Volume 1 2022

This study investigates scaling behavior of linear population decoders for motor cortex neural spike data as a function of simultaneously recorded neuron count and training set size within the context of computational neuroscience and neural data analysis, an area of growing scientific importance given its implications for brain-computer interface design principles and optimal electrode array sizing for motor prosthetics. Using linear discriminant analysis and partial least squares decoders trained on binned spike count population vectors with leave-one-out cross-validation, we examine neural population activity dimensionality increasing with recorded unit count, enabling linear decoder accuracy gains that plateau per a log-linear model in recordings from 6 macaque M1/PMd sites (64-188 simultaneously recorded units per session, 2,400 trial epochs total) drawn from chronic multi-electrode array recordings in macaque motor cortex during center-out reaching task performance. Results indicate that linear decoder accuracy for 8-direction reaching discrimination scales from 68.4% at n=10 neurons to 94.2% at n=180 neurons, following a diminishing-returns curve well approximated by a log-linear model (r2=0.97) (p < 0.001), with 94.2% decoding accuracy at n=180 neurons as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to computational neuroscience and neural data analysis and carry actionable implications for the design of programs and policies targeting brain-computer interface design principles and optimal electrode array sizing for motor prosthetics.

This study investigates gravimetric and volumetric capacitance enhancement in Ti3C2Tx MXene electrodes via molecular spacer intercalation and surface termination ratio control within the context of electrochemical energy storage and 2D materials science, an area of growing scientific importance given its implications for high-power-density storage for electric vehicles, portable electronics, and grid ultracapacitor banks. Using cyclic voltammetry, galvanostatic charge-discharge, and electrochemical impedance spectroscopy in 1M H2SO4 electrolyte under N2 atmosphere, we examine intercalated spacer molecules expanding interlayer d-spacing, facilitating proton access to electroactive Ti surface sites and increasing effective double-layer capacitance in 28 MXene electrode variants across 5 spacer molecule types and 3 surface termination conditions tested at scan rates 5-500 mV/s drawn from nitrogen-atmosphere electrochemical cell at room temperature with 1M H2SO4 as electrolyte. Results indicate that DMSO-intercalated, fluorine-reduced Ti3C2Tx achieves gravimetric capacitance of 612 F/g at 5 mV/s with 84% retention at 100 mV/s and 91% capacity after 10,000 cycles (p < 0.001), with 612 F/g capacitance, 54.2% improvement over unmodified MXene as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to electrochemical energy storage and 2D materials science and carry actionable implications for the design of programs and policies targeting high-power-density storage for electric vehicles, portable electronics, and grid ultracapacitor banks.

This study investigates quantifying the relationship between urban green infrastructure coverage and land surface temperature heat island intensity across 48 U.S. metropolitan areas within the context of urban ecology and environmental remote sensing, an area of growing scientific importance given its implications for urban heat action planning, tree canopy equity programs, and climate resilience infrastructure investment. Using Landsat-8 thermal infrared land surface temperature retrieval cross-referenced with high-resolution land cover classification for urban green fraction estimation, we examine urban green cover reducing heat island intensity via evapotranspiration cooling, albedo modification, and surface shading in 48 metropolitan statistical areas analyzed across 8 Landsat scenes (2018-2021) covering summer peak-heat periods drawn from summer daytime land surface temperature measurements across 48 U.S. metropolitan statistical areas. Results indicate that each 10 percentage point increase in metropolitan green cover fraction is associated with a 0.84 C reduction in heat island intensity, accounting for 52% of variance in cross-city heat island magnitude (p < 0.001), with 0.84 C heat island reduction per 10% green cover increase, 52% variance explained as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to urban ecology and environmental remote sensing and carry actionable implications for the design of programs and policies targeting urban heat action planning, tree canopy equity programs, and climate resilience infrastructure investment.

This study investigates hierarchical model-based reinforcement learning for dexterous 7-DoF robotic arm manipulation under sparse binary success rewards within the context of robot learning and deep reinforcement learning, an area of growing scientific importance given its implications for autonomous warehouse logistics, assistive robotics for elderly care, and precision industrial assembly. Using hierarchical policy with high-level goal-conditioned world model planning and low-level motor primitive execution trained with hindsight experience replay, we examine learned world model enabling multi-step lookahead planning at subgoal level, propagating sparse rewards through dense simulated reward landscape for efficient credit assignment in training over 5 manipulation task categories with 1M simulation steps per task; 200 physical robot trials on Franka Panda for validation drawn from PyBullet physics simulation with Franka Emika Panda physical robot validation in structured manipulation environment. Results indicate that H-MBRL achieves 88.4% mean task success across 5 benchmarks at 1M steps, requiring 3.2x fewer environment steps than the best model-free baseline to reach 70% success (p < 0.001), with 88.4% task success at 1M steps, 3.2x sample efficiency improvement as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to robot learning and deep reinforcement learning and carry actionable implications for the design of programs and policies targeting autonomous warehouse logistics, assistive robotics for elderly care, and precision industrial assembly.

