Volume 1 2020
Racial and Gender Disparities in AI-Based Resume Screening: An Audit Study of 12 Corporate Automated Hiring Platforms
Josephine A. Osei; Marcus L. Williams; Tara M. Krishnaswamy
This study investigates racial and gender disparities in callback rates produced by AI-based resume screening systems across 12 corporate automated recruitment platforms within the context of AI ethics and computational social science, an area of growing scientific importance given its implications for AI hiring system auditing requirements, equal employment opportunity enforcement, and fair ML model development standards. Using correspondence audit study submitting 4,800 resumes (400 per platform) with factorial manipulation of applicant name racial/gender signaling while holding all qualifications constant, we examine AI screening model training data encoding historical hiring biases translating into differential callback rates for racially/gender-coded names even with equivalent qualifications in 4,800 resume submissions to 12 AI hiring platforms across 8 industry sectors with 4 name conditions (Black female, Black male, White female, White male) drawn from online resume submission portals of 12 large U.S. corporations using AI-based applicant tracking systems for initial resume screening. Results indicate that AI screening systems produce mean callback rates of 34.2% for White male names versus 21.4% for Black female names (12.8 pp gap), with largest disparities in finance (18.4 pp) and technology (16.2 pp) sectors (p < 0.001), with 12.8 pp gap (34.2% White male vs. 21.4% Black female callback rate) as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to AI ethics and computational social science and carry actionable implications for the design of programs and policies targeting AI hiring system auditing requirements, equal employment opportunity enforcement, and fair ML model development standards.
Read more
P-Wave Amplitude Decay and Magnitude Estimation Uncertainty in Regional Earthquake Early Warning Systems: Analysis of 8,412 Pacific Northwest Events
Rebecca N. Frost; Kenji T. Hayashi; Marco A. D'Amato
This study investigates P-wave amplitude decay characteristics and magnitude estimation uncertainty in the ShakeAlert Pacific Northwest earthquake early warning system using 8,412 catalogued seismic events within the context of seismology and earthquake hazard engineering, an area of growing scientific importance given its implications for ShakeAlert network design optimization, station density gap identification, and alert threshold tuning for Pacific Northwest communities. Using P-wave amplitude-distance attenuation analysis using 1D and 3D regional velocity models, magnitude estimation bias quantification, and warning time versus false alert rate optimization, we examine geometric spreading and anelastic attenuation governing P-wave amplitude decay with distance, introducing magnitude estimation uncertainty that scales with event depth and source-station azimuthal distribution in 8,412 seismic events (M1.5-7.1) from 428 PNSN stations with hypocentral distances 10-400 km during the 2018-2020 operational ShakeAlert period drawn from Pacific Northwest Seismic Network (PNSN) station array across Washington, Oregon, and northern California. Results indicate that magnitude estimation RMSE of 0.42 units for events within 100 km of sufficient station density, degrading to 0.84 units beyond 200 km, with false alert rate of 1.8% achievable at 4.8-second median warning time for M5+ (p < 0.001), with RMSE 0.42 at <100 km vs. 0.84 at >200 km; 4.8s median warning for M5+ as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to seismology and earthquake hazard engineering and carry actionable implications for the design of programs and policies targeting ShakeAlert network design optimization, station density gap identification, and alert threshold tuning for Pacific Northwest communities.
Read more
Spike-Specific CD4+ and CD8+ T Cell Memory Formation and 12-Month Durability After mRNA-1273 Booster Vaccination in Immunocompromised Adults
Sunita R. Agarwal; Patrick M. O'Brien; Chloe E. Beaumont
This study investigates spike-specific CD4+ and CD8+ T lymphocyte memory magnitude, phenotype, and 12-month durability following mRNA-1273 booster vaccination in immunocompromised adults versus healthy controls within the context of vaccinology and clinical immunology, an area of growing scientific importance given its implications for booster vaccination timing recommendations for immunocompromised populations and cellular immunomonitoring strategies for vulnerable vaccine recipients. Using intracellular cytokine staining and activation-induced marker assay measuring spike-specific T cell responses at pre-boost, day 14, day 90, and 12-month timepoints by flow cytometry, we examine mRNA-1273 spike antigen expression driving CD4+ T follicular helper priming of germinal center B cells and CD8+ cytotoxic T cell expansion, with immunosuppression attenuating but not eliminating memory formation in 248 participants: 84 solid organ transplant recipients, 80 hematologic malignancy patients, 48 HIV-infected adults, and 36 healthy controls; all received mRNA-1273 booster drawn from academic transplant, oncology, and HIV clinics at Grandview University Medical Center with PBMC collection and cryopreservation for batch flow cytometry. Results indicate that mRNA-1273 booster generates spike-specific CD4+ responses in 78.6% of immunocompromised participants (versus 97.2% healthy controls) with 12-month response persistence in 62.4% of responders across immunocompromised groups (p < 0.001), with 78.6% CD4+ response rate in immunocompromised vs. 97.2% healthy controls; 62.4% 12-month persistence as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to vaccinology and clinical immunology and carry actionable implications for the design of programs and policies targeting booster vaccination timing recommendations for immunocompromised populations and cellular immunomonitoring strategies for vulnerable vaccine recipients.
