Headlines

Researchers have discovered Δ246p53, a conserved 18-kilodalton p53 isoform translated from codon 246 that responds to DNA damage and enhances p53 family-driven senescence through interactions with full-length p53 and ΔNp63.

A new mouse study shows that bone marrow mesenchymal stem cells treat chemotherapy-induced premature ovarian failure by rebalancing the Bcl2/Caspase3 apoptosis pathway, restoring follicles and hormone levels.

Gene therapies are used to treat many different blood disorders. However, blood stem cells often fail to achieve their full potential after transplantation.

New research shows that gamma irradiation drives p53 wild-type medulloblastoma cells into a senescent state that senolytic drugs such as Navitoclax and Copanlisib can then selectively eliminate, suggesting a way to reduce the side effects of craniospinal radiotherapy in children.

The Essential Cycles of Regenerative Agriculture The Real Cost of Not Knowing Regenerative agriculture has done a lot of good for a lot of operations, but somewhere along the way it also turned into a marketing word. Biological inoculants. Soil-boosting amendments. Proprietary blends with names that sound like they were focus-grouped by a pharmaceutical company. […] The post The Essential Cycles of Regenerative Agriculture appeared first on Understanding Ag.

A Texas filing puts Oracle Health's 2025 Cerner breach at nearly 20 million people. See what data was exposed and what patients should check now.

A new bioinformatic study in Reproductive Sciences shows that a risk model built from telomere-related senescence genes can independently predict survival outcomes in cervical cancer patients.

Perficient has launched an AI suite to help Medicare Advantage plans raise their CMS star ratings, which decide access to more than $13B a year in federal bonus payments. It arrived a day after CMS published 2027 ratings showing fewer plans qualifying, and the consultancy is owned by Swedish buyout group EQT. Perficient has launched […] This story continues at The Next Web

Researchers from the University Hospital of Bonn and the University of Bonn have succeeded in dramatically rejuvenating human cells in a test tube. To do so, they directly reprogrammed red blood cell precursors into stem cells from which neurons can be derived. During this process, the molecular clocks that indicate a cell's age were reset. For example, blood cells from an 80-year-old were transformed into stem cells with a molecular age of less than 20 years.

NYU Langone ophthalmology researchers presented trial evidence that a ROCK inhibitor eyedrop can spare corneal transplants, staged therapies for corneal neovascularization, hidden vitamin A deficiency cases, and progress toward whole-eye transplantation at the AAO 2026 meeting.

This guide is built to make that decision easier. We spent months researching the top 10 hair transplant clinics in Turkey, analysing thousands of verified reviews, and cross-checking surgeon credentials against ISHRS and JCI databases. The result is an independent ranking of the top 10 hair transplant clinics in Turkey.

A study of 414 undergraduates at Arsi University finds 70 percent use generative AI tools like ChatGPT for self-diagnosis and self-medication, with digital literacy and weak health-seeking behavior among the key drivers.

Biomedical engineers at Duke University have developed a three-layer model of human blood vessels that enables researchers to better understand how Hutchinson-Gilford progeria syndrome (HGPS), a rare genetic disorder that causes accelerated aging, affects the vascular system.

Scientists have engineered human stem cell-derived spheroids containing myelinating oligodendrocytes and reactive microglia that model demyelination and successful remyelination, offering a drug-screening platform for multiple sclerosis therapies.

Nvidia commits $1 billion over five years to support US science, quantum computing and AI research, expanding its role in the federal Genesis Mission. The post Nvidia’s $1 Billion Research Plan Targets AI, Quantum Computing and Medicine appeared first on TechRepublic.

A scoping review of 19 studies finds that multimodal AI fusing raw images with non-image data in mental and cognitive healthcare is dominated by a single dataset, intermediate fusion strategies, and very little real-world clinical validation.

Blood stem cells are rare cells in the bone marrow that produce the blood and immune cells needed throughout life, ensuring that the body's blood and immune system are renewed. This is why they are also used in bone marrow transplants and novel gene therapies for treating a range of blood disorders. However, during gene therapy, the stem cells need to be processed in the laboratory before transplantation, which can cause them to lose some of their regenerative capacity and reduce the chance of the overall treatment being successful.