This study investigates predictors of session adherence and dropout in telehealth-delivered CBT for major depressive disorder across three delivery platform types within the context of digital psychiatry and implementation science, an area of growing scientific importance given its implications for telehealth mental health program design, platform selection for diverse populations, and targeted dropout prevention. Using prospective cohort study tracking session completion, PHQ-9 scores, and platform engagement metrics over a 12-week CBT protocol with logistic regression dropout prediction, we examine platform usability, videoconference therapeutic alliance quality, and socioeconomic technology barriers mediating adherence to digital mental health treatment in 824 adults (mean age 38.2 years, 64% female) enrolled in telehealth CBT across 3 platforms over 18 months drawn from primary care practices with integrated behavioral health in three metropolitan health systems. Results indicate that overall CBT completion was 62.4% with significantly higher completion on blended platforms (74.2%) versus asynchronous-only (51.4%); PHQ-9 improvement was 8.4 points in completers versus 3.2 points in dropouts (p < 0.001), with 62.4% overall completion, 74.2% in blended delivery as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to digital psychiatry and implementation science and carry actionable implications for the design of programs and policies targeting telehealth mental health program design, platform selection for diverse populations, and targeted dropout prevention.

This study investigates composition-dependent electrocatalytic CO2 reduction to formate over copper-bismuth bimetallic nanoparticle catalysts with in-situ X-ray absorption spectroscopy active-site characterization within the context of electrocatalysis and inorganic chemistry, an area of growing scientific importance given its implications for industrial CO2 utilization via electrochemical reduction to formic acid for hydrogen storage and green chemical synthesis. Using linear sweep voltammetry and chronoamperometry in H-cell at -0.8 to -1.4 V vs. RHE with GC product quantification and in-situ XAS characterization, we examine Bi3+ active sites stabilizing COOH* intermediate favoring formate selectivity over competing H2 and CO pathways at optimal Cu:Bi compositions in 18 CuxBi(1-x) compositions at 5 applied potentials (-0.8 to -1.4 V vs. RHE) with 24-hour stability chronoamperometry on optimal composition drawn from H-cell electrochemical measurements under CO2-saturated 0.5M KHCO3 at room temperature with synchrotron in-situ XAS at beamline 7-3. Results indicate that Cu0.2Bi0.8 composition achieves peak formate Faradaic efficiency of 91.4% at -1.0 V vs. RHE with partial current density of 18.4 mA/cm2, substantially outperforming both pure metal endpoints (p < 0.001), with 91.4% Faradaic efficiency for formate at -1.0 V vs. RHE as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to electrocatalysis and inorganic chemistry and carry actionable implications for the design of programs and policies targeting industrial CO2 utilization via electrochemical reduction to formic acid for hydrogen storage and green chemical synthesis.

This study investigates tighter PAC-Bayes generalization bounds for multi-layer neural networks via data-driven perturbation-aware Gaussian prior distributions centered at sharpness-aware minima within the context of statistical learning theory and deep learning mathematics, an area of growing scientific importance given its implications for neural network certification in safety-critical deployment and understanding implicit regularization in overparameterized deep learning. Using stochastic weight perturbation-based PAC-Bayes bound computation with learned prior distributions aligned to loss-landscape curvature at convergence, we examine perturbation-aware prior aligned to loss curvature landscape enabling posterior concentration near flat minima, reducing the KL divergence term and producing tighter non-vacuous bounds in experiments over 5 dataset-architecture pairs with 20 random initialization seeds each; bound computed after full training with SGD plus cosine LR schedule drawn from standardized training on each dataset with explicit non-vacuous PAC-Bayes bound computation post-training on held-out validation sets. Results indicate that the perturbation-aware prior bound is 1.84x tighter than the best prior PAC-Bayes bound on CIFAR-10 with ResNet-18 (18.4% vs. 33.8%) while remaining non-vacuous and computationally tractable (p < 0.001), with 18.4% generalization bound vs 33.8% best prior bound on CIFAR-10 (1.84x improvement) as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to statistical learning theory and deep learning mathematics and carry actionable implications for the design of programs and policies targeting neural network certification in safety-critical deployment and understanding implicit regularization in overparameterized deep learning.

