest the presence of pleiotropy. In addition, the lack of association remained in the leave-one-out analysis. This MR study indicates no causal association of circulating concentrations of vitamin B12, vitamin E, folate, retinol, β-carotene, iron, zinc, and copper with AF.

Likely community transmission of COVID-19 infections between neighboring, persistent hotspots in Ontario, Canada.

Introduction: This study aimed to produce community-level geo-spatial mapping of confirmed COVID-19 cases in Ontario Canada in near real-time to support decision-making. This was accomplished by area-to-area geostatistical analysis, space-time integration, and spatial interpolation of COVID-19 positive individuals. Methods: COVID-19 cases and locations were curated for geostatistical analyses from March 2020 through June 2021, corresponding to the first, second, and third waves of infections. Daily cases were aggregated according to designated forward sortation area (FSA), and postal codes (PC) in municipal regions Hamilton, Kitchener/Waterloo, London, Ottawa, Toronto, and Windsor/Essex county. Hotspots were identified with area-to-area tests including Getis-Ord Gi*, Global Moran's I spatial autocorrelation, and Local Moran's I asymmetric clustering and outlier analyses. Case counts were also interpolated across geographic regions by Empirical Bayesian Kriging, which localizes high concentrations of COVID-19 positive tests, independent of FSA or PC boundaries. The Geostatistical Disease Epidemiology Toolbox, which is freely-available software, automates the identification of these regions and produces digital maps for public health professionals to assist in pandemic management of contact tracing and distribution of other resources. Results: This study provided indicators in real-time of likely, community-level disease transmission through innovative geospatial analyses of COVID-19 incidence data. Municipal and provincial results were validated by comparisons with known outbreaks at long-term care and other high density residences and on farms. PC-level analyses revealed hotspots at higher geospatial resolution than public reports of FSAs, and often sooner. Results of different tests and kriging were compared to determine consistency among hotspot assignments. Concurrent or consecutive hotspots in close proximity suggested potential community transmission of COVID-19 from cluster and outlier analysis of neighboring PCs and by kriging. Results were also stratified by population based-categories (sex, age, and presence/absence of comorbidities). Conclusions: Earlier recognition of hotspots could reduce public health burdens of COVID-19 and expedite contact tracing.

Likely community transmission of COVID-19 infections between neighboring, persistent hotspots in Ontario, Canada

Introduction: This study aimed to produce community-level geo-spatial mapping of confirmed COVID-19 cases in Ontario Canada in near real-time to support decision-making. This was accomplished by area-to-area geostatistical analysis, space-time integration, and spatial interpolation of COVID-19 positive individuals. Methods: COVID-19 cases and locations were curated for geostatistical analyses from March 2020 through June 2021, corresponding to the first, second, and third waves of infections. Daily cases were aggregated according to designated forward sortation area (FSA), and postal codes (PC) in municipal regions Hamilton, Kitchener/Waterloo, London, Ottawa, Toronto, and Windsor/Essex county. Hotspots were identified with area-to-area tests including Getis-Ord Gi*, Global Moran’s I spatial autocorrelation, and Local Moran’s I asymmetric clustering and outlier analyses. Case counts were also interpolated across geographic regions by Empirical Bayesian Kriging, which localizes high concentrations of COVID-19 positive tests, independent of FSA or PC boundaries. The Geostatistical Disease Epidemiology Toolbox, which is freely-available software, automates the identification of these regions and produces digital maps for public health professionals to assist in pandemic management of contact tracing and distribution of other resources. Results: This study provided indicators in real-time of likely, community-level disease transmission through innovative geospatial analyses of COVID-19 incidence data. Municipal and provincial results were validated by comparisons with known outbreaks at long-term care and other high density residences and on farms. PC-level analyses revealed hotspots at higher geospatial resolution than public reports of FSAs, and often sooner. Results of different tests and kriging were compared to determine consistency among hotspot assignments. Concurrent or consecutive hotspots in close proximity suggested potential community transmission of COVID-19 from cluster and outlier analysis of neighboring PCs and by kriging. Results were also stratified by population based-categories (sex, age, and presence/absence of comorbidities). Conclusions: Earlier recognition of hotspots could reduce public health burdens of COVID-19 and expedite contact tracing.

