The bacterial 16S rRNA gene is widely used to characterize host-associated and environmental microbiomes, most commonly through sequencing short hypervariable regions. Recent improvements in PacBio sequencing chemistry and concatenation approaches can now enable high-throughput, full-length 16S rRNA gene sequencing with high accuracy and depth. However, errors introduced during library preparation remain a major limitation, particularly during PCR amplification of full-length amplicons, where error accumulation may be elevated due to longer sequence lengths. These challenges are amplified when samples vary widely in microbial biomass, making it difficult to select a single optimal number of PCR cycles. Here, we evaluated PCR cycle autonormalization for PacBio Kinnex full-length 16S rRNA gene sequencing across seven agriculturally relevant specimen types. We compared conventional fixed-cycle PCR protocols (20, 24, and 30 cycles) with an autonormalization approach in which individual reactions were terminated during exponential amplification based on real-time fluorescence thresholds. Under the workflow tested here, autonormalized libraries generally retained a high proportion of sequences following denoising and chimera removal, exhibited low residual error rates (<0.005%), and yielded relatively even read distributions across heterogeneous sample inputs. Overamplified reactions (30 cycles) showed elevated residual error rates and greater sequence loss, particularly in samples with higher microbial biodiversity, whereas low-cycle libraries produced more variable read output among specimens. Importantly, the PCR protocol had relatively minor effects on overall community composition compared with specimen type. These results support PCR cycle autonormalization as a useful workflow strategy for heterogeneous full-length 16S library preparation, while also highlighting the importance of library design, pooling strategy, and downstream processing in shaping technical outcomes.IMPORTANCEAmplicon-based sequencing of the 16S rRNA gene is a foundational tool in microbiome research, yet PCR amplification remains a major source of library-preparation error. This challenge is magnified for full-length 16S rRNA sequencing and for workflows that process specimen types with widely varying microbial biomass. Selecting a single PCR cycle number can underamplify low-biomass samples or overamplify high-titer samples, increasing artifacts and sequence loss during downstream processing. Here, we show that PCR cycle autonormalization can be integrated into a PacBio full-length 16S rRNA workflow and, under the conditions tested, provides low residual error rates and relatively even sample representation across heterogeneous inputs. Autonormalization also enables blind pooling of amplicons without post-PCR quantification or equimolar normalization, reducing hands-on time and sample loss. These benefits make cycle autonormalization particularly valuable for high-throughput and production-scale library preparation applications handling diverse specimen types.
The development of rapid, on-site analytical methods and test strips for alkaline phosphatase activity (APA) is crucial for addressing drinking water safety crises, as it enables eutrophication assessment and algal bloom warnings under phosphorus-limited conditions. In this study, we incorporate aggregation-induced emission luminogen (AIEgen) guest TPE-4PA into bimetallic lanthanide-based infinite coordination polymers (Tb/Eu-GMP ICPs). TPE-4PA exhibits unprecedented binding preference for Tb3+ over Eu3+, triggering coordination-induced emission (CIE) at 450 nm and modulating the antenna effect (AE) and the tandem energy transfer (TER) from Tb3+ to Eu3+. Subsequently, acetylacetone (acac) was further introduced to yield TPE-4PA@Tb/Eu-GMP-acac ICPs with an appropriate full-color fluoresce response capability. The addition of alkaline phosphatase (ALP) hydrolyzes GMP and TPE-4PA dual substrates, resulting in reduced characteristic emission of the bimetallic Ln-ICP host at 493, 548, 593, 619 and 703 nm, increased monomer emission of TPE-4PA at 392 nm, as well as the solvent-driven AIE of the enzymatic product TPE-4OH at 430 nm, which constitutes a novel ratiometric ALP sensing mechanism with pale-purple to blue fluorescence color shifts. Featuring high sensitivity, doubly assured selectivity, and operational simplicity, this "all-in-one" probe enables real-time APA monitoring via dual-substrate kinetics and is suitable for on-site ALP test strip development. Moreover, the unique color transition point of the dual-substrate-based TPE-4PA@Tb/Eu-GMP-acac ICPs probe can be designed to signal an APA surge prior to algal bloom onset, which holds significant potential for the development of microalgae pollution early-warning systems, thereby safeguarding drinking water safety.