An explainable generative deep learning framework called SYNERGY-VAE has uncovered three hidden population subgroups with distinct health signatures and depression prevalence ranging from 6.8 to 10.9 percent, enabling more accurate subgroup-specific risk prediction.

New research shows that the bile acid receptor FXR coordinates monocyte recruitment and Kupffer cell proliferation through CCR2 and CSF1/PI3K/Akt signaling, making the receptor and its macrophage targets a promising therapeutic axis for enhancing liver regeneration.

As more people turn to AI chatbots for quick, 24/7 medical advice, a new perspective paper co-authored by faculty at Binghamton University, State University of New York, explores the risks of this practice that are largely invisible to users—and health care systems.

Researchers at ETH Zurich and Roche have combined active learning with geometric and self-supervised deep learning to predict C–H borylation regioselectivity in drug-like molecules, correctly identifying borylation sites on all ten prospective test substrates.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. We’re putting too much faith in AI’s ability to say no Today’s AI models are trained to refuse a vast number of prompts. If you ask your chatbot how to poison…

A Nature study shows that alveolar type 2 stem cells migrate across compartmental boundaries and, guided by immune-derived SPP1 and Notch signals from ciliated cells, transdifferentiate into club and ciliated cells to repair injured bronchioles.

Laboratory cultures of three Heterocypris ostracod species have revealed complete developmental sequences from egg to adult, providing crucial reference data for identifying fossil shells and interpreting past environmental change.

After the collapse of the Soviet Union, external forces openly sought to manipulate the newly independent states, Alexander Lukashenko recalled

New research shows that enzyme-digested edible bird's nest amplifies nerve growth factor signaling through the TrkA-ERK1/2-CREB-BDNF pathway, priming PC12 cells to differentiate into neurons.

New research shared exclusively with TIME shows chatbots are good at spotting distress, but frequently fail to refer users to resources

Dental follicle stem cells can be used as a basis for regenerative cell therapy in the treatment of periodontitis, the prevalent form of gum disease, and a condition that becomes more common with age. Researchers here note that IGFBP5 expression is reduced in senescent dental follicle stem cells, and find that increased expression of IGFBP5 reduces the pace at which these cells become senescent. Cellular senescence in the cultured cells used in cell therapy is a problem that the regenerative medicine community is becoming increasingly aware of, and various efforts have been made in various contexts to try to mitigate the issue. Here, researchers employ this new knowledge regarding the role of IGFBP5 to design a better cell therapy to promote periodontal regeneration, and carry […]

A genome-scale CRISPR screen in human hepatocytes has revealed an extensive network of regulators governing the aryl hydrocarbon receptor, including the E3 ubiquitin ligase UBR5, with implications for treating metabolic and inflammatory disease.

Conversational diagnostic artificial intelligence in ambulatory primary care: a prospective feasibility study  The LancetClinical Trial Indicates A.I. Chatbots Could Aid Urgent Care  The New York TimesStudy in The Lancet suggests AI could improve patient-physician relationships.  blog.googleAI chatbot could spot health problems before you see your doctor  InYourAreaGPs could use AI chatbots to diagnose patients ahead of visits, says study  The Independent

Conversational diagnostic artificial intelligence in ambulatory primary care: a prospective feasibility study  The LancetClinical Trial Indicates A.I. Chatbots Could Aid Urgent Care  The New York TimesStudy in The Lancet suggests AI could improve patient-physician relationships.  blog.googleAI chatbot could spot health problems before you see your doctor  InYourAreaGPs could use AI chatbots to diagnose patients ahead of visits, says study  The Independent

A controlled evaluation of eight ambient voice technologies found that 30 to 68 percent of AI-transcribed surgical records contained clinically significant errors, with the most dangerous mistakes concentrated in medication doses and laboratory results.

A self-reinforcing injectable hydrogel converts body movement into electrical signals that trigger calcium-dependent histone acetylation and accelerate bone regeneration in rats.

