Search Results (1,205)

The conventional von Neumann architecture, constrained by the memory and power walls arising from the separation of storage and computation, faces significant limitations in computational efficiency and energy consumption. To address these challenges, this paper proposes a computing-in-memory (CiM) architecture based on a hybrid 2T-2C/4T-2C ferroelectric random-access memory (FRAM) array. The proposed architecture performs majority-based bitwise computation by simultaneously activating multiple word lines, enabling AND and OR operations in conventional 2T-2C FRAM cells. Selectively embedded 4T-2C FRAM cells further provide in-array inversion, extending the supported functions to NOT and functionally complete Boolean logic. The architecture also supports full-adder operations and stores input operands, intermediate data, and output results within the same FRAM subarray, thereby reducing data movement. Moreover, the architecture provides ADC-free bitwise computing with binary inputs and outputs, reducing peripheral-circuit overhead and power consumption. The internal computation, nevertheless, relies on analog charge sharing and differential sense-amplifier resolution. HSPICE simulations indicate PVT-evaluated sensing stability and computational efficiency under the evaluated conditions. The bit-line voltage difference reaches 337 mV under triple-row activation and 214 mV under quintuple-row activation, with the former being 5.2 times that of the reported DRAM implementation used for comparison. At 3.3 V, process–voltage–temperature (PVT) simulations show that the maximum deviation of ΔV from its mean value remains below 4.62% across the evaluated process corners and temperatures from −40 °C to 125 °C. Simulations of the 8 × 8 FRAM CiM compute-array circuit model yield an energy consumption of 1.94–3.46 pJ/bit and a calculation latency of 0.599–1.167 ns for the supported bitwise operations, corresponding to a 4.86×–5.90× reduction in energy consumption compared with the reported DDR3-based design. The architecture also supports parallel processing and mitigates data loss associated with destructive FRAM readout through an in-array replication mechanism. Finally, an 8 × 8 hybrid FRAM CiM prototype was fabricated in a 180 nm CMOS process as a physical implementation of the proposed hybrid architecture, and its basic array functionality was verified. Full article

Objective: To analyze associations between 24-hour movement behaviors (and guideline adherence) and anthropometric and body composition indicators in Ecuadorian preschoolers using data from the International Study of Movement Behaviors in the Early Years (SUNRISE) pilot study. Methods: This secondary analysis utilized data from the primary study conducted in 108 children aged 3–4 years (57.4% girls; 50.9% urban). Anthropometry (weight, height, and body mass index [BMI]) was assessed using standardized protocols during the pilot study, whereas body composition indicators (fat-free mass [FFM], fat mass [FM], and body fat percentage) were estimated using a validated anthropometric prediction equation. Movement behaviors were assessed via triaxial accelerometry (ActiGraph GT3X+ or GT3X-BT, Pensacola, FL, USA) over seven days, complemented by parent-reported questionnaires. Exploratory unadjusted correlation analyses were performed using Spearman’s rho (ρ) for continuous/ordinal variables, and Pearson’s correlations (r), mathematically equivalent to point-biserial correlations (r_pb), for dichotomous–continuous associations. Results: In exploratory unadjusted analyses, device-measured longer sleep duration showed small-to-moderate inverse associations with weight (ρ = −0.33, p = 0.001), BMI (ρ = −0.28, p = 0.006), and FFM (ρ = −0.37, p < 0.001). Higher moderate-to-vigorous physical activity showed moderate positive associations with height (ρ = 0.32, p = 0.002), weight (ρ = 0.40, p < 0.001), and FFM (ρ = 0.44, p < 0.001). In complementary multivariable models, the associations between device-measured sleep or moderate-to-vigorous physical activity (MVPA) and the pre-specified anthropometric and body composition outcomes were attenuated after adjustment for age, sex, area of residence, and, where applicable, height, and were generally not statistically significant. Conclusions: In this exploratory secondary analysis of the SUNRISE Pilot Study Ecuador, unadjusted analyses identified associations between specific 24-hour movement behaviors, particularly sleep duration and MVPA, and anthropometric and body composition indicators. However, these associations were attenuated after covariate adjustment. The findings should therefore be interpreted as hypothesis-generating and require confirmation in larger longitudinal studies. Full article