A robust strategy was developed to customize the hydrogel's composition, concentration, and crosslink density, thereby providing a method for the screening and optimization of skin repair hydrogels. Specifically, methyltetrazine-modified collagen (Col-T), norbornene-modified RGD peptide, and norbornene-modified hyaluronic acid with varying degrees of modification (HA-Nlow, HA-Nmed, and HA-Nhigh), were synthesized. Upon mixing Col-T, RGD-N, and one of the HA-N derivatives, a bioorthogonal reaction was immediately initiated, thereby forming an in situ crosslinked, shape-adaptable hydrogel. An extracellular matrix-mimetic hydrogel composed of collagen, RGD peptide, and hyaluronic acid was optimized using human epidermal stem cells (hEpdSCs), human dermal fibroblasts (HDFs), and human umbilical vein endothelial cells (HUVECs). The optimized hydrogel effectively promoted the hEpdSCs proliferation, the proliferation and migration of HDFs, and the migration and tubular formation of HUVECs. In comparison, GelMA exhibited significant cytotoxicity against hEpdSCs due to the use of photoinitiator LAP. The hydrogel exhibited anti-hemolytic, pro-coagulant, and tissue-adhesive properties, and significantly accelerated the healing of 15 mm × 15 mm full-thickness wound after a single application without any additives. This hydrogel was associated with enhanced hair follicle-like structure formation and reduced inflammation-related responses in the wound area. Furthermore, its ready-to-use and biodegradable nature made it highly suitable for clinical applications.
暂无摘要（点击查看详情）
暂无摘要（点击查看详情）
Performing ex situ normothermic machine perfusion (NMP) for ≥24 h represents an opportunity to evaluate and treat livers, but is limited by the lack of support from relevant extrahepatic organs. Incorporation of systems of renal replacement therapy, including hemodiafiltration (HDF), appears useful in this regard during prolonged ex situ liver NMP. This study aimed to demonstrate the impact and benefits associated with incorporation of continuous HDF during 24-h ex situ NMP in a relevant preclinical model, including transplantation and post-transplant follow-up. Porcine livers (n = 28) underwent 24-h ex situ NMP with either partial perfusate exchange at 12 h and no HDF (NHDF, n = 11) or HDF initiated 2 h after NMP start (n = 17). Biochemical, histological, endothelial, and metabolomic parameters were assessed. A subset of grafts undergoing NMP + HDF (n = 8) were transplanted into recipients. Incorporation of HDF during NMP maintained stable pH and electrolyte levels, effectively preventing hypernatremia, hypochloremia, and hypocalcemia developing without HDF. HDF cleared metabolic wastes (e.g. urea) and inflammatory cytokines (IL-1B, IL-2, IL-6, IL-8, and IL-18), resulting in reduced injury and oxidative stress markers after 24 h (Suzuki score 0.8 ± 0.4 HDF vs. 2.7 ± 1.3 NHDF, p <0.001). Vasoprotective endothelial response mechanisms, including KLF2 and eNOS gene and protein expression, were upregulated, whereas stellate cell activation and sinusoidal contraction were reduced among HDF-treated grafts. HDF reduced metabolomic alterations arising in livers during 24-h NMP, and adequate graft maintenance using NMP + HDF was demonstrated by full functional and metabolic recovery during post-transplant follow-up. Continuous HDF promotes a more physiological biochemical and metabolic environment, reduces inflammation and oxidative stress, and preserves homeostatic endothelial response mechanisms in livers undergoing 24-h ex situ NMP, facilitating successful transplantation in a complex preclinical model. In this preclinical study, livers were normothermically perfused for 24 h ex situ, both with and without continuous HDF. Incorporation of HDF offered relevant improvements in numerous on-device measures, including the maintenance of physiological biochemical parameters; removal of injurious metabolic wastes; and improvement of injury and stress responses in parenchymal and nonparenchymal cells. A subset of livers were successfully transplanted and demonstrated full functional and metabolic recovery during follow-up. These findings indicate that advanced renal replacement therapies, such as HDF, are a key aspect of improving and prolonging ex situ normothermic liver perfusion, although there is ongoing need to develop more physiological metabolic support protocols for livers while on such devices.
