Abstract: The Changdu Basin in eastern Tibet was an important habitat for Jurassic dinosaur faunas, yet reports of dinosaur remains are scarce. Research on the composition of dinosaur assemblages, palaeoenvironments, and palaeobiogeography remains insufficient. This paper reports a dinosaur tracksite from the Middle Jurassic Dabuka Formation in Duolongwa Village, Chaya County, and discusses ichnomorphology, the trackmakers, their gait and speed, and sedimentary facies. Two tracklayers are observed. Theropod tracks from the DLW1 tracklayer are assigned to Eubrontidae ichnogen. et ichnosp. indet. (Morphotype-1). The trackmakers had hip heights of 0.98-1.77 m and body lengths of 2.56-4.67 m and primarily moved at a walking gait. Gasosaurus constructus is the most probable trackmaker. Sauropod tracks from the DLW2 tracklayer are assigned to Brontopodus isp. (Morphotype-2). The trackmakers had an estimated hip height of 2.96 m and a body length of 10.94 m. They were likely juveniles belonging to the Shunosaurus-Mamenchisaurus fauna. Sedimentological analyses indicate that the Duolongwa tracksite was a river point-bar-floodplain depositional system. The discovery of the Duolongwa tracksite demonstrates close links between the Middle Jurassic dinosaur fauna of eastern Tibet and cotemporaneous dinosaur faunas of the Sichuan-Yunnan areas, providing new evidence for faunal exchange among dinosaur assemblages across Southwest China.

Abstract: Martínez Alberga and Roitzheim [6] classified model structures on the five-point pentagon and proved their Bousfield reachability. We describe every model on \( H(p,q) \) by two weak-block partitions and two acyclic-fibration threshold vectors. Two endpoint conditions complete the classification and give explicit counts. We extend the classification to every nonempty finite horizontal sum \( H(p_1,...,p_k) \) by imposing the endpoint conditions on each pair of branches. Every classified model is reachable. Two cuts remove the interactions at the shared endpoints. The remaining model is constructed branch by branch; at most two further set-localizations restore the target. Classification and reachability therefore give complementary descriptions: which models exist, and how to construct them.

Abstract: We extend the finite-scale radial monotonicity result for the Poincaré ball to three settings: the inward radial direction, the tangential direction, and the case of a variable Euclidean step \(\varepsilon(r)\). In the inward case we show that the hyperbolic step \(\delta_-(r,\varepsilon)\) is strictly increasing in \(r\) on the domain \(r>2\mathrm{arctanh}(\varepsilon)\). In the tangential case, interpreting the hyperbolic step as the hyperbolic distance to the displaced point, we prove strict monotonicity of \(\delta_\perp(r,\varepsilon)\) on the domain \(a^2+\varepsilon^2<1\), where \(a=\tanh(r/2)\). We then establish a sphere-averaged analogue \(\bar{\delta}(r,\varepsilon)\) on the restricted domain \(\varepsilon<a<1-\varepsilon\) (which requires \(\varepsilon<1/2\)), where pointwise monotonicity holds for every direction of the displacement, and a volume-based characterization \(\delta_{\mathrm{vol}}(r,\varepsilon)\) on the full outward domain \(0\leq r<2\mathrm{arctanh}(1-\varepsilon)\). Finally, for a variable step \(\varepsilon(r)\) we derive a necessary and sufficient criterion for monotonicity of \(\delta_+(r,\varepsilon(r))\), show that it holds automatically whenever \(\varepsilon'(r)\geq0\), and exhibit an explicit counterexample of the form \(\varepsilon(r)=c(1-\|z\|)\) with \(c\in(0,1)\) for which monotonicity fails. The extension of the sphere-averaged result to the full domain \(a+\varepsilon<1\) remains open.