This study investigates the role of transnational scientific networks and shared atmospheric fallout data in enabling negotiation of the 1963 Partial Test Ban Treaty within the context of history of science and Cold War diplomatic history, an area of growing scientific importance given its implications for contemporary science diplomacy for climate change, pandemic preparedness, and emerging technology governance frameworks. Using archival analysis of State Department cables, Atomic Energy Commission records, Pugwash Conference proceedings, and declassified intelligence estimates from the National Archives and Eisenhower Presidential Library, we examine shared atmospheric fallout measurement data creating a common scientific framework transcending ideological division, enabling treaty verification proposals acceptable to both superpowers in 428 archival documents from 6 collections, 14 Pugwash Conference proceedings, and 8,217 atmospheric test monitoring records (1954-1963) drawn from National Archives (Washington DC), Eisenhower Presidential Library, and Pugwash Conference digital archives. Results indicate that Pugwash Conference scientific exchanges directly informed 14 of 21 PTBT technical verification provisions, with atmospheric fallout consensus data cited in 8 of 12 key U.S.-Soviet diplomatic notes from 1961-1963 (p = 0.003), with 14 of 21 PTBT technical provisions traceable to Pugwash epistemic community input as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to history of science and Cold War diplomatic history and carry actionable implications for the design of programs and policies targeting contemporary science diplomacy for climate change, pandemic preparedness, and emerging technology governance frameworks.

This study investigates Batrachochytrium dendrobatidis prevalence, host community transmission dynamics, and long-term amphibian species richness decline across five global tropical biodiversity hotspots within the context of disease ecology and conservation biology, an area of growing scientific importance given its implications for amphibian conservation management, Bd biocontrol prioritization, and captive assurance colony design for critically endangered species. Using systematic amphibian visual encounter surveys with Bd qPCR swab prevalence testing and 10-year resurvey comparison against historical baseline species richness data, we examine Bd cutaneous chytridiomycosis disrupting amphibian osmoregulation and electrolyte balance, with community-level transmission maintained by tolerant reservoir species even after vulnerable species collapse in 3,284 individual amphibians swabbed from 124 survey sites across 5 biodiversity hotspot regions, resurveyed from historical baselines established 8-22 years prior drawn from highland tropical sites (800-3,200 m elevation) in 5 biodiversity hotspot regions with documented historical Bd invasion events. Results indicate that mean Bd prevalence of 38.4% is significantly associated with amphibian species richness decline of 41.8% relative to pre-invasion baselines, with greatest declines at mid-elevation sites (1,400-2,200 m) where Bd thermal optimum coincides with peak species richness (p < 0.001), with 41.8% species richness decline relative to pre-invasion baseline as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to disease ecology and conservation biology and carry actionable implications for the design of programs and policies targeting amphibian conservation management, Bd biocontrol prioritization, and captive assurance colony design for critically endangered species.

This study investigates longitudinal association between residential fine particulate matter (PM2.5) exposure and spirometric lung function trajectory in school-age children over 5 years within the context of environmental epidemiology and pediatric pulmonology, an area of growing scientific importance given its implications for pediatric air quality standard development, asthma school intervention targeting, and housing equity policy for air quality. Using annual spirometry (FEV1, FVC, FEV1/FVC) over 5 years combined with residential address-linked EPA monitoring and land-use regression model PM2.5 assignment, we examine PM2.5-induced pulmonary inflammation and oxidative stress suppressing alveolar development and accelerating lung function growth trajectory deficit in developing airways in 2,184 children from 64 elementary schools in 8 California air basin zones followed from 2017-2022 with annual spirometry drawn from school-based spirometry clinics with residential PM2.5 assigned from nearest EPA monitoring station augmented by land-use regression. Results indicate that each 10 ug/m3 increase in annual mean PM2.5 is associated with a 3.2% lower FEV1 % predicted at 5-year follow-up, with effects twice as large in children with prevalent asthma (adjusted beta -6.4%, 95% CI -9.2 to -3.6) (p < 0.001), with 3.2% lower FEV1 per 10 ug/m3 PM2.5 increase; 6.4% in children with asthma as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to environmental epidemiology and pediatric pulmonology and carry actionable implications for the design of programs and policies targeting pediatric air quality standard development, asthma school intervention targeting, and housing equity policy for air quality.