Redeploy manure resources to enhance the agro-pastoral cycle

Returning manure to the land is a critical link in the internal cycle of agricultural systems, but excess manure leads to water eutrophication. The traditional manure re-use method brings pathogenic microorganisms, heavy metals, antibiotic resistance genes (ARGs), insect eggs, and other contaminants into the soil, posing a great threat to the ecological environment and human health. Clarifying the spatial distribution patterns of manure nutrient supply and farmland nutrient demand can help guide a more efficient and harmless way to return manure to farmland. This work counted data on cultivation and breeding in 356 cities on the Chinese mainland from 2015 to 2019 and calculated the livestock breeding volume (LB), total environmental capacity (C), and remaining environmental capacity (RC) accordingly. The Spatial Autocorrelation Model (SAC) was used to analyze the distribution patterns of the three. Data results show that China currently has the potential to double LB, but most cities in the west have excess manure due to the mismatched distribution of LB and C. The hot spot analysis results demonstrate the priority/general areas of manure management and the export/import areas of manure resources. The results of the outlier analysis show that some cities located at the boundary of RC Cold/Hot spot areas (e.g., Chengdu City) can perform resource replacement nearby to relieve local environmental pressure. This study analyzes the potential and realistic resistance to utilizing manure as an organic nutrient resource and provides a reference for developing manure management links.

Global influenza surveillance systems to detect the spread of influenza-negative influenza-like illness during the COVID-19 pandemic: Time series outlier analyses from 2015-2020.

BackgroundSurveillance systems are important in detecting changes in disease patterns and can act as early warning systems for emerging disease outbreaks. We hypothesized that analysis of data from existing global influenza surveillance networks early in the COVID-19 pandemic could identify outliers in influenza-negative influenza-like illness (ILI). We used data-driven methods to detect outliers in ILI that preceded the first reported peaks of COVID-19.Methods and findingsWe used data from the World Health Organization’s Global Influenza Surveillance and Response System to evaluate time series outliers in influenza-negative ILI. Using automated autoregressive integrated moving average (ARIMA) time series outlier detection models and baseline influenza-negative ILI training data from 2015–2019, we analyzed 8,792 country-weeks across 28 countries to identify the first week in 2020 with a positive outlier in influenza-negative ILI. We present the difference in weeks between identified outliers and the first reported COVID-19 peaks in these 28 countries with high levels of data completeness for influenza surveillance data and the highest number of reported COVID-19 cases globally in 2020. To account for missing data, we also performed a sensitivity analysis using linear interpolation for missing observations of influenza-negative ILI. In 16 of the 28 countries (57%) included in this study, we identified positive outliers in cases of influenza-negative ILI that predated the first reported COVID-19 peak in each country; the average lag between the first positive ILI outlier and the reported COVID-19 peak was 13.3 weeks (standard deviation 6.8). In our primary analysis, the earliest outliers occurred during the week of January 13, 2020, in Peru, the Philippines, Poland, and Spain. Using linear interpolation for missing data, the earliest outliers were detected during the weeks beginning December 30, 2019, and January 20, 2020, in Poland and Peru, respectively. This contrasts with the reported COVID-19 peaks, which occurred on April 6 in Poland and June 1 in Peru. In many low- and middle-income countries in particular, the lag between detected outliers and COVID-19 peaks exceeded 12 weeks. These outliers may represent undetected spread of SARS-CoV-2, although a limitation of this study is that we could not evaluate SARS-CoV-2 positivity.ConclusionsUsing an automated system of influenza-negative ILI outlier monitoring may have informed countries of the spread of COVID-19 more than 13 weeks before the first reported COVID-19 peaks. This proof-of-concept paper suggests that a system of influenza-negative ILI outlier monitoring could have informed national and global responses to SARS-CoV-2 during the rapid spread of this novel pathogen in early 2020.

A causal relationship between leukocyte telomere length and multiple sclerosis: A Mendelian randomization study.