Read more
Phase Change Material Integration in Building Envelopes for Passive Thermal Regulation: Energy Savings and Peak Load Reduction Across Hot-Dry, Mixed-Humid, and Cold Climate Zones
Amelia K. Park; Luca M. Rossi; Fatou B. Diallo
This study investigates cooling energy savings, peak load reduction, and thermal comfort improvement from PCM-integrated building envelopes across three U.S. climate zones in EnergyPlus simulation validated by field measurement within the context of building energy engineering and sustainable architecture, an area of growing scientific importance given its implications for PCM product selection guidance by climate zone and building type for code adoption and LEED energy credit pathways. Using EnergyPlus whole-building energy simulation of PCM-integrated envelopes across ASHRAE Climate Zones 2B, 4A, and 6A validated against 12-month field measurement data from 3 instrumented test buildings, we examine PCM latent heat absorption during peak solar gain hours reducing envelope thermal transmittance and shifting heat flux peak 2-4 hours, reducing HVAC peak demand and total cooling energy in 36 building archetype simulations (4 building types x 3 climate zones x 3 PCM configurations) validated with 3 full-scale instrumented buildings over 12 months drawn from ASHRAE Climate Zones 2B (Phoenix, AZ), 4A (Charlotte, NC), and 6A (Minneapolis, MN) EnergyPlus TMY weather files. Results indicate that PCM envelope integration reduces cooling EUI by 18.4% in hot-dry climate (Zone 2B) and 12.4% in mixed-humid (Zone 4A) but only 4.8% in cold climate (Zone 6A), with peak demand reduction of 28.4%, 18.4%, and 8.4% respectively (p < 0.001), with 18.4% cooling EUI reduction in hot-dry; 28.4% peak demand reduction as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to building energy engineering and sustainable architecture and carry actionable implications for the design of programs and policies targeting PCM product selection guidance by climate zone and building type for code adoption and LEED energy credit pathways.
Read more
Eight-Week Mindfulness-Based Stress Reduction and Cognitive Function in Older Adults With Mild Cognitive Impairment: A Randomized Active-Controlled Trial
Margaret A. Sullivan; Deepak R. Menon; Ingrid T. Larsen
This study investigates effects of 8-week Mindfulness-Based Stress Reduction on cognitive function trajectories in older adults with mild cognitive impairment compared to an active health education control within the context of geriatric psychology and cognitive neuroscience, an area of growing scientific importance given its implications for clinical MCI management programs, community brain health initiatives, and cognitive aging intervention development. Using double-blind assessor randomized controlled trial with MBSR (8 weekly 2.5-hour sessions) versus active health education control, measuring neuropsychological battery at baseline, 8 weeks, and 12 months, we examine mindfulness training reducing stress-related cortisol elevation and amygdala hyperactivity, reducing default mode network interference with hippocampal memory encoding and prefrontal executive control in 182 older adults with MCI (91 per arm, mean age 72.4 years, 58% female, mean MoCA 21.8) from two memory clinic sites drawn from community MBSR program delivery in clinic-adjacent meditation rooms and online for remote participants enrolled 2018-2020. Results indicate that MBSR produces 2.4-point greater MoCA improvement at 8 weeks versus control (p<0.001) and 1.8-point advantage maintained at 12 months, with RAVLT delayed recall showing largest between-group effect size (d=0.58) (p < 0.001), with +2.4 MoCA points at 8 weeks; d=0.58 for RAVLT delayed recall as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to geriatric psychology and cognitive neuroscience and carry actionable implications for the design of programs and policies targeting clinical MCI management programs, community brain health initiatives, and cognitive aging intervention development.