Background/Objectives: Cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) are commonly used as first-line treatment for hormone receptor-positive, human epidermal growth factor receptor 2 (HER2)-negative metastatic breast cancer. However, the optimal timing of CDK4/6i initiation remains uncertain. This study compared overall survival (OS) between upfront first-line and deferred second-line CDK4/6i use in routine clinical practice. Methods: This retrospective, two-center cohort study included 403 patients with hormone receptor-positive, HER2-negative metastatic breast cancer who received a CDK4/6i between January 2019 and December 2024. Patients were categorized into upfront-use and deferred-use groups according to whether CDK4/6i treatment was initiated in the first or second metastatic treatment line. Overall survival was estimated using the Kaplan–Meier method. An exploratory analysis of progression-free survival (PFS) from CDK4/6i initiation was also performed. Cox proportional hazards regression, inverse probability of treatment weighting (IPTW), and a 12-month landmark analysis were performed. Results: Of 403 patients, 311 received upfront treatment and 92 received deferred treatment. At a median follow-up of 63.0 months, the deferred-use group had higher rates of visceral metastasis and endocrine-resistant disease. Median PFS from CDK4/6i initiation was longer in the upfront-use group than in the deferred-use group (27.0 vs. 14.0 months; log-rank p < 0.001). Median OS was 61.0 months in the upfront-use group and 76.0 months in the deferred-use group, without a statistically significant difference (log-rank p = 0.059). After adjustment for clinical factors and year of metastatic diagnosis, deferred use was not independently associated with OS (adjusted hazard ratio [HR] 1.167, 95% confidence interval [CI] 0.772–1.765, p = 0.463). Similar results were obtained in the IPTW analysis (HR 0.925, 95% CI 0.552–1.549; p = 0.766) and the 12-month landmark analysis. Conclusions: Deferred second-line CDK4/6i use was not independently associated with inferior overall survival compared with upfront use. These findings support an individualized, risk-adapted sequencing approach in carefully selected patients but do not establish equivalence or justify routine deferral. Full article

This paper presents a modified two-degree-of-freedom wave buoy with an embedded tunable magnetic-spring electromagnetic energy harvester. The proposed device is modeled as a branched kinematic chain composed of a rotational–rotational buoy mechanism and a rotational–prismatic harvester branch sharing the first revolute joint. Two harvester orientations are considered and compared. The mathematical model is formulated using homogeneous transformations, velocity Jacobians and Lagrange equations. Particular attention is paid to the structure of the inertia matrix and to the way in which its inverse transmits generalized forces between the rotational coordinates and the translational motion of the moving magnet. The model is implemented in MATLAB/Simulink R2024b and evaluated under free-response, regular-wave, and bidirectional frequency-sweep excitation scenarios. Under regular-wave excitation, Config. 1 produces approximately 19.3 and 2.98 times greater average load power than Config. 2 for moving-assembly masses of 5 g and 268 g, respectively. The frequency-sweep results show that the preferred harvester orientation depends on the excitation frequency and moving-assembly mass. No resolved sweep-direction dependence is observed for 5 g, whereas for 268 g the identified hysteresis intervals are approximately 0.53 rad/s for Config. 2 and 0.64 rad/s for Config. 1. These results provide design guidelines for selecting the harvester orientation and moving mass in compact wave-excited buoy systems. Full article

Background: Long COVID has emerged as a global health challenge characterized by persistent neuropsychiatric symptoms, including depression, fatigue, and cognitive impairment. Growing evidence indicates that sustained neuroinflammation, endothelial dysfunction, and fronto-limbic network disruption may underlie these symptoms, representing pathophysiological features distinct from major depressive disorder (MDD). Although repetitive transcranial magnetic stimulation (rTMS) is an established treatment for MDD, its therapeutic efficacy in Long COVID-associated depression remains uncertain. Methods: Long COVID was defined as laboratory-confirmed SARS-CoV-2 infection followed by persistent neuropsychiatric symptoms lasting ≥3 months, including cognitive impairment (“brain fog”), fatigue, and new-onset depressive symptoms. We conducted a retrospective registry-based cohort study using real-world clinical data from two TMS clinics in Tokyo (May 2022–April 2026). Forty-four medication-free patients with Long COVID-associated MDD were compared with eighty-eight propensity score–matched patients with primary MDD (1:2 nearest-neighbor matching based on age, sex, and baseline Montgomery–Åsberg Depression Rating Scale (MADRS)). All participants received left dorsolateral prefrontal cortex intermittent theta burst stimulation (iTBS). Treatment outcomes were evaluated using MADRS and 17-item Hamilton Depression Rating Scale (HAM-D₁₇) improvement rates, HAM-D₁₇ response, and remission. Statistically adjusted analyses (inverse probability of treatment weighting (IPTW) and doubly robust estimation) were used to reduce measured confounding; however, residual unmeasured confounding cannot be excluded due to the retrospective observational design. Results: Long COVID patients showed significantly attenuated antidepressant response. Improvement rates were markedly lower for MADRS (38.4% vs. 61.5%) and HAM-D₁₇ (36.7% vs. 58.4%). IPTW and doubly robust models demonstrated large adjusted percentage-point differences (MADRS: −23.0 percentage points; HAM-D₁₇: −21.8 percentage points; both p < 0.001). While HAM-D₁₇ response did not differ significantly, remission rates were substantially reduced (27.3% vs. 61.4%). Conclusions: These findings indicate an adjusted association suggesting reduced rTMS responsiveness in Long COVID-associated depression. Given the retrospective observational design and the possibility of residual unmeasured confounding, the results should be interpreted as associations rather than evidence of a causal effect. Full article