This study describes the heterologous expression and stability profiling of a manganese superoxide dismutase from Thermus aquaticus (TaqMn-SOD). Codon-optimized TaqMn-SOD fused with a C-terminal 6His tag was expressed in Escherichia coli. The apoenzyme (ApoTaqMn-SOD) was successfully recovered and purified under optimized induction conditions followed by post-lysis thermal treatment (85 °C for 90 min), affording a yield of >100 mg of purified protein per liter of culture. The purified TaqMn-SOD exhibits a specific activity of about 3,000 U/mg and exceptional stability profiles: retaining full activity at 85 °C for 4 h; maintaining > 56% activity across a broad pH range of 3.0-12.0; tolerating ethanol concentrations up to 40%, preserving > 80% activity after 48 h and 50-80% activity after 90 d in 5-20% ethanol. Moreover, its activity was not significantly affected by various divalent metal ions (excepting Fe2+), organic solvents, detergents, denaturants, or the chelating agent EDTA, with n-hexane enhancing activity by 47%. The enzyme was partially inactivated (47%) in simulated gastric fluid for 10 min, but remained stable in simulated intestinal fluid and water. These results suggest that TaqMn-SOD holds potential for applications in high-temperature food processing, thermally-stable cosmetic manufacturing, agricultural protection, and pharmaceutical formulations requiring stability under extreme conditions.
Accurate decoding of motor intent from biosignals is an important step toward intuitive upper-limb prosthetic-control interfaces. We propose a novel high-dimensional multimodal deep learning framework that fuses surface electromyography (sEMG) and B-mode ultrasound (US) images to estimate metacarpophalangeal and proximal interphalangeal joint angles continuously. The framework employs a shared Encoder-Decoder-Regression architecture integrating convolutional neural networks (CNNs), transposed convolutions, an action-conditioned multi-head cross-attention module (ATT) that uses the commanded action as a query, and long short-term memory (LSTM) layers to jointly capture spatiotemporal features from both modalities. To improve cross-subject generalization and reduce data requirements for new users, we introduce a transfer learning strategy with parameter freezing. Experiments on data from seven able-bodied subjects show that, compared with sEMG-only and US-only baselines, the fusion model reduces test local root mean square error (RMSE) by 2.187° (23.385%) and 0.890° (11.054%), and increases test local correlation (Pearson's r) by 0.069 (10.02%) and 0.039 (5.48%) (p < 0.05), supporting the potential of multimodal fusion for future prosthetic-control interfaces. A preliminary validation on one amputee participant further supports the feasibility of applying the framework under a residual-limb sensing condition. Ablation studies further confirm that the full CNN+LSTM+ATT model achieves the best performance, reducing test local RMSE by 0.933° (11.524%) and increasing test local correlation by 0.033 (4.56%) (p < 0.05). Furthermore, fine-tuning the pretrained model with only 25% of a new subject's data yields performance comparable to full retraining, highlighting the framework's data efficiency.
Understanding disease activity and therapy goals of patients with rheumatoid arthritis (RA) is important to provide patient-centered specialty pharmacy care and achieve positive therapy outcomes. The Routine Assessment of Patient Index Data 3 (RAPID3) is a validated metric to track patient-reported RA disease activity. Consistent and frequent monitoring of RAPID3 scores may provide clinically meaningful insights. To describe the trajectory of RAPID3 scores over time and factors related to changes in RAPID3 scores among patients new to specialty medications. Adult patients with RA who initiated specialty medications were included if they filled medications at least 3 times and had at least 2 RAPID3 scores documented in 6 months. Linear mixed effects regression models estimated trends in RAPID3 scores over time with respect to patient identifiers. Logistic regression models adjusted for baseline RAPID3 severity and estimated adjusted odds ratios (aOR) and associated 95% CIs of achieving a minimal clinically important improvement (≥3.8 points) or low severity/near remission (LS/NR) RAPID3 severity score at 6 months, by patient characteristics. Of 312 patients with RA, 195 (63%) had baseline high severity RAPID3 scores. Patients with baseline high severity achieved significant RAPID3 decreases over 6 months (5.9 points; P < 0.0001). Nearly half of patients (N = 147, 47%) had a clinically important improvement and 105 (34%) were LS/NR at follow-up; 67 patients (21%) met both criteria. After adjusting for baseline RAPID3 severity category, medication change (aOR, 0.39; 95% CI, 0.21-0.71) and opting out of full therapy management (aOR, 0.36; 95% CI, 0.19-0.65) were inversely associated with achieving a clinically important improvement. Being LS/NR at follow-up was also inversely associated with medication change (aOR, 0.19; 95% CI, 0.075-0.48) and opting out of full therapy management (aOR, 0.27; 95% CI, 0.12-0.63). Low medication adherence (aOR, 0.28; 95% CI, 0.11-0.72) and Medicaid insurance (aOR, 0.52; 95% CI, 0.29-0.93) were also inversely associated with being LS/NR at follow-up. RAPID3 is a valuable tool in the specialty pharmacy setting, helping assess patient outcomes beyond traditional metrics. Specialty pharmacies should support their patients in striving to reach clinically important improvement and/or achieving LS/NR as measures of quality pharmacy care during the crucial months after therapy initiation and throughout care. Active collaboration between patients and pharmacists is crucial in reaching positive RA disease activity outcomes. To support this, implementing programs that foster patient engagement with pharmacists may be important for optimizing care delivery.