Silicon Valley billionaires are obsessed with 'reprogramming' their bodies for longevity. The first human trial suggests they may be onto something.  Yahoo News UKSee more headlines & perspectives on Google News

Silicon Valley billionaires are obsessed with 'reprogramming' their bodies for longevity. The first human trial suggests they may be onto something.  Yahoo News UKSee more headlines & perspectives on Google News

Silicon Valley billionaires are obsessed with 'reprogramming' their bodies for longevity. The first human trial suggests they may be onto something.  YahooSee more headlines & perspectives on Google News

Silicon Valley billionaires are obsessed with 'reprogramming' their bodies for longevity. The first human trial suggests they may be onto something.  YahooSee more headlines & perspectives on Google News

Stem cell-derived sperm have been a long-pursued goal. Now researchers have produced spermatogonia-like cells that resemble those found in the testes after birth.

Liver cancer is the fifth most commonly diagnosed cancer and the second leading cause of cancer death worldwide. Hepatocellular carcinoma (HCC) accounts for 70-85% of primary liver cancers, and its five-year recurrence rate can reach 70-80%.

A survey of 113 Palestinian nursing students found that more positive perceptions of artificial intelligence were independently associated with stronger self-reported e-health literacy.

arXiv:2610.12377v1 Announce Type: new Abstract: Target trial emulation has become a standard framework for causal inference from observational data. Within this paradigm, the clone-censor-weight (CCW) methodology provides a practical way to deal with complex treatment and adherence definitions when treatment regimes are not distinguishable at baseline. Despite its increasing use, the statistical interpretation of CCW remains limited. In this work, we formalize the CCW methodology from a statistical viewpoint based on admissible sets and stochastic interventions. We characterize the causal estimands targeted by CCW, establish consistency of CCW estimators under standard identification assumptions, and provide theoretical guarantees for bootstrap-based inference methods.

arXiv:2610.11919v1 Announce Type: new Abstract: In this paper we consider random walks moving in a dynamic random environment that is completely resampled with probability $p$ after each step of the random walker. This resampling yields a straightforward regeneration structure and limit theorems, such as a law of large numbers with a limiting speed $v(p)$ that depends on $p$. We investigate several properties of $v(p)$ as a function of $p$, such as differentiability and monotonicity. Similarily, we study the limiting diffusivity $\sigma^2(p)$ in a central limit theorem.

arXiv:2610.10857v1 Announce Type: new Abstract: Behavior cloning (BC) in non-Markovian environments is a challenging problem because policies have to reason over contextual information over long horizons. Existing policy architectures rely on recurrent or attention-based mechanisms to capture long-term dependencies. However, recurrent models suffer from hidden-state collapse and gradient instability under backpropagation through time, while attention-based models are fundamentally limited by context length. To address these issues, we propose Keyframe Mnemonics, a novel self-supervised method that $\textit{discovers}$ a set of information-critical observations ($\textit{mnemonics}$) by learning an objective from randomly sampled past observations and using it as a reward for keyframe selection. We then train a BC policy that conditions on the discovered keyframes to model the action distribution. Under certain task-structure assumptions, our formulation provides context retention

arXiv:2610.10774v1 Announce Type: new Abstract: Audio generation models can translate natural-language descriptions into sound, but their outputs are typically waveforms. Their audio quality is constrained by audio compression, and their outputs do not readily support direct edits to notes, timbral parameters, or modulation relationships. We present AutoSynth, which represents MIDI performance events, fixed synthesizer parameters, and variable-length modulation routes as a unified sequence for a synthesizer, and learns their dependencies with an audio-conditioned autoregressive model. A single model supports both tasks. Given reference audio, the model directly predicts a synthesizer program; given text, it uses a pretrained audio generation model and converts the generated audio into a program. Training consists of two stages: supervised learning on large-scale audio-program pairs automatically constructed from a small set of native presets, followed by group-relative policy

A French retrospective study of 161 allogeneic hematopoietic cell transplantation recipients finds that day three SOFA scores and new organ support requirements predict ICU mortality better than admission values, though overall prediction accuracy remains limited.