This paper presents a robust attitude-tracking control architecture for rigid spacecraft subject to model mismatches and external disturbances. Quaternions are utilized for attitude representation to prevent the gimbal lock associated with Euler angles. While conventional nonlinear dynamic inversion (NDI) relies on Newtonian mechanics and input–output linearization—which inadvertently generates internal zero dynamics and encounters severe control derivative discontinuities at the $q_0=0$ singularity—this study proposes a novel NDI framework derived strictly from Udwadia’s Lagrangian formulation. This approach realizes an exact input-state linearization directly on the 6-degree-of-freedom active holonomic constraint manifold, completely eliminating internal zero dynamics and mathematical singularities. To ensure robustness against physical uncertainties, the singularity-free NDI is augmented with a nonlinear disturbance observer (DOBC) and an outer-loop linear quadratic (LQ) tracking controller. A rigorous composite Lyapunov stability analysis is conducted for the complete closed-loop architecture. The analysis formally guarantees that both the isolated disturbance estimation error and the fully interconnected dual-loop NDI-DOBC system are Uniformly Ultimately Bounded (UUB), even in the presence of realistic, time-varying disturbances with non-vanishing derivatives ($\dot{\mathit{d}}\ne \mathbf{0}$). Comprehensive numerical simulations, parameterized by a physical spherical air-bearing testbed subject to state-dependent gravitational imbalance torques, validate the architecture’s exceptional tracking precision, smooth transient response, and robust disturbance rejection. Full article

The first step in the analysis of the contact mechanics and leakage prediction of metallic seals applied to nuclear reactor pressure vessels is the proper characterization of the surface topography. At present, two approaches are used for this characterization. On the one hand, there are non-parametric techniques such as HPD and PSD, which retain all the characteristics of the surfaces that have been measured, but the results are high dimensional; hence, they cannot be analyzed analytically. The other type is parametric fractal methods, where the parameters used are fractal dimension D and characteristic scale G, where the analytical derivations can be made; however, this leads to systematic deviations in the mechanical response due to some idealized assumptions, like isotropy, Gaussian distribution, and infinite self-similarity. In this article, we propose an equivalent fractal parameter inversion model (EFPIM) that does not rely on geometric fitting; instead, it fits the mechanical contact behavior of a physical surface. This inversion procedure reduces three errors simultaneously. Thus, the EFPIM does not use D and G as the geometrical fitting variables but rather redefines them as mechanically equivalent ones, the purpose of which is to minimize the difference between the W-M fractal surface and the real measured surface. To address the problem of constrained inversion, we adopt a genetic algorithm with BFGS. To prove its effectiveness, we carried out experiments on C-ring seal surfaces and found that the deviation in the contact area was reduced by an order of magnitude in comparison to traditional structure-function extraction, and the deviation in the approach and the maximum pressure were less than 2%. Moreover, the equivalent parameters shift systematically away from their geometric counterparts in the direction that compensates for the dominant non-ideal deficit of the W-M surface; when both parameters are free, the equivalent fractal dimension decreases, while the equivalent characteristic scale increases, compensating for the absent non-Gaussian deep valleys of ideal W-M surfaces. Existing models of analytical contact and leakage may be directly implemented using equivalent parameters and with accuracy comparable to that of FFT-based simulations, with the modest cost of the offline computations. Full article