A 32-year-old woman with panhypopituitarism following surgical resection, chemotherapy, and radiotherapy for a pituitary germinoma underwent assisted reproductive treatment. Management involved comprehensive hormone replacement therapy, ovulation induction with exogenous gonadotropins, in vitro fertilization (IVF), and thromboprophylaxis with low molecular weight heparin due to a diagnosed thrombophilia. A fresh embryo transfer resulted in a successful singleton pregnancy. The pregnancy progressed uneventfully and reached full term and a healthy neonate was delivered. This case highlights the feasibility of pregnancy in women with complex endocrine and thrombotic conditions and adds to the limited literature supporting IVF in women with significant hypothalamic-pituitary axis compromise.
Deployment complexity and specialized hardware requirements hinder the adoption of deep learning models in neuroimaging. We present MindGrab, a lightweight, fully convolutional model for volumetric skull stripping across the evaluated imaging modalities. MindGrab's architecture is designed from first principles using a spectral interpretation of dilated convolutions, and demonstrates state-of-the-art performance on the tested benchmarks (mean Dice score across datasets and modalities: 95.9 ± 1.6), with up to 40-fold speedups and substantially lower memory demands compared to established methods. Its minimal footprint allows for fast, full-volume processing in resource-constrained environments, including direct in-browser execution. MindGrab is delivered via the BrainChop platform as both a simple command-line tool (pip install brainchop) and a zero-installation web application (brainchop.org). By removing traditional deployment barriers without sacrificing accuracy, MindGrab makes state-of-the-art neuroimaging analysis broadly accessible.
Here, we present a protocol to acquire high resolution, extended depth of field images of insect specimens by photographic focus stacking using a modular digital imaging system. The method provides a standardized workflow linking equipment assembly, calibration, image acquisition, and post processing. Using a full frame mirrorless camera (61 MP) coupled to microscope objectives and synchronized strobe illumination, the protocol achieves pixel scales from 0.76 m-0.19 m and produces artifact free composites through sub-micron focus increments (0.2 m). The procedure can capture and process approximately 20 final images per week under routine laboratory conditions. Compared with existing stacking solutions, this low-cost hybrid setup (< 30% of the cost of commercial systems) maximizes accessibility while maintaining diffraction limited image quality. Representative applications include the production of color calibrated identification plates for taxonomy, biodiversity digitization, and outreach. The protocol's standardized structure facilitates reproducibility across laboratories and field stations, supporting large scale insect imaging campaigns in both resource limited and institutional environments.
Behavior guidance is recommended as a core component of pediatric dentistry, yet the effectiveness of specific non-pharmacological techniques for autistic patients has not been systematically quantified. The researchers systematically searched PubMed, Embase, Web of Science, Cochrane Library, Scopus, APA PsycInfo, CINAHL, and AMED for randomized controlled trials of non-pharmacological behavior guidance techniques in autistic children undergoing dental procedures. Eligible studies enrolled children with a clinical diagnosis of autism spectrum disorder and compared a structured behavior guidance strategy with routine behavior management or another guidance technique, and reported anxiety and/or cooperation outcomes using validated measures. Electronic searches identified 1,242 records; after deduplication, 487 titles and abstracts were screened, 9 full texts were assessed, and 5 trials (n = 445; sample size 19-162) were included. All were conducted in specialist pediatric dental services and focused on non- or minimally invasive procedures, including examination, prophylaxis, and fluoride application. Interventions comprised a multisensory sensory-adapted dental environment, visual pedagogy, electronic media-based guidance (video modeling and video goggles), and immersive virtual reality. Owing to heterogeneity, quantitative synthesis was feasible for only two VR trials that reported Frankl Behavior Rating Scale scores. In the larger parallel-group RCT, VR improved cooperation versus conventional care (mean difference [MD] 0.55, 95% CI 0.23-0.88); the smaller crossover study showed a similar direction (MD 0.39, 95% CI -0.35 to 1.13). Pooled in a fixed-effect model, VR was associated with a moderate improvement of about half a Frankl point (pooled MD 0.52, 95% CI 0.22-0.82). Non-pharmacological behavior guidance techniques may improve clinically relevant outcomes in autistic children during routine dental care. However, pooled quantitative evidence in the present review was available only for virtual reality-related improvements in cooperation, whereas evidence for other techniques, including sensory-adapted environments, remained limited and was based primarily on narrative synthesis.