Read more
Omniphobic PVDF Hollow Fiber Membrane Surface Engineering for Wetting-Resistant Direct Contact Membrane Distillation of High-Salinity Oilfield Produced Water
Hana M. Kato; Rodrigo E. Salazar; Anna T. Petersen
This study investigates omniphobic surface engineering of PVDF hollow fiber membranes via hierarchical nanoparticle-fluoropolymer coating for wetting-resistant direct contact membrane distillation of oilfield produced water within the context of chemical engineering and membrane separation technology, an area of growing scientific importance given its implications for produced water desalination for oilfield water reuse, zero liquid discharge systems, and hypersaline industrial wastewater treatment. Using contact angle goniometry, liquid entry pressure measurement, and 72-hour DCMD flux stability testing with surfactant-containing high-TDS produced water at 60/20 C feed/permeate, we examine hierarchical re-entrant surface geometry preventing Wenzel-state wetting by reducing solid-liquid contact fraction, with low-surface-energy fluorosilane coating providing thermodynamic wetting resistance to surfactant-laden brines in 28 membrane variants (7 surface treatments x 4 nanoparticle loadings) tested in 72-hour DCMD stability runs against 5 produced water chemistries (TDS 50,000-180,000 mg/L) drawn from bench-scale DCMD test cell with hot/cold recirculation loops at feed temperature 60 C and permeate temperature 20 C. Results indicate that 1.0 wt% SiNP-PFOTS omniphobic membrane maintains 98.4% salt rejection and 24.2 LMH water flux for 72 hours against 120,000 mg/L TDS produced water with 0.1 mM SDS surfactant, versus wetting failure at 8.4 hours for unmodified PVDF (p < 0.001), with 72-hour wetting resistance vs. 8.4-hour failure for unmodified PVDF; 98.4% salt rejection maintained as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to chemical engineering and membrane separation technology and carry actionable implications for the design of programs and policies targeting produced water desalination for oilfield water reuse, zero liquid discharge systems, and hypersaline industrial wastewater treatment.
Read more
Ancient Genomic Evidence for Multiple Migration Waves and Denisovan Admixture in Prehistoric Southeast Asian Island Populations, 3000-500 BP
Simone L. Bachmann; Ricardo A. Campos; Nadia P. Okafor
This study investigates palaeogenomic evidence for migration wave timing, Austronesian expansion routes, and Denisovan introgression signals in prehistoric Southeast Asian island populations from ancient DNA within the context of palaeogenomics and prehistoric population genetics, an area of growing scientific importance given its implications for Austronesian expansion reconstruction, prehistoric Pacific population movement modeling, and Southeast Asian prehistory synthesis. Using shotgun and target-enrichment ancient DNA sequencing with ADMIXTURE, TreeMix, D-statistics, and qpAdm population structure and admixture analysis, we examine multiple Austronesian migration pulses from Taiwan displacing Papuan-related hunter-gatherer populations with variable admixture, retaining Denisovan introgression signatures in proportion to Papuan ancestry in 84 ancient individuals from 24 archaeological sites across Philippines, Sulawesi, Maluku, and Timor-Leste (4-38x coverage, mean 12.4x) drawn from 24 cave and midden archaeological sites across island Southeast Asia with skeletal remains dated 500-3000 BP by AMS radiocarbon. Results indicate that D-statistics confirm Denisovan introgression in 68.4% of individuals with >30% Papuan ancestry (mean D-stat 0.024), with a third pre-Austronesian migration wave identified by qpAdm in 12 Wallacean individuals dated 1800-2400 BP (p < 0.001), with Denisovan signal in 68.4% of high-Papuan-ancestry individuals; third migration wave in 12 Wallacean individuals as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to palaeogenomics and prehistoric population genetics and carry actionable implications for the design of programs and policies targeting Austronesian expansion reconstruction, prehistoric Pacific population movement modeling, and Southeast Asian prehistory synthesis.