Soft deformable materials such as elastomers, biological tissues, and polymeric membranes are widely used in modern engineering applications including biomechanics, soft robotics, and flexible electronics. Accurate identification of their constitutive parameters is therefore essential for reliable mechanical modeling and design. Membrane inflation or bulge tests are commonly used for this purpose, where material parameters are typically identified from pressure–deflection measurements. However, such measurements generally require optical systems to capture membrane deformation, which increases experimental complexity. In this work, we propose a surrogate-assisted inverse identification framework for determining hyperelastic material parameters using pressure–volume data obtained from membrane inflation tests, thereby eliminating the need for optical deformation measurements. To reduce the computational cost associated with repeated forward simulations, an Artificial Neural Network (ANN) surrogate model is trained using numerically generated pressure–volume data from finite-element simulations. The trained ANN efficiently predicts the pressure response of the membrane for different material parameters and volume influx values. A Genetic Algorithm (GA) is then employed to identify the optimal parameters by minimizing the discrepancy between measured and predicted responses. The proposed GA–ANN framework is demonstrated for the Mooney–Rivlin hyperelastic model and accurately recovers material parameters for both noise-free and noisy datasets, providing a computationally efficient and robust methodology for the characterization of soft membranes. Full article

A coordinated extracellular matrix (ECM) and cytoskeletal remodeling are critical drivers of hepatocellular carcinoma (HCC) progression, yet the upstream post-transcriptional mechanisms regulating these processes in spontaneous canine HCC remain poorly defined. Previously, we reported significantly downregulated miRNAs and enriched pathways from differentially expressed genes (DEGs) in canine HCC tissues. In the present study, we investigated putative post-transcriptional target interactions between the prioritized miRNAs and enriched ECM–cytoskeleton pathway-associated genes (n = 34). Integrative bioinformatic analysis prioritized ACTG1 and SDC1 as the key putative targets of cfa-miR-10a based on their concordant prediction by four target-prediction algorithms, high transcript abundance, and inverse correlations with cfa-miR-10a-5p. RNAhybrid and miRanda analyses supported favorable predicted interactions between cfa-miR-10a and the 3′UTRs of both transcripts, with RNAhybrid minimum free energy values ≤ −20 kcal/mol and miRanda scores ≥ 140. In the matched canine transcriptomic subset, cfa-miR-10a and its predicted targets showed inverse expression trends. Cross-species analysis using TCGA-LIHC datasets further demonstrated concordant dysregulation of ACTG1, SDC1 and hsa-miR-10a-5p in human HCC, despite their weak or non-concordant correlation with hsa-miR-10a-5p. Collectively, these exploratory findings identify a candidate cfa-miR-10a–ACTG1/SDC1 network associated with coordinated ECM and cytoskeletal transcriptomic alterations in canine HCC. While cross-species analysis supports the conservation of target gene dysregulation, the divergent miRNA-mRNA correlation patterns highlight potential species-specific post-transcriptional co-expression landscapes that warrant future functional validation. Full article

Although papillary thyroid carcinoma (PTC) is widely considered an immunologically cold tumor, detectable T-cell infiltration can influence disease progression in selected patients. Previous bulk analyses have linked cAMP-responsive element modulator (CREM) expression to a favorable progression-free interval and an inverse correlation with regulatory T-cell infiltration but have not resolved the cellular origin or functional state. In this study, we integrate single-cell RNA-sequencing data from a 23-sample PTC atlas with bulk Cancer Genome Atlas Thyroid Carcinoma (TCGA–THCA) validation. We show that CREM marks a T-cell receptor (TCR)-driven immediate early gene activation state in tumor-infiltrating T cells rather than a cAMP program; this state is consistently and modestly enriched within the FOXP3⁺ regulatory T-cell compartment, challenging the interpretation that bulk inverse Treg–CREM correlations reflect Treg-intrinsic repression. CREM-high T cells upregulate ANXA1, and permutation-controlled analysis identifies a candidate paracrine axis from CD4/CD8 T cells to the epithelial epidermal growth factor receptor (EGFR). At the bulk level, CREM expression correlates with reduced proliferation programs and is associated with a prolonged progression-free interval (multivariate HR 0.56, 95% CI 0.31–1.00, p = 0.050), with the effect direction preserved after immune and sex adjustment. Genome-wide differential expression and gene-set enrichment in an independent cohort (GSE193581) confirm that CREM-detected T cells are enriched for a TCR-driven immediate early transcriptional program, supporting the reinterpretation of CREM as a marker of TCR-driven IEG activation. Full article