Ultrasound‑mediated microbubble cavitation can induce transient tumor perfusion loss (TTPL), yet the pressure‑dependent magnitude and short‑timescale recovery of this effect remains poorly defined. This study investigated how acoustic pressure governs both the extent and duration of TTPL following a single cavitation exposure. Subcutaneous hepatocellular carcinoma tumors (HEPG2 human cell line) in athymic nude mice (n = 15) received a 1‑second cavitation treatment at peak‑negative pressures of 1.4, 2.8, or 4.1 MPa. Due to the small f-number of the transducer employed, the estimated average peak negative pressures of these conditions within the tumor were 0.6, 1.1, and 1.7 MPa respectively. Tumor perfusion was evaluated using contrast‑enhanced ultrasound (CEUS) immediately (within 1 min), 5, 15, 30, and 60 min after treatment. Perfused area loss was quantified with a maximum intensity projection time‑area curve (MIP‑TAC) metric. Cavitation activity was assessed using passive cavitation detection (PCD), and histology evaluated acute tissue effects. Low acoustic pressure produced only partial perfusion loss with full recovery within 5 min. Moderate acoustic pressure induced substantial TTPL followed by near complete recovery by 15 min. High‑pressure treatment caused complete perfusion loss in all tumors, with initial recovery at 15 min but also with a subsequent decline over the hour. Elevated broadband energy recorded with PCD confirmed inertial cavitation in the moderate and high conditions. Histology revealed damage‑associated staining in 1/5 moderate‑pressure tumors and 3/5 high‑pressure tumors, consistent with pressure‑dependent mechanical injury, possibly contributing to the gradual decline in tumor perfusion observed after initial rebound in the high-pressure condition.
Articular cartilage has a specialised extracellular matrix that provides tensile strength and resistance to compression, but repair capacity is limited. Matrix remodelling during growth is essential for long-term tissue function, yet the underlying protein-level adaptations remain poorly characterised in large-animal models relevant to human joint biology. Using non-targeted, label-free mass spectrometry-based proteomics, we profiled full-thickness articular cartilage from goats across seven postnatal ages from neonatal to adult (n = 3 per age). Cartilage proteins were extracted using guanidine-based solubilisation and analysed by mass spectrometry. Selected proteins were further examined by immunohistochemistry. We identified 799 proteins across the seven ages, of which 157 matrisome components grouped into six categories. Development was associated with increased abundance of proteins involved in matrix organisation and stabilisation, including COL6A1, LOX, TIMP3 and CILP. Enrichment analysis revealed a shift from collagen biosynthesis and fibrillogenesis in early postnatal cartilage to elastic fibre organisation, integrin-matrix interactions and glycosaminoglycan metabolism in mature tissue, consistent with transition from matrix assembly to maintenance. Lysozyme increased with age, suggesting a structural role that warrants further study. Several proteins enriched in mature cartilage, including CILP, HTRA1, FN1 and SPP1, have also been implicated in osteoarthritis, suggesting that some molecular features of mature ECM maintenance are shared with diseased tissue. Immunohistochemistry confirmed stable COL2 localisation, loss of deep-zone COL10 staining with maturation and emergence of superficial PRG4 expression in adult cartilage. Our findings define the proteomic trajectory of cartilage maturation and provide a molecular reference for joint development and matrix ageing.
Hydromorphone is a potent semi-synthetic opioid analgesic frequently administered for acute pain management in hospitalized patients. Opioids have been shown to cause muscle rigidity, akinesia, and catalepsy in human and animal models (Vankova, 1996), but full syndromic catatonia induced by opioids is not well documented. To our knowledge, only four published cases describe opioid-induced catatonia, with reported presentations predominantly characterized by motor findings such as rigidity, and only one case requiring treatment with lorazepam (Yeoh, 2022; Huang, 2007; Di Rosa, 2014; Ketigian, 2023). We present a unique case of hydromorphone-induced mixed state catatonia that met standardized diagnostic criteria and resolved completely with lorazepam. This case underscores that catatonia remains under-recognized outside of psychiatric settings, where its features may be misattributed to delirium, highlighting the importance of early psychiatric consultation.