ObjectivesMultiple sclerosis (MS) is a chronic inflammatory autoimmune and degenerative disorder of the central nervous system. Telomeres are protective structures located at the ends of linear chromosomes, and leukocyte telomere length (LTL) is closely connected with cell aging and senescence. However, the relationship between LTL and the risk of MS remains unknown.MethodsWe performed a two-sample Mendelian randomization (MR) to evaluate whether LTL was causally associated with MS risk.ResultsIn our MR analysis, 12 LTL-related variants were selected as valid instrumental variables, and a causal relationship between LTL and MS was suggested. The risk of MS nearly doubled as the genetically predicted LTL shortened by one standard deviation (SD) under the inverse variance weighted (IVW) fixed effect model (odds ratio (OR) = 2.00, 95% confidence interval (CI): 1.52-2.62, p = 6.01e-07). Similar estimated causal effects were also observed under different MR models. The MR–Egger regression test did not reveal any evidence of directional pleiotropy (intercept = -0.005, stand error (SE) = 0.03, p = 0.87). The Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO) analysis also indicated no directional pleiotropy or outliers for any LTL-related IVs (p-global test = 0.13). In addition, a leave-one-out sensitivity analysis showed similar findings, which further emphasized the validity and stability of the causal relationship.ConclusionsOur results suggest a potential causal effect of LTL on the risk of MS. Genetically predicted shorter LTL could increase the risk of MS in the European population. LTL should be noted and emphasized in the pathogenesis and treatment of MS.

The evolutionary history and mechanistic basis of female ornamentation in a tropical songbird.

Ornamentation, such as the showy plumage of birds, is widespread among female vertebrates, yet the evolutionary pressures shaping female ornamentation remain uncertain. In part this is due to a poor understanding of the mechanistic route to ornamentation in females. To address this issue, we evaluated the evolutionary history of ornament expression in a tropical passerine bird, the White‐shouldered Fairywren, whose females, but not males, strongly vary between populations in occurrence of ornamented black‐and‐white plumage. We first use phylogenomic analysis to demonstrate that female ornamentation is derived and that female ornamentation evolves independently of changes in male plumage. We then use exogenous testosterone in a field experiment to induce partial ornamentation in naturally unornamented females. By sequencing the transcriptome of experimentally induced ornamented and natural feathers, we identify genes expressed during ornament production and evaluate the degree to which female ornamentation in this system is associated with elevated testosterone, as is common in males. We reveal that some ornamentation in females is linked to testosterone and that sexes differ in ornament‐linked gene expression. Lastly, using genomic outlier analysis we identify a candidate melanogenesis gene that lies in a region of high genomic divergence among populations that is also differentially expressed in feather follicles of different female plumages. Taken together, these findings are consistent with sex‐specific selection favoring the evolution of female ornaments and demonstrate a key role for testosterone in generating population divergence in female ornamentation through gene regulation. More broadly, our work highlights similarities and differences in how ornamentation evolves in the sexes.

Automatic clustering-based approach for train wheels condition monitoring

ABSTRACT The main goal of this paper is to present an unsupervised methodology to identify railway wheel flats. This automatic damage identification algorithm is based on the acceleration evaluated on the rails for the passage of traffic loads and deals with the application of a two-step procedure. The first step aims to build a confidence boundary using baseline responses evaluated from the rail, while the second step involves the damages’ classification based on different severity levels. The proposed procedure is based on a machine learning methodology and involves the following steps: (i) data acquisition from sensors, (ii) feature extraction from acquired responses using an AR model, (iii) feature normalization using principal component analysis, (iv) data fusion, and (v) unsupervised feature classification by implementing outlier and cluster analyses. To evaluate whether the number of sensors used to detect and classify wheel flat can be optimized, the influence of sensors’ number is performed.

Short birth interval and associated factors in rural India: A cross-sectional study.

Short Birth Interval (SBI) is one of the main causes of adverse maternal and child health outcomes. A 33-month birth-to-birth interval between two successive live births should be followed to minimize the risk of adverse maternal and child health. This study aimed to examine the prevalence of SBI and the associated factors in rural India. Information on 98,522 rural mothers from the fourth round of National Family Health Survey data was analyzed. Bivariate statistics, logistic regression, Moran's I, and Cluster and Outlier Analysis have been used to assess the prevalence and spatial pattern of SBI in rural India. Results revealed that about half of the mothers in rural India had experienced SBI. Rural Indian mothers whose child was not alive (OR = 1.76, 95% CI = 1.63-1.90), were not using any contraceptive methods (OR = 1.42, 95 % CI = 1.37-1.48) and not breastfeeding (OR = 2.73, 95% CI = 2.50-2.97) were more likely to experience SBI. On the other hand, rural mothers from the middle, richer and richest wealth quintiles (OR = 0.91, 95% CI = 0.86-0.97; OR = 0.84, 95% CI = 0.80-0.92; OR = 0.60, 95% CI = 0.55-0.66) and of age over 30 years (OR = 0.38., 95% CI = 0.36-0.39) were less likely to experience SBI. Analysis of spatial patterns revealed clear east-west differences inthe prevalence ofSBI. There was strong clustering of high values of SBI in most districts across the central, northern, western, and southern regions. The study suggests the need to introduce appropriate interventions and programs focused on reducing the prevalence of SBI in rural India.