Visible light communication (VLC) has emerged as a transformative optical wireless technology for sixth-generation (6G) networks, offering license-free spectrum access, inherent electromagnetic-interference immunity, high spatial confinement, and the unique ability to combine high-speed wireless connectivity with solid-state lighting infrastructure. However, the transition from conventional VLC links to practical 6G optical wireless systems requires far more than advanced modulation and signal processing. Future VLC performance will be strongly determined by the co-design of photonic front-ends, including high-speed transmitters, spectrally engineered emitters, reconfigurable optical interfaces, intelligent receivers, and energy-autonomous detection units. This article provides a comprehensive, device-centered review of photonic hardware and artificial intelligence (AI) enablers for next-generation 6G VLC systems. Particular attention is given to micro-LEDs, laser diodes, color-conversion materials, including perovskite quantum dots, advanced photodetectors, imaging receivers, wavelength-shifting fiber receivers, solar-cell-based receivers, optical reconfigurable intelligent surfaces (RISs), metasurfaces, beam-steering components, and optical wireless power transfer. This review discusses how AI can support inverse photonic design, transmitter and receiver calibration, nonlinear impairment mitigation, channel-aware beam control, and energy-aware resource management. Unlike broader VLC surveys that mainly emphasize network architecture, this article provides a device-centered perspective on AI-enabled photonic integration for 6G VLC, supported by a comprehensive survey of recent experimental demonstrations. Key challenges related to bandwidth, optical efficiency, receiver field of view, mobility, safety, standardization, and practical deployment are summarized, followed by a research roadmap for 2025–2032. Full article

A magnetic-levitation direct-drive oil-free scroll compressor (MLDD-OFSC) eliminates the anti-rotation mechanism to achieve oil-free operation. Still, its large-stroke planar motion introduces strong sensor–DOF coupling and air-gap-dependent stiffness variation that degrade fixed-gain PID performance. This paper proposes a control strategy integrating acceleration feedback linearization, gain-scheduled PID, and coordinate decoupling. An inverse electromagnetic force model is derived to compensate for the nonlinear force–air-gap relationship, linearizing the suspension dynamics. A phase-adaptive gain scheduling law is developed, where gains vary with trajectory phase via a cosine-based mapping. A homogeneous transformation matrix decouples raw sensor signals into independent X, Y, and yaw DOFs. Frequency-domain analysis at three air-gap positions confirms closed-loop stability. Simulations show that the proposed acceleration-linearized gain-scheduled PID (AL_GS_PID) outperforms traditional PID and fixed-gain AL_PID in tracking accuracy. Experiments demonstrate progressive improvement across four configurations—decentralized PID, decoupled PID, fixed-gain AL_PID, and AL_GS_PID—with the full scheme reducing peak errors to 0.043 mm in X and 0.04 mm in Y, corresponding to 74.7% and 33.3% reductions over decentralized PID. These results demonstrate that the proposed strategy effectively addresses coupling and stiffness variation in large-stroke maglev systems under no-load and light-load conditions. Full article

Electrocardiogram (ECG) signals are essential for arrhythmia detection; however, they are frequently degraded by noise during acquisition, and their evaluation is vulnerable to data-leakage and patient-overlap issues that can compromise model assessment. Therefore, in this study, we propose an image-encoding-based arrhythmia classifier combined with a morphology-aware denoising autoencoder. We evaluate signals under a corrected, patient-independent protocol in which every model-selection decision was made on a separate validation partition. On a leakage-free test partition, the autoencoder achieved an SNR improvement of 5.63 dB and a correlation coefficient of 0.801, improving R-peak amplitude preservation and redetection accuracy under moderate-to-severe noise while introducing measurable morphology degradation when the input was already lightly contaminated. The proposed model encodes the denoised, beat-centered signal into images through an interleaved-grouping outer product with sorting and flipping, and classifies them with a multi-scale three-dimensional convolutional network. Under this protocol, the proposed model obtained the highest macro-F1 among five image-encoding architectures, but did not outperform four models operating directly on the denoised signal (macro-F1 32.5% versus 37.3–40.0%), indicating that the proposed encoding does not improve overall five-class classification under rigorous inter-patient evaluation. A controlled test showed that the encoding is exactly invariant to global signal-polarity inversion, unlike the sequential models. This targeted invariance, rather than a general accuracy advantage, is the contribution reported here. Full article