Abstract: Telecom taught the world how to run heterogeneous generations in parallel for decades (1G–5G and beyond). This paper formalizes that analogy for artificial intelligence (AI), where rule‑based, statistical/ML, deep learning, foundation models, and agentic systems will coexist for the foreseeable future. We make five contributions: (1) a formalized mapping between coexistence phenomena in mobile networks and AI multi‑version ecosystems; (2) a layered architecture for version‑aware AI orchestration with capability negotiation and sandboxed execution; (3) two analytical risk models quantifying interoperability and security exposure under coexistence; (4) a structured threat analysis with attack paths unique to cross‑model and cross‑agent amplification; and (5) a quantitative evaluation protocol encompassing semantic drift, safety policy violation rates, and change‑failure risk during phased upgrades. We detail engineering and governance controls that mirror proven telecom practices (multi‑RAT, phased migration, telemetry‑driven operations) which can help multi variant AI systems.

Abstract: The goal of this work is simple and traditional: to provide a formalization and generalization of the notion of entropy in the spirit of Carnot. The tools we employ to achieve our objective are quite novel however. We have no need for state spaces, paths in state spaces, path integration,cycles in state spaces, systems in equilibrium, quasi-static transformations, the usual dynamical notion of reversibility, or even, the first law of thermodynamics. Instead, we rely on a primitive notion of process and the effects a process has on a system, typically, on the surroundings of a system. Carnot cycles are replaced in our framework by bithermal processes. We recover that the efficiency of a bithermal process is maximal when the process is reversible and maximal efficiency is determined by the temperature of the reservoirs. We also derive that all cyclic processes are dissipative, at best isentropic. The formalism assigns a degree of irreversibility to every process. The argument we present is quite general. It does not assume systems to be at equilibrium; it simply requires systems to have a state at the beginning and at the end of the process.

Abstract: GENERAL CONTROL NON-DEREPRESSIBLE 5 (GCN5) is a histone acetyltransferase that plays a central role in regulating plant development and responses to environmental stimuli. In root development, GCN5 promotes the maintenance of the stem cell niche and proliferation of transit-amplifying cells. SPATULA (SPT) is a basic helix-loop-helix (bHLH) transcription factor with established roles in gynoecium and carpel development, as well as broader functions in plant growth. In roots, SPT negatively regulates quiescent center size, root meristem size, and overall root growth, without substantially affecting root patterning or cell-type differentiation. To investigate a potential genetic interaction between SPT and GCN5 in root meristem development, we generated and characterized spt;gcn5 double mutants. Notably, loss of SPT substantially suppressed some developmental defects associated with gcn5 mutations, including abnormalities in root meristem size, meristem organization, and root growth. These data suggest that a complex genetic interaction between SPT and GCN5 regulates important aspects of root meristem growth and organization.

Abstract: Background: Surgical oncology education covers a broad range of topics including general cancer knowledge, multidisciplinary clinical reasoning, operative skills, communication, survivorship, and career development. The literature is scattered across educational levels and disciplines, including undergraduate, postgraduate, surgical, medical oncology, interprofessional, and continuing education. This review provides an overview of how the field is structured, highlights key outcomes, and identifies gaps that should inform future research. Methods: An integrative narrative review with scoping features was conducted based on an extensive search of oncology education literature, complemented by targeted supplementary searches. These included undergraduate oncology, surgical oncology training, tumor board and multidisciplinary education, communication and ethics, survivorship, patient-partnered learning, career development, and competency frameworks. The review included peer-reviewed conceptual papers, curriculum frameworks, reviews, and empirical studies involving medical students, residents, fellows, practicing clinicians, or interprofessional learning transferable to other settings. The studies were organized by learner level, cancer site, educational approach, and outcome type. Due to significant variability in interventions, populations, and measures, no combined effect estimate was calculated. Results: The field comprises three interconnected layers: core cancer skills for future generalists; multidimensional reasoning involving surgery, systemic therapy, radiotherapy, pathology, imaging, palliation, and patient preferences; and specialized surgical practice, including technical skills, disease-site expertise, supervision, autonomy, research, and leadership. Undergraduate education typically focuses on fundamental principles and common cancers, while postgraduate surgical training centers on specific disease sites, procedures, and levels of responsibility. Interactive learning methods such as case discussions, tumor-board simulations, standardized patients, patient stories, mentorship, and technical simulations tend to boost immediate knowledge and confidence. However, evidence linking these methods to long-term knowledge retention, clinical performance, operative independence, and patient outcomes is limited. The literature on communication is most advanced in concept and measurement, whereas areas like survivorship, interprofessional performance, leadership, health advocacy, and teacher training are less developed. Conclusions: Surgical oncology education is better characterized by competencies across the entire cancer pathway than by a single specialty or method. Future research should connect curriculum exposure to outcomes such as objective reasoning, communication, technical skills, teamwork, and patient care. It should also assess delayed transfer and evaluate whether existing frameworks apply across health systems and cultural contexts.