Simulation and tabletop exercises are widely used in disaster preparedness education, yet comparative evidence remains limited. Pilot findings informed the refinement of this study. To evaluate and compare prelicensure baccalaureate nursing students' perceived knowledge acquisition, confidence, and satisfaction following participation in either a full-scale (FS) simulation or a tabletop (TT) disaster preparedness exercise using standardized facilitation and structured debriefing. A quasi-experimental design included 316 students assigned by clinical practicum to FS or TT across 3 semesters. Validated survey tools measured perceived knowledge, simulation effectiveness, satisfaction, confidence, and debriefing quality. Both modalities demonstrated significant pre-post increases in perceived knowledge with comparable baseline scores. TT exercises showed greater increases in confidence and learning, whereas the FS simulation more strongly supported communication and decision-making. FS and TT modalities are complementary approaches for disaster education when aligned with best practices.
Constructing functional connectivity networks from electroencephalogram (EEG) channels and using graph neural networks for emotion recognition have emerged as a significant technical route in EEG emotion recognition. However, most existing approaches are limited to estimating brain graph networks based on EEG full-channel signals, failing to adequately explore the representations between and within brain regions. To address this limitation and further investigate the interactions between channels and regions, an explainable cross-level topological network (ECTN) is proposed for EEG emotion recognition, which is designed to capture EEG functional interactions from channel-level to region-level. Within the ECTN framework, three modules are designed, namely cross-region topological feature fusion module, specific-region position-guided attention module, and bidirectional gated fusion module. Specifically, EEG functional interactions are explicitly decoupled into two complementary views: global region interactions and local region dynamics. Additionally, the bidirectional gated fusion module leverages the inclusion relationships between channels and brain regions to further integrate region-level and channel-level features. The ECTN model is evaluated on the publicly available SEED series of datasets, SEED, SEED-IV and SEED-V. Experimental results indicate that our method achieves superior performance, effectively validating the benefits of exploring channel-wise and region-wise interactions.
In the context of the global big data deluge, concerted efforts are being made to address the challenges faced by large scientific facilities. These efforts are focused on providing users with the full potential offered by real-time, remote and self-driving experiments, where AI-driven analysis can guide experimental decisions in real time, while ensuring that the data pipelines adhere to the findability, accessibility, interoperability and reuse principles throughout their entire facility lifecycle. Besides all the efforts being made, a user-centric and user-friendly centralization of the overall scientific computing framework at the large scientific facilities remains a work in progress. To address this challenge the Big Data Science Center at the Shanghai Synchrotron Radiation Facility has developed and deployed a centralized, cohesive and user-friendly platform on top of its already existing superfacility framework, which is designed to manage the complete data lifecycle at large scientific facilities. This user-centric platform has transformed the user experience, shifting focus from complex data operations to scientific interpretation. Consequently, the accessibility of the facility to users has been considerably enhanced, thereby expediting the pace at which their discoveries are made.
The optimization of hydroxypropylated cassava starch (HPS) films reinforced with açaí residue (AR) for sustainable packaging applications using low glycerol content (7.5 wt%) is reported. A 23 full factorial design of experiments (DoE) combined with a random forest (RF) algorithm was applied to optimize the hydroxypropylation reaction by evaluating the effects of propylene oxide/hydroxyl groups molar ratio (PO/OH), AR percentage, and reaction temperature. Hydroxypropylation significantly improved film flexibility even at 7.5 wt% glycerol amount, from around 6% to 13%, while AR incorporation maintained tensile strength and Young's modulus similar to the control sample, of around 6 MPa and 250 MPa, respectively. The optimized films also exhibited thermal stability comparable to that of native starch films with maximum decomposition rate at around 250 °C and rapid disintegration under soil burial conditions, occurring within 1 day. Statistical analysis and machine learning consistently indicated that higher molar ratio and fiber content favor enhanced mechanical performance. Overall, the results demonstrate that optimized hydroxypropylation enables the production of starch-based films with improved properties, reduced synthetic plasticizer content, and potential for agro-industrial waste upcycling following green chemistry principles.