New research in Nature Aging shows that the age-related decline of chaperone-mediated autophagy distorts the senescent cell secretome and disables macrophage phagocytosis, and that pharmacologically restoring the pathway clears senescent cells and eases lung fibrosis in mice.

Silicon Valley billionaires are obsessed with 'reprogramming' their bodies for longevity. The first human trial suggests they may be onto something.  Business InsiderEye Cells May Have Been Rejuvenated for the First Time in People  Time MagazineLife Biosciences Announces First-in-Human Data from Ongoing Phase 1 Trial Evaluating ER-100 in Optic Neuropathies  Life BiosciencesCellular rejuvenation reaches a human trial  Longevity.TechnologyLife Biosciences Reports Early Results From Human Cell Reprogramming Trial  SuaraGarut.ID

Silicon Valley billionaires are obsessed with 'reprogramming' their bodies for longevity. The first human trial suggests they may be onto something.  Business InsiderEye Cells May Have Been Rejuvenated for the First Time in People  Time MagazineLife Biosciences Announces First-in-Human Data from Ongoing Phase 1 Trial Evaluating ER-100 in Optic Neuropathies  Life BiosciencesCellular rejuvenation reaches a human trial  Longevity.TechnologyLife Biosciences Reports Early Results From Human Cell Reprogramming Trial  SuaraGarut.ID

Older people could be monitored safely in their own homes without the need for intrusive cameras or wearable devices thanks to new research by Nottingham Trent University (NTU).

Epidemiological studies link light at night (LAN), shift work, chronic jet lag (CJL), and core clock gene mutations to higher cancer risk and aggressive spread, especially in breast and colorectal cancer.

An integrated analysis of human genetics, patient-derived cell models and electronic health records has identified 19 Alzheimer's drug targets in European ancestry populations and seven in African ancestry populations, with EPHX2 inhibition emerging as a leading therapeutic candidate.

Technology inspired by 3D printing is already being introduced in hospitals and specialized pharmacies to produce personalized medicines tailored to individual patients. In her doctoral thesis at Åbo Akademi University, Mahsa Bahman investigated how the approach can improve dosing accuracy and make customized treatments more widely available.

A novel laboratory setup separating two pegmatitic melts with a fluid phase shows that lithium isotopes fractionate between melt and fluid, with the fluid preferentially enriched in the heavier isotope lithium-7.

A new review in npj Parkinson's Disease examines the strategies needed to improve survival, differentiation and functional integration of stem cell-derived dopaminergic grafts as cell therapy for Parkinson's disease gains conditional approval in Japan.

Researchers have functionally characterized CD34/CD123/CD25/CD99-positive leukemic progenitor cells in FLT3-ITD-mutated AML, showing they engraft and reinitiate leukemia in mice and can be targeted by combining gilteritinib with LCK inhibition.

Johnnie "DJ" Jones underwent heart transplant surgery Wednesday after Mayo Clinic doctors in Jacksonville detected a serious cardiac condition.

A meta-analysis of randomized trials combined with Mendelian randomization finds that physical activity improves key aging indicators, with peak benefits estimated at about 11.5 marginal MET-hours per week.

Artificial intelligence (AI) has had a transformative effect on health care. The technology has been tested and applied in almost every area, from disease diagnostics to AI-powered digital clinics.

A multicenter Japanese study of 135 patients shows robot-assisted pelvic exenteration for advanced rectal cancer achieved 98 percent complete resection with dramatically lower blood loss and no 30-day deaths, marking rapid adoption of robotics in one of surgery's most radical operations.

New research shows that the metabolic protein ANGPTL8 drives the aging of hematopoietic stem and progenitor cells by binding the PirB receptor and suppressing the PI3K/AKT pathway.

Researchers show that a selective embedding data loading strategy alternating X-ray and CT scans lets deep learning models reach competitive medical image classification accuracy with far less training data.