Abstract: Facial expression recognition (FER) systems often explain predictions with saliency maps that neither identify specific facial behaviour nor indicate how a decision could change. We present AU-GRACE, which couples an action-unit (AU) relation graph with an explicit AU bottleneck to provide counterfactual recourse as small sets of plausible AU edits. Facial Action Coding System (FACS)/EMFACS knowledge supplies reliability-gated soft constraints, allowing departures from prior edges weakly supported by data. The emotion head accesses the image through the AU vector and a norm-penalised residual, so AU edits intervene on the decision. An amortised solver generates edits in 6.4 ms. AU-GRACE achieves 96.78%, 69.19% and 94.92% accuracy on RAF-DB, AffectNet-7 and FERPlus, respectively, under the evaluation protocols defined by the dataset authors. Counterfactual validity is 94.8%, with 2.3 edited AUs and 86.4% plausibility. Agreement with six certified FACS coders reaches κ¯=0.71, close to their inter-coder agreement. The edits remain effective after rendering and re-classification by an independent model. These results show that competitive recognition accuracy can be combined with AU-level recourse through an explicit AU bottleneck.

Abstract: This study examines time-varying return connectedness among WTI crude oil, gold, the U.S. Dollar Index (DXY), and the Stock Exchange of Thailand (SET). A discount-factor TVP-VAR with separately estimated stochastic-volatility paths is applied to 4,206 synchronized daily observations from February 2008 to December 2025. Generalized forecast-error variance decompositions, directional connectedness, generalized impulse responses, prespecified crisis comparisons, and sensitivity checks are employed. Full-sample connectedness is low (mean TCI = 1.8004), with own-return shares of 97.7067%-98.6715%. During the cleaned Global Financial Crisis (GFC) window, mean TCI rises to 2.2968 versus 1.6315 outside the GFC; however, the lower median and insignificant rank-based tests indicate that the increase is concentrated in the upper tail. The clearest GFC feature is a gold-centered configuration: Gold-to-Oil transmission increases by 1.3723 percentage points, while Oil's net position remains close to zero and specification-sensitive. Gold is the strongest net transmitter and DXY the strongest receiver during the GFC, but their roles change across COVID-19 and Russia-Ukraine regimes. Impulse responses attenuate within approximately two days. The findings show that energy-financial transmission is weak on average, short-lived, and crisis-dependent, supporting conditional rather than structural interpretations of cross-market risk.

Abstract:

Three-dimensional (3D)-printed clear aligners represent a new alternative for treating orthodontic malocclusions. However, their biocompatibility remains a critical factor requiring careful evaluation. In the current study, the in vitro cytotoxicity of three innovative 3D-printed resins (Tera Harz TC-85 DAC, TA-28, and TR-07, Graphy, Seoul, Korea) and two conventional thermoformed materials (Smart Track, Invisalign, Align Technology, USA, and TruGEN, Spark, Ormco, USA) was assessed by measuring the cell viability of human primary gingival fibroblasts (HGFs) exposed to specimens via direct contact. The MTT assay revealed that, at 7 days, 3D-printed resins exhibited a statistically significant reduction in HGF metabolic activity compared to both the control group and the thermoformed groups (p < 0.0001). This was categorized as "slight cytotoxicity" according to the Ahrari et al. classification. A similar overall biological trend was observed at 14 days, with no progressive impairment in HGF viability, indicating that the materials maintain a stable in vitro biocompatibility profile over time. Notably, all tested materials maintained cell viability levels above the 70% threshold, in compliance with the International Organization for Standardization (ISO) 10993-5 standards. While thermoformed materials are currently considered the conventional reference in orthodontics, the tested 3D-printed resins revealed no relevant in vitro cytotoxic risk, aligning with their ongoing integration into clinical orthodontic practice.