Read more
Sparse Window Transformer Attention for 3D Point Cloud Semantic Segmentation: Benchmark Performance on SemanticKITTI and nuScenes LiDAR Datasets
Leon T. Fischer; Ai-Ling M. Wu; Soren N. Eriksson
This study investigates sparse window-based transformer attention for 3D LiDAR point cloud semantic segmentation with benchmark evaluation on SemanticKITTI and nuScenes outdoor driving datasets within the context of 3D computer vision and autonomous driving perception, an area of growing scientific importance given its implications for autonomous vehicle semantic perception, outdoor robotics scene understanding, and HD mapping for self-driving systems. Using sparse voxelized window attention transformer architecture with hierarchical feature aggregation, trained with focal loss on SemanticKITTI and nuScenes segmentation benchmarks, we examine sparse window partitioning reducing transformer quadratic attention complexity to linear in point count, enabling efficient global context with local geometric detail preservation through hierarchical aggregation in SemanticKITTI: 22,000 training scans / 4,071 test scans; nuScenes: 28,130 training / 6,019 test with 16 semantic classes drawn from SemanticKITTI outdoor urban LiDAR benchmark and nuScenes autonomous driving dataset with 20 semantic class annotations. Results indicate that SWTrans achieves 78.4% mIoU on SemanticKITTI test (state-of-art at submission) and 81.2% on nuScenes val, with 2.4x faster inference than PointTransformer via sparse attention (p < 0.001), with 78.4% mIoU SemanticKITTI (vs. 72.0% PointTransformer), 2.4x faster inference as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to 3D computer vision and autonomous driving perception and carry actionable implications for the design of programs and policies targeting autonomous vehicle semantic perception, outdoor robotics scene understanding, and HD mapping for self-driving systems.
Read more
Chronic Atmospheric Nitrogen Deposition Reduces Plant Species Richness and Shifts Community Composition Across a European Temperate Grassland Biodiversity Gradient
Annelise M. Hoffmann; Brendan T. O'Connell; Camille R. Dubois
This study investigates long-term atmospheric nitrogen deposition effects on plant species richness, functional group composition, and trait diversity across a European temperate grassland N-deposition gradient within the context of plant community ecology and environmental biogeochemistry, an area of growing scientific importance given its implications for critical load policy calibration for European N-deposition regulation and grassland biodiversity conservation management. Using paired historical resurvey design comparing 1990s versus 2018-2020 plant community composition against EMEP-modeled N-deposition accumulation, with multivariate ordination and linear mixed models, we examine reactive nitrogen enrichment promoting nitrophilous grass and tall forb dominance that competitively exclude low-stature stress-tolerant specialist species, reducing overall species richness and forb functional diversity in 184 permanent grassland plots resurveyed in 2018-2020 against 1990s baselines across 8 European countries (Germany, Netherlands, UK, Sweden, Denmark, France, Czech Republic, Switzerland) drawn from semi-natural grassland sites with 20-50+ year management continuity across an N-deposition gradient from Atlantic maritime to continental European climate zones. Results indicate that cumulative N-deposition accounts for 48.4% of variance in species richness decline across plots; each additional 100 kg N/ha cumulative deposition is associated with a 2.84 species/m2 loss, with forb richness declining 3.2x faster than grass richness (p < 0.001), with 48.4% variance explained; 2.84 species/m2 per 100 kg N/ha cumulative deposition as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to plant community ecology and environmental biogeochemistry and carry actionable implications for the design of programs and policies targeting critical load policy calibration for European N-deposition regulation and grassland biodiversity conservation management.
Read more
Recreational Screen Time, Sleep Duration, and Executive Function in Preadolescent Children: A Cross-Sectional Study of 3,842 U.S. Children Aged 9-11
Sandra E. Nakamura; Bernard A. Kofi; Lisa M. Reyes
This study investigates associations among daily recreational screen time, objectively measured sleep duration, and neuropsychological executive function in 9-11 year old U.S. children from the ABCD Study within the context of developmental pediatrics and cognitive neuroscience, an area of growing scientific importance given its implications for pediatric screen time guidance development, school sleep health programs, and digital well-being policy for children. Using parent-reported recreational screen time and actigraphy-measured sleep duration linked to NIH Toolbox Cognition Battery executive function composite at Year 2 ABCD follow-up visit, we examine prolonged screen time displacing sleep duration and delaying sleep onset, compounding direct executive function impairment from reduced offline memory consolidation and prefrontal development disruption in 3,842 children aged 9-11 years (mean 9.9 years, 49% female, 52% White, 20% Hispanic, 16% Black) from 21 U.S. ABCD Study sites drawn from 21 U.S. ABCD Study research sites with actigraphy over 3 nights and in-clinic NIH Toolbox Cognition Battery at Year 2 follow-up (2020). Results indicate that screen time >2 hours/day is associated with 0.22 SD lower executive function composite (p<0.001), with sleep duration mediating 38% of the association; combined >2hr screen time AND <9hr sleep associated with 0.48 SD lower scores (p < 0.001), with -0.22 SD executive function per >2hr screen time; 38% mediated through sleep; -0.48 SD in combined risk group as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to developmental pediatrics and cognitive neuroscience and carry actionable implications for the design of programs and policies targeting pediatric screen time guidance development, school sleep health programs, and digital well-being policy for children.
Read more