New research shows that transient inhibition of the mitochondrial enzyme succinate dehydrogenase triggers coordinated metabolic and epigenetic reprogramming across heart cell types, reactivating regenerative programs and suppressing fibrosis after myocardial infarction.

A long-term study of 22 patients shows that vascularized fibular grafting achieved bone healing and infection control in nearly three-quarters of severe infected non-unions of the lower limb, though consolidation often took years and a competing-risk analysis was needed to capture the true timeline.

A phase II pilot trial found that a single fecal microbiota transplantation combined with atezolizumab and bevacizumab was safe and produced tumor responses in one third of liver cancer patients whose disease had resisted immunotherapy.

A scoping review of 118 studies finds that deep learning models excel at detecting parasites in microscopy images under experimental conditions, but very few have been validated or deployed in real clinical laboratories.

A multi-institution feasibility study found that patient-specific 3D-printed models of pulmonary arteriovenous malformations were successfully fabricated for all cases and helped interventionalists plan embolization procedures and educate patients.

His idea is an approval system for AI — but what would it take to get a green light?

Simulated Patients Before Real Trials A Copenhagen startup has gone on record predicting that drug trials from Biogen

’s virtual summit returns on November 4 to examine how AI is reshaping R&D timelines and real-world patient translation. From big pharma to AI-natives, this year’s program features the leaders of many of today’s outstanding companies and organizations powering the AI revolution! The post The State of AI in Drug Discovery appeared first on GEN - Genetic Engineering and Biotechnology News.

The tissues of the mammalian body span a broad range of regenerative capacity, from the highly regenerative liver to the minimal regenerative capabilities of the brain and the heart. We might speculate that the evolutionary road to a reduced capability of regeneration in the central nervous system results need to retain structure that stores memory. The heart incorporates a complex electrical system, and the need to retain the structure of this system may also place constraints on the level of regeneration that can take place following injury. Heart damage, such as that resulting from a heart attack and consequent transient loss of blood supply to regions of the heart, heals poorly, and is typically characterized by fibrosis that is disruptive to heart tissue structure and […]

Researchers have shown that the matrix protein fibronectin reprograms macrophages in damaged dental pulp into pro-angiogenic cells, enhancing blood vessel formation and regenerating pulp–dentin-like tissue in a preclinical model.

A study of more than 155,000 adults across 32 countries finds that the healthy-ageing benefits of good lifestyles vary up to fivefold between nations, depending on economic development, healthcare access, climate resilience, and social conditions.

Technology inspired by 3D printing is already being introduced in hospitals and specialized pharmacies to produce personalized medicines tailored to individual patients.

Healthleap raises $38 million to expand AI that analyzes hospital records for overlooked health risks, as questions remain about clinical accuracy and patient care. The post Healthleap Raises $38M for AI That Spots Hidden Health Risks in Hospitals appeared first on TechRepublic.

New research shows carboplatin drives the expansion of TP53-mutant blood clones far more strongly than PARP inhibitors, while inherited homologous recombination deficiency unexpectedly blunts this expansion.

In early-stage cervical cancer, radical hysterectomy is an established treatment. The operation can be performed either through open surgery or robot-assisted minimally invasive surgery. Until now, there have been limited data from large randomized studies on differences in complications and recovery between the methods.

A phase-2 randomized trial shows intermittent dasatinib plus quercetin significantly improved liver fibrosis and resolved MASH in nearly half of treated patients, with single-cell sequencing confirming depletion of senescence and fibrosis signatures.

Three years ago, George Kassiotis and his team made an unexpected discovery: A specific way of producing a protein called calbindin was found to help cancer cells survive by preventing them from aging. This left Kassiotis, who runs the Crick's Retroviral Immunology Laboratory, confused. Why would a mechanism that boosts cancer growth exist in the first place?

Optogenetics, recognized by the 2026 medicine Nobel, makes cells respond to light, including eye cells that normally relay signals to the brain.