Abstract: Safety training can teach an AI to make fewer mistakes. It can also teach it to hide themistakes it still makes. Both can produce better test scores. This paper explains how thathappens, examines experiments that have produced it, and asks what follows for the risk ofa major accident. A passed test is reassuring when the test had a good chance of catchingthe problem. If that chance is small, even many passes may tell us little. A second problemarises when safety work removes common, small failures more easily than rare, severe ones.The failures left over can become worse on average while the total number falls. These effectscan develop while everyone involved is trying to make the system safer. Competition canthen encourage firms and governments to give AI more responsibility before they understandthe remaining danger. My forecast is that enforceable limits on the development of the mostcapable AI systems will follow a major accident. The practical response is to check whattests can catch, preserve ways of observing failures, and limit what a system can damagewhile uncertainty remains.

Abstract: Before/after studies of planned special events routinely report elevated emissions and attribute the increase to event-induced congestion. We show that this attribution can be wrong, and we give a cheap test that separates the two causes. Using one-minute detector data from LATTICE, an operational freeway analytics platform deployed on the Las Vegas network, we analyze the 2026 Electric Daisy Carnival on Interstate 15 across both travel directions. Northbound, the arrival direction, loses 9.5 mph during a five-hour evening window. Southbound, over the same days and the same weather, loses nothing. Yet CO₂ rises by a nearly identical amount in both directions (+11.6% vs. +11.8%), which means the headline emissions increase is driven by weekend travel demand, not by the congestion. Treating the uncongested direction as a control, a difference-in-differences estimate isolates the congestion-specific component. Only PM_2.5 retains a substantial residual (+8.9 percentage points); CO₂ and NO_x retain essentially none. The residuals order monotonically by each pollutant’s measured low-speed emission sensitivity (Spearman ρ = 1.00, n = 4), a sensitivity we recover independently from the same detector stream. The practical consequence is that event-emissions claims based on a single direction should be treated as demand estimates unless a control corridor is included.

Abstract: The E80 is an open-source, 8-bit, single-cycle soft processor written in VHSIC Hardware Description Language (VHDL) from gates, multiplexers, and flip-flops, without an arithmetic package. Its 39 instructions were derived from 371 program samples in eight sources of introductory computer architecture material. They cover almost 80% of the operations recorded, averaged per source. A toolchain takes an assembly program to a simulation waveform with one keystroke. Presets cover three low-cost field-programmable gate array (FPGA) boards, one on a fully open-source flow. This extended version removes a false combinational loop from the datapath, so the design passes timing analysis on all three flows without workarounds. A serial light-emitting diode (LED) matrix mirrors the simulation signals on the board. Execution can be slowed or stepped one instruction at a time. A self-checking program decides the conformance of every instruction from the final program counter alone. The largest program in the surveyed material, a recursive maze search written for a 16-bit machine, was ported without abridgement into 208 of the 255 memory words. The effect on learning remains unmeasured.

Abstract: While the existential math for non-smooth weak solutions in ideal fluid bounds has historically been demonstrated via chaotic convex integration loops, an explicit, constructive analytical closed-form derivation governing singular hypersonic compressible magnetohydrodynamic (MHD) shear-strip boundary layers has been entirely absent from the scientific literature. We resolve this profound deficit by constructing a definitive, fully explicit analytical closed-form architecture that stabilizes coupled multi-field transport gradients. By mapping the coupled density-weighted velocity and magnetic induction fields through a gauge-covariant intrinsic tensor Friedrichs mollifier matted with a multi-dimensional singular Riesz transform vector mapping , the high-frequency ultraviolet loop interactions are structurally neutralized. This process forces the spontaneous, inner generation of a fourth-order biharmonic hydro-magnetic dissipative spectral cage , avoiding any ad-hoc parameters or empirical truncations. Under the absolute topological control of the Aubin-Lions Compactness Embedding Lemma, the continuous functional scaffold contracts uniformly to zero (Lc → 0), establishing strong Cauchy convergence toward rough local solutions of the classical compressible MHD field equations. We prove that at this sharp local limit, the induced biharmonic dissipation matrix collapses into a stable, non-vanishing strictly positive local metric invariant ( > 0), analytically isolating the anomalous dissipation profile at the exact Onsager Hölder regularity exponent threshold α = 1/3. Furthermore, the predictive validity of this explicit framework is directly verified against empirical astronomical data from the Event Horizon Telescope (EHT) and the Chandra X-ray Observatory for the relativistic accretion disk of M87*, yielding a 100% deterministic predictive matching with a zero boundary layer error profile.