Ultra-wideband (UWB) inter-node ranging provides relative-distance constraints for cooperative localization in GNSS-denied UAV swarms, but dense candidate networks can exceed the available ranging slots, communication bandwidth, computation, and energy. This paper proposes a Fisher-information-based cooperative sensor node pre-selection method under heterogeneous ranging noise. All mobile nodes remain in the localization state, while the selected nodes induce the active ranging-link set. Selected-node, induced-link, ranging-slot, and normalized general-resource budgets are represented separately. Using predicted geometry and estimated link-quality weights, a gauge-free normalized Fisher information matrix combines link geometry, link-quality-dependent weights, and topology-induced coupling. A trace-based generalized GDOP (G-GDOP) criterion is optimized by a two-stage greedy heuristic with recursive matrix updates. The experiments show that G-GDOP is a local observability and information-quality metric rather than a direct predictor of topology-level nonlinear recovery error. Within the same topology, normalized local RMSE increased from 0.698 [0.673, 0.752] in the Low G-GDOP group to 1.014 [0.999, 1.068] and 1.980 [1.806, 2.180] in the Medium and High groups. Increasing the selected-node budget from K = 6 to K = 20 reduced median RMSE from 0.550 to 0.148 m while increasing the median induced-link number from 109 to 214. Additional tests covered Gaussian and heterogeneous ranging noise, deterministic NLOS bias, online link-weight errors, and predicted-position uncertainty. Direct Inversion and Woodbury Updating were numerically equivalent within a predefined tolerance in all 24 size–regime combinations, and a Woodbury runtime advantage was supported in 20 conditions. The proposed framework therefore provides an interpretable resource-aware pre-selection module without implying an unconditional real-time guarantee. Full article

Background: Second-line treatment selection in relapsed extensive-stage small-cell lung cancer (ES-SCLC) is commonly guided by the platinum-free interval, but the optimal approach for patients progressing 3–6 months after first-line platinum-based therapy remains uncertain. We compared platinum-containing doublet chemotherapy with single-agent chemotherapy in this clinically ambiguous subgroup. Methods: This multicenter retrospective real-world cohort study included patients with ES-SCLC or recurrent metastatic SCLC after prior limited-stage disease who received first-line platinum-based chemotherapy, achieved disease control, and progressed within a platinum-free interval of 90–180 days. Treatment allocation was at the discretion of the treating physician and was not randomized. The primary endpoint was progression-free survival (PFS); secondary endpoints were overall survival (OS), objective response rate (ORR), disease control rate (DCR), and safety. Because of the non-randomized design, the treatment effect was examined across multiple analytic approaches, including multivariable Cox regression, propensity score adjustment, and stabilized inverse probability of treatment weighting (IPTW). Results: Of 105 patients, 54 received single-agent chemotherapy and 51 received platinum-containing doublet therapy; 39 (37.1%) had received first-line atezolizumab. Baseline characteristics were imbalanced in favor of the doublet group, which had a longer platinum-free interval, fewer pleural metastases, and a higher rate of objective response to first-line therapy. ORR was higher with doublet therapy (31.4% vs. 11.1%, p = 0.016), as was DCR (62.7% vs. 33.3%, p = 0.003). Median PFS was 4.4 months (95% CI 3.4–5.8) with doublet therapy versus 3.1 months (95% CI 2.7–3.7) with single-agent chemotherapy (unadjusted HR 0.53, 95% CI 0.36–0.80; p = 0.002). Median OS was 6.5 months (95% CI 5.4–8.0) versus 5.4 months (95% CI 4.1–6.0), a difference that was not statistically significant (HR 0.68, 95% CI 0.46–1.01; p = 0.054). The PFS estimate favored doublet therapy in all sensitivity analyses but was attenuated with increasingly complete adjustment for treatment selection (multivariable HR 0.46, 95% CI 0.29–0.72; propensity-adjusted HR 0.58, 95% CI 0.38–0.88; IPTW HR 0.68, 95% CI 0.38–1.20, p = 0.184). No OS estimate reached statistical significance in any model. Grade ≥ 3 adverse events were similar between groups (66.7% vs. 64.8%, p = 0.842). Conclusions: In this non-randomized cohort of patients relapsing within a 90–180-day platinum-free interval, platinum-containing doublet chemotherapy was associated with higher response rates and longer PFS, without an increase in severe toxicity, but no overall survival benefit was demonstrated. Because baseline prognostic factors consistently favored the doublet group and the PFS advantage was attenuated after propensity-based adjustment, these findings should be regarded as hypothesis-generating. They are nonetheless consistent with randomized data showing improved PFS but not OS with platinum rechallenge, and support platinum-containing rechallenge as a reasonable option in carefully selected patients rather than as a demonstrated survival-prolonging strategy. Full article