Kidney transplant and wait-listing rates were lower for beneficiaries with Medicare Advantage vs. traditional Medicare, according to study data published in JAMA.The 21st Century Cures Act, passed in 2021, permitted all individuals with end-stage kidney disease to enroll in Medicare Advantage plans, according to Ankur Shah, MD, MPH, associate professor of medicine at the Warren Alpert Medical School of Brown University, and colleagues. However, data exploring whether Medicare insurance type impacts access to kidney transplant have been limited, the researchers wrote.“Earlier work had shown

Researchers at Shanghai Jiao Tong University have developed a fault diagnosis framework that uses recurrent attentional reinforcement learning to adaptively search for the most informative temporal fragments in machinery sensor data, achieving 76.4 percent accuracy on rolling bearings and 100 percent on a hydraulic system.

A secondary analysis of the RACC trial assesses perioperative safety and early quality of life after robot-assisted vs open radical hysterectomy in patients with early-stage cervical cancer.

In a UK first, doctors in Edinburgh have transplanted frozen testicular tissue into a young man left infertile by chemotherapy, in only the second such procedure reported worldwide.

A new commentary argues that lung nodule localisation is entering a precision era, as stabilised fluorescent matrices aim to solve the diffusion problem that undermines surgical margins in robotic sublobar resection.

Tax-exempt hospitals now sell longevity programs from $5,500 to as much as $45,000 a year. Here's what's included, what insurance covers, and what critics ask.

The DNA Series: This is the last blog in the DNA Series (boo!), and we’re delving into the world of gene editing. For this series, … The post DNA Series: Gene Editing appeared first on SAIFood.

A new Pediatric Research commentary argues that insurance coverage and reimbursement policy, not the underlying science, now determine which children with rare genetic diseases can access curative gene therapies.

Innovative techniques using electrical energy to treat bladder cancer may lower recurrence rates and minimize the need for radical surgical interventions.

Shared family factors, such as genes and upbringing, are likely to explain a large part of the link between high screen use and mental health problems in young people. This is according to a new study from Karolinska Institutet published in Nature Human Behaviour.

Researchers in Vietnam have built a hybrid vision-language AI that diagnoses grapevine diseases with 98.7 percent accuracy while explaining its reasoning and resisting the background overfitting that cripples conventional CNNs in real fields.

Researchers in Madrid have developed an in vitro stimulation model showing that B cells from patients with common variable immunodeficiency can begin activating but fail to sustain antibody production, accompanied by a shift toward immune-inhibitory gene expression and altered cytokine profiles.

arXiv:2610.10346v1 Announce Type: new Abstract: Robotic minimally invasive surgery has revolutionized surgical practice. However, as the surgeons are mechanically separated from the surgical tools, loss of tactile feedback occurs. We present a compact force sensor with three, millimeter scale sensing elements (taxels) that can be affixed to the grasping face of a robotic surgical tool. The sensor features a miniaturized design using Hall effect sensors and magnets embedded in a deformable elastomer contact layer. Sensor calibration is achieved with a 6 degree of freedom (DoF) parallel robot. Each taxel can detect applied normal and shear forces with average RMSE of 2.133 kPa. The calibrated sensor was validated on phantom and ex vivo porcine tissue, demonstrating tactile sensing during retraction as well as slip. This work introduces a small-scale Hall effect based tactile sensor, enabling sensing for force-sensitive robotic surgical applications.

arXiv:2610.10256v1 Announce Type: new Abstract: Reality may establish that an outcome occurred without identifying which evolving procedure produced it or why. This distinction matters in production ML systems whose code, configuration, and artifacts change while external feedback accumulates. We examine it in a human-directed, agent-engineered quantitative trading system, using oracle to mean an external source of realized outcomes rather than a complete correctness specification. Across one year, the account gained and outperformed a broad market index, while annual alpha was not statistically distinguishable from zero under the main retrospective specification. Retrospectively selected subperiods include adverse relative performance and conditional candidate-level weakness under declared approximate references. Engineering records document changes during the episode, and complete recommendation-to-runtime binding is unavailable. The archive does not establish a common frozen