Abstract: General-purpose audio foundation models (FMs) — wav2vec 2.0, HuBERT, WavLM, AST, CLAP — have transformed speech and environmental-sound processing, but their value for underwater acoustic target recognition (UATR) is unknown: UATR operates on radiated vessel noise far outside the pretraining distribution of any of these models, and our companion work has shown that even models trained on underwater audio transfer catastrophically across water areas [2]. We benchmark six public FMs (wav2vec2-base, wav2vec2-large, HuBERT-base, WavLM-base+, AST, CLAP-HTSAT) as frozen feature extractors on all four UATR corpora of our series (Oceanship, QiandaoEar22, VTUAD, ShipsEar; 258,893 segments), under two label views (per-corpus full-label and a shared two-class view) and two regimes: within-corpus linear probing and cross-corpus zero-shot probe transfer (72 ordered pairs). Findings: (1) Within each corpus, linear probes on frozen FM embeddings beat the majority-class baseline on macro-F1 on every model–corpus pair of the full-label view (on raw accuracy, four of six models sit below the corrected VTUAD majority rate — an imbalance artifact we flag); the accuracy-strongest full-label FM on QiandaoEar22 and ShipsEar (CLAP) matches trained log-mel ResNet-18 probes within corpora in accuracy but trails them in macro-F1 on three of four corpora, and trails fine-tuned ResNet-18 (e.g., QiandaoEar22 macro-F1 0.258 vs 0.292 for the fine-tuned reference of [4], §4.1), with environmental-audio pretraining (AST, CLAP) beating speech pretraining in 6 of 8 full-label cells. (2) Model scale does not help: wav2vec2-large never beats wav2vec2-base on both metrics of any corpus — macro-F1 degrades in 3 of 4 corpora (accuracy in 2 of 4). (3) Cross-corpus transfer is where FMs pay off: under the series protocol and against all available trained references, the best frozen-FM probe — selected per direction on the primary metric adopted for this benchmark, macro-F1 (chosen because of the documented class imbalance, §4.1–4.2), as a descriptive oracle upper envelope — beats the trained ResNet-18′s zero-shot macro-F1 on 11 of 12 anchored directions (point estimates against the tested anchors under the fixed probe pipeline; in 8 directions the paired-bootstrap 95% CI excludes zero for all three anchor seeds). On Oceanship→QiandaoEar22, a linear probe on frozen WavLM embeddings reaches macro-F1 0.736, numerically exceeding the available ResNet-18 target-oracle reference (0.551) of our trained-model UDA benchmark [1] under the series protocol; a 5,000-run permutation control (the true score sits at the 96.3rd–97.4th percentile of random readouts) and comparability analyses indicate that this exceptional direction should be interpreted cautiously rather than as a controlled superiority claim (§4.3, §4.5). When the FM is instead chosen by source-validation performance — the deployable selection rule — the transferred probe still beats the trained source_only baseline in 9 of 12 directions under the same fixed probe pipeline. (4) All FM results are deterministic (frozen pipeline, seed 42); we therefore report exact values without seed variance and attach paired-bootstrap intervals to all twelve cross-corpus comparisons, with a recording-cluster-level bootstrap as a dependence check (§4.3, §4.5). Frozen audio-FM embeddings are not a replacement for in-domain training; within the literature we surveyed, no earlier study has reported a four-corpus, full 12-direction frozen-embedding cross-water benchmark under a shared label-mapping protocol, and within the companion series frozen FM probes are the first representation family shown to transfer across water areas without any adaptation.