Evaluation of Spatial and Temporal Performance of Deep Learning Models for Travel Demand Forecasting: Application to Bike-Sharing Demand Forecasting

Deep learning approaches are widely employed for forecasting short-term travel demand to respond to real-time demand. Although it is critical for demand forecasting to be evenly distributed in the spatial and temporal views to support real-time mobility service operations, in related studies, the predictive performance of models has been evaluated only in terms of aggregated errors. Therefore, the present study was conducted to investigate the distribution of errors to explore spatiotemporal correlations. Six deep learning models with the same architecture, except for the base module, consisting of three stacked layers, were constructed. These models were used to forecast demands for a station-based bike-sharing service in Seoul, South Korea. To attain our goals, global and local Moran’s I of the errors was introduced to evaluate the spatial and temporal performances of the deep learning approaches. The results showed that the model with convolutional long short-term memory layers, which are effective at predicting spatiotemporal data, outperformed the other models in terms of aggregated performance. However, the global Moran’s I of the errors in the model reflects spatial dependency over the regions. This suggests that the best predictive performance of the model does not necessarily imply that it performs well in demand forecasting in all regions. Furthermore, cluster and outlier analyses of the errors indicated that excessive or insufficient predictions were clustered or dispersed throughout the regions. These results can be used to enhance the model by introducing the spatial correlation index into the loss function or by incorporating additional features for handling spatial correlations.

An Outlier Analysis on Multi-Dimensional and Time-Series Data

An Outlier Analysis on Multi-Dimensional and Time-Series Data

Abstract LB059: Subtype classification of pediatric high-grade glioma tumors by comparative transcriptomics

Abstract Pediatric high-grade glioma (pHGG) is a highly malignant and poorly understood cancer driven by diverse genetic and epigenetic mechanisms. Here, we use comparative RNA sequencing, outlier analysis, and spectral clustering approaches to analyze transcriptomic data of 1,543 pediatric brain tumor specimens from the UCSC Treehouse Childhood Cancer Initiative (Treehouse) and Open Pediatric Brain Tumor Atlas (OpenPBTA) to identify subpopulations of pHGG patients with characteristic gene expression profiles. We find that approximately half (45%) of pHGG tumors from OpenPBTA exist in three subgroups defined by high outlier-level expression either of: mitochondrially-encoded 12S and 16S rRNAs; genes enriched in the HSF1-mediated heat shock response and activation pathways; or six C/D box snoRNA (SNORD) genes originating from the paternally-expressed SNORD116 locus involved in Prader-Willi syndrome, a complex neurodevelopmental disorder. Interestingly, the same set of HSF1-dependent pathway genes is also significantly upregulated in a subset (~11%) of pHGG tumors from Treehouse, validating this finding in two independent compendia with different transcript isolation strategies (Treehouse, polyA selection; OpenPBTA, ribodepletion). Our work identifies distinct classes of tumors with outlier-level expression of genes with previously unknown roles in pHGG and provides a framework for subtyping tumors by comparative transcriptomics that is adaptable to any cancer type. We are currently investigating the molecular roles of HSF1-response genes and the imprinted SNORD116 gene cluster in pHGG. Our ongoing research into the biomolecular signatures and mechanisms of the three major tumor classes of pHGG as defined in our study will contribute to a greater understanding of pHGG disease manifestation and progression, and will inform strategies of tailored therapeutic interventions for children with this devastating disease. Citation Format: Gina D. Mawla, A. Geoffrey Lyle, Ellen T. Kephart, Katrina Learned, Holly C. Beale, Joshua E. Goldford, Olena M. Vaske. Subtype classification of pediatric high-grade glioma tumors by comparative transcriptomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB059.