arXiv:2610.09702v1 Announce Type: new Abstract: Ordinary prompt-based editing can cause latent watermark detection to fail without explicitly targeting the watermark. We benchmark eight watermark methods against five editors across four generative backbones, four editing strengths, and five semantic categories, with edit-validity and threshold checks. Separating editing from seven subsequent distortions reveals that editing alone primarily distinguishes Tree-Ring, while added distortions expose a broader spectrum of detection survival. Sequential edits reveal a second hidden difference: score separation can decline while detection rates remain near their ceiling. Across methods, standardized clean score separation ($d'$) organizes composite-survival tiers, whereas spatial overlap adds little to predicting edit-only survival beyond clean detectability. Embedding-strength interventions in two methods link higher clean separation to higher post-edit separation. In HSTR, the margin

arXiv:2610.09558v1 Announce Type: new Abstract: Drug target discovery requires distinguishing molecules that causally drive disease from those that are merely associated with it. Training and evaluating AI agents to perform this workflow end-to-end is difficult because real world biobanks lack known causal ground truth and participant-level data is access controlled. We introduce DrugTargetWorld, a framework that procedurally generates simulated biobanks, or "worlds," with known but concealed causal structure. Each world contains genotypes, proteins, health records, outcomes, and synthetic magnetic resonance imaging (MRI) for 54,000 participants. Agents must construct a disease phenotype, identify causal driver proteins, infer the beneficial direction of modulation, and optionally conduct virtual 'wet lab' experiments. We evaluated nine agents in 540 episodes across 20 cardiovascular worlds and three experimental budgets. Opus 5 and GPT-5.6 Sol achieved the highest mean composite

arXiv:2610.09211v1 Announce Type: new Abstract: We present a recipe for adapting English-pretrained autoregressive TTS foundation models to underrepresented languages, demonstrated on Bangladeshi Bangla. Existing Bangla TTS corpora are small and single-speaker, and to our knowledge no open zero-shot voice-cloning system is available for the Bangladeshi register. We contribute a phonetically- and gender-balanced two-tier Bangladeshi Bangla corpus balanced via a tiered Jensen-Shannon divergence objective over conjunct clusters (juktakkhor), together with three fine-tuning changes: a merge-consistent tokenizer extension, Bangla text normalization, and a prompt-masked dual-loss objective. These changes preserve zero-shot cloning across the language switch. Our BanglaEval protocol applies Wilcoxon signed-rank tests with Bonferroni correction over native-speaker ratings. BanglaBox attains near-natural Naturalness, outperforms commercial and open-source baselines on Naturalness, Speaker

arXiv:2610.09146v1 Announce Type: new Abstract: Large language models (LLMs) and vision-language models (VLMs) are usually frozen after deployment, so they do not learn from the cases they solve. This is especially concerning in medicine, where new clinical evidence, updated guidelines, and new therapies can change established practice. Fine-tuning can update the model, but it requires access to model weights and additional training. Parameter-free methods avoid training, but they may overfit a fixed validation set, lack reliable domain knowledge, or lose visual details by saving experience only as text. To address these limitations, we present a model-agnostic framework that allows frozen LLMs and VLMs to learn from deployment experience through three forms of external expertise: a Skill that guides reasoning and tool use, a Knowledge Memory that stores reliable facts supported by earlier cases or trusted external evidence, and a Multimodal Knowledge Base that keeps visual examples

arXiv:2610.09134v1 Announce Type: new Abstract: Robot world models support policy evaluation, planning, and synthetic data generation, but these applications require predictions that distinguish successful actions from failures. Across four released checkpoints from two architecture families, we observe weak sensitivity to action changes and success-like predictions on verified failures. Although recent work incorporates failures into model training, which data can repair released checkpoints without changing their architecture or training objective still remains underexplored. To this end, we introduce CureWM, which constructs alternative actions from successful demonstrations across a severity grid, verifies their outcomes through execution in simulation or on hardware, and fine-tunes released models on the resulting failures and surviving successes alongside nominal demonstrations. This construction provides controlled action contrasts from shared starting contexts. On 484 held-out