Abstract: This study examines the first seventeen years of Imam Mohammad ibn Saud's rule in Diriyah, from his accession in 1727 CE (1139 AH) to his alliance with Sheikh Mohammad ibn Abd al-Wahhab in 1744 CE. Long treated as a prelude to that alliance, the period is analyzed here as the founding phase of the First Saudi State, as formally recognized by the 2022 Royal Order establishing Founding Day. Combining the analytical-historical method with a structural approach, the study applies a three-level critical framework (external criticism, internal criticism, and contextual analysis) to the Najdi chronicles, a hostile contemporary narrative, and more than twenty regional, archaeological, numismatic, and memory studies published between 2021 and 2026. The results show that Diriyah expanded little in territory during these years yet built a coherent political order under severe structural constraint. That order rested on four elements: a defensive balance toward Banu Khalid and other stronger neighbors; durable local alliances; the careful management of settled–Bedouin relations within a subsistence oasis economy; and a pre-ideological discourse of lineage, security, and justice that converted practical performance into symbolic capital. This strategy of latency kept the ruling house cohesive, preserved the emirate's autonomy, and gave Diriyah the stability that later allowed it to adopt and sustain a broader religious and political project. The findings support the historiographical dating of the Saudi state to 1727, show that the accumulation of legitimacy preceded the growth of coercive and fiscal capacity, and identify the founding phase as the institutional and symbolic reference point of the Kingdom's contemporary national narrative and of Saudi Vision 2030.

Abstract: Executive functions and socio-emotional competencies play an important role in emerging literacy and numeracy. Hence, it is important to consider how functioning in the cognitive domain co-occurs with socio-emotional functioning and how these two domains of early childhood development are associated with emerging literacy and numeracy. At the same time, there are few cross-national person-centred studies that address how individual differences in executive functions co-occur with individual differences in socio-emotional competencies. Drawing on the data of 18,875 children (mean age = 5.52 years; 49.30% female) who participated in the International Early Learning and Child Well-being Study 2025, the current study examines latent profiles of executive and socio-emotional competencies in early childhood and their associa-tion with emerging numeracy and literacy. Latent profile analyses yielded five distinct profiles of early childhood executive functions and socio-emotional competencies. The first profile was characterised by globally low functioning (12.9%). The second profile comprised children with comparatively low social-behavioural functioning (24%), whereas the third reflected broadly average, balanced functioning (28.5%). The fourth included children with pronounced executive functioning and emotional strengths but comparatively weaker social and behavioural functioning (20.4%). The fifth profile comprised children who exhibited profound social and behavioural strengths (14.3%). Significant cross-jurisdictional differences were found in profile membership. These differences may reflect broader contextual, cultural, or educational conditions, alt-hough the mechanisms underlying them require further investigation. Significant dif-ferences in emerging literacy and numeracy were also found across profiles, with the children in the fourth profile exhibiting the best numeracy and literacy competencies. The findings suggest that executive and socio-emotional strengths do not necessarily align in early childhood, and that their distinct configurations are associated with dif-ferences in emerging literacy and numeracy.

Abstract: AlphaFold 3 (AF3) has transformed structural bioinformatics through generative diffusion modeling of biomolecular assemblies. However, intrinsically disordered regions (IDRs)—which comprise over 30% of the human proteome and mediate essential cell signaling, oncogenesis, and phase separation—violate the classic thermodynamic energyminimization assumptions inherent to static structure prediction. In the absence of explicit solvent and physiological binding partners, AF3's diffusion trunk exhibits a pronounced compaction bias, collapsing unconstrained polypeptide chains into artificial, rigid secondary structures (α-helices) with spuriously elevated confidence (pLDDT ≥ 70–90).Here, we present AF-RECAL, an automated multi-modal machine learning meta-evaluator designed to audit and recalibrate AlphaFold 3 confidence metrics. Evaluating 53,035 residues across 118 full-length human disease proteins against experimental DisProt ground truth, we observe that AF3 assigns confident structure (pLDDT ≥ 70) to 29.86% of all verified disordered residues (2,649 / 8,871), consistent with independent literature reporting structural mismatches in 22% of IDR residues at pLDDT ≥ 80 (23.47% in our cohort). AF-RECAL addresses these errors by cross-examining three complementary information streams: (1) AF3 internal spatial geometry via Predicted Aligned Error (PAE) multi-scale matrices and a local-to-long-range contact density ratio; (2) evolutionary sequence representations and masked-languagemodel Shannon entropy from Meta's ESM-2 (650M parameter) protein language model; and (3) windowed biophysical polymer chemistry.Under strict ≤ 30% sequence identity clustering (114 independent clusters, eliminating homology leakage across isoforms and paralogs), AF-RECAL achieves an out-of-fold AUROC of 0.7721 and a Precision-Recall AUC (PR-AUC) of 0.4063, improving under full Leave-One-Cluster-Out (LOCO) cross-validation across all 114 clusters to 0.7888 AUROC and 0.4205 PR-AUC (mean per-cluster PR-AUC: 0.5131 ± 0.3437). Adding 3D spatial geometry to an identical ESM-2 650M + chemistry baseline yields an empirical +27.6% relative gain in PR-AUC (0.3183 → 0.4063, reaching0.4205 under LOCO-CV); under cluster-level paired bootstrap testing across 114 clusters, this difference exhibits p = 0.1380 (95% CI: [-0.0246, 0.2083]), establishing a strong positive empirical trend. AF-RECAL achieves an uncorrected nominal PR-AUC gain over metapredict v3 (0.3014, p = 0.0520) and native AF3 (0.2889, p = 0.0320). A verified 100iteration Y-randomization permutation test confirms biological learning over chance (z = +75.11σ AUROC, z = +130.83σ PR-AUC, p = 0.0000), and probability reliability analysis demonstrates a Brier calibration score of 0.1653, outperforming metapredict v3 (0.2253). Across operational high-confidence regimes (70.6% of residues), AF-RECAL achieves 84.58% accuracy, providing structural biologists a practical, post-hoc quality filter to flag spurious structural features before downstream virtual screening and experimental validation.

Abstract: Hybrid renewable energy systems (HRESs) are an effective tool for addressing rural electrification challenges in sub-Saharan Africa (SSA). However, standalone renewable configurations often suffer from intermittency and limited dispatchability, particularly in remote areas with restricted grid access. This study examines the integration of a bi-ogas generator into a PV/wind turbine/battery hybrid system for the electrification of Dala Zoulgo, a remote village in Far North Cameroon, to meet residential, community, and agricultural electricity needs while minimizing cost of energy, net present cost, and CO2 emissions. HOMER Pro was used to simulate and techno-economically compare multiple configurations, with and without biogas integration. The optimal architecture combined a 65.4 kW of photovoltaic array, one wind turbine, a 75-kW biogas generator, and 50 battery storage units under a load-following dispatch strategy, with a net present cost and levelized cost of energy of USD 1.17M and USD 0.340/kWh, respectively, out-performing configurations without biogas or with different renewable combinations. Sensitivity analysis showed biogas integration remains optimal across most of the bio-mass price-availability range tested, while wind speed is the most influential resource parameter on system cost. These results demonstrate that locally available biogas re-sources can improve the techno-economic and socio-environmental viability of hybrid renewable energy systems, offering a replicable pathway for rural electrification in similar off-grid contexts across SSA.

Abstract: Background/Objectives: The aim of this study was to compare the outcomes of pulpotomy done in primary molars under different treatment settings, namely: Basic Behavior Guidance (BBG), Oral Sedation (OS) and General Anesthesia (GA). Methods: In this is retrospective study, dental records of 79 pediatric dental patients treated in university dental hospital between March 2019 to January 2022 were reviewed. Data collected include: (1) age at the time of receiving the treatment, (2) age at the recall visit, (3) gender, and (4) type of clinical setting when providing the dental treatment (BBG, OS, or GA). A total of 154 primary molars received pulpotomy procedure. Recall visits include clinical and radiographic examination. Results: Clinically, six teeth in the BBG group exfoliated. SSC was lost in all groups, but no significant difference was found (P= 0.236). For the percussion and palpation responses, no significant differences were found between the three groups (p= 0.245 and 0.395 respectively). Radiographic findings showed no significant difference between the three groups when roots and pulp canals and bone resorption were examined (p= 0.313 and 0.193 respectively). Signiant differences were found when dentinal bridge formation was assessed (p=0.003). When the quality of the pulpotomy, success of the SSC, and the success of the pulpotomy were compared, no significant differences were found (P>0.05). Conclusions: No significant differences were found between the three groups when the pulpotomy outcomes were assessed. This retrospective study highlights the effect of treatment settings. Future prospective study is recommended.