当前共找到 1276 篇文献分享,本页显示第 41 - 60 篇。
41.
孤舟蓑笠翁
(2025-06-15 09:44):
#paper 【doi】10.1038/s41591-025-03562-5;【发表年份】2025年;【期刊】Nature Medicine;【标题】Prospective, multicenter validation of a platform for rapid molecular profiling of central nervous system tumors。【内容总结】研究人员想开发一个又快又准的方法来诊断脑肿瘤,因为现有技术太慢太贵,尤其在全球资源有限的地方不好用,所以他们搞了个叫Rapid-CNS2的新工具,结合纳米孔测序和自适应采样技术,能在手术中30分钟内给出肿瘤的大致分类和基因拷贝数信息,24小时内提供完整的分子报告,包括突变、融合和甲基化数据,还搭配了一个叫MNP-Flex的智能分类器处理甲基化信息;他们在多个医院测试了301个肿瘤样本,结果超级好——甲基化家族分类准确率92.9%,基因突变检测与常规方法91.67%一致,拷贝数分析与金标准完全匹配,MNP-Flex更牛,家族和子类准确率分别达99.6%和99.2%,术中测试15分钟就搞定83%的样本,整体诊断时间从几周缩到40小时,让医生能更快制定个性化治疗方案。
42.
惊鸿
(2025-06-13 23:47):
#paper
Nature
Pub Date : 2025-06-11
DOI : 10.1038/d41586-025-01453-w
Computer processors built from 2D materials
2025年6月《自然》期刊报道的两项独立研究(Ghosh等和Ao等)首次实现了基于二维材料的完整计算机处理器原型。Ghosh团队构建了约1000个二硫化钼晶体管的单指令集芯片,Ao团队则集成了约2500个二维晶体管制成兼容RISC-V架构的32位微处理器。这两项突破验证了二维材料(如过渡金属硫化物)在系统级集成电路中的可行性,其核心价值在于:二维材料的原子级厚度(约0.7纳米)可突破硅基晶体管在10纳米尺度以下性能衰减的物理限制。
然而,当前技术仍面临三重挑战:一是材料缺陷导致器件均一性不足,实验室原型良率远低于商用硅芯片;二是制造工艺不兼容现有半导体产线,需开发全新的二维材料晶圆级集成技术;三是载流子迁移率等关键性能参数仍需优化。研究团队通过创新设计(如Ao采用双栅结构提升稳定性)部分缓解了这些问题,但距离产业化尚有距离。
从技术演进角度看,这两项工作标志着二维电子学从器件研究迈向系统集成的重要转折点。短期看,二维材料可能作为硅基芯片的补充模块(如特定低功耗单元);长期若突破规模化瓶颈,或将为后摩尔时代提供新的技术路径。其科学意义在于实证了原子级薄层材料承载复杂计算架构的能力,为纳米尺度电子器件发展提供了关键数据支撑。
Nature,
2025-6-12.
DOI: 10.1038/d41586-025-01453-w
Abstract:
No abstract available.
43.
孤舟蓑笠翁
(2025-06-13 20:26):
#paper 【doi】10.1038/s41591-025-03692-w;【发表年份】2025年;【期刊】Nature Medicine;【标题】T and B cell responses against Epstein-Barr virus in primary sclerosing cholangitis。该研究简略地探讨了EB病毒(EBV)与原发性硬化性胆管炎(PSC)的免疫关联,发现PSC患者的T/B细胞对EBV反应异常增强。具体通过比较504名PSC患者和904名健康人的T细胞受体(TCR)测序,鉴定出1008个PSC相关克隆型,其中多数靶向EBV抗原(如BZLF1和EBNA3),且受PSC风险HLA等位基因(如HLA-B*08:01)限制;通过噬菌体免疫沉淀测序(PhIP-seq)分析120名PSC和202名健康人血清,显示PSC患者抗EBV抗体反应显著升高(尤其靶向BMRF1等裂解期蛋白);从PSC肝脏浸润B细胞提取的单克隆抗体也证实靶向EBV抗原(如BFRF3);最后通过超116万人电子病历分析,发现传染性单核细胞增多症(EBV引起)患者患PSC风险比健康人高12倍(95% CI: 6.3-22.9),远高于其他免疫疾病(如克罗恩病),表明EBV感染与PSC发病存在强流行病学关联,提示EBV再激活可能在PSC病理中起关键作用。
Nature Medicine,
2025-6-11.
DOI: 10.1038/s41591-025-03692-w
Abstract:
Abstract Primary sclerosing cholangitis (PSC) is an idiopathic, progressive and incurable liver disease. Here, we aimed for systematic analyses of adaptive immune responses in PSC. By profiling the T cell …
>>>
Abstract Primary sclerosing cholangitis (PSC) is an idiopathic, progressive and incurable liver disease. Here, we aimed for systematic analyses of adaptive immune responses in PSC. By profiling the T cell repertoires of 504 individuals with PSC and 904 healthy controls, we identified 1,008 clonotypes associated with PSC. A substantial fraction of these clonotypes was restricted to known PSC human leukocyte antigen susceptibility alleles and known to target Epstein–Barr virus (EBV) epitopes. We further utilized phage-immunoprecipitation sequencing to determine antibody epitope repertoires of 120 individuals with PSC and 202 healthy controls, which showed a higher burden of anti-EBV responses in PSC than controls. EBV-specific monoclonal antibodies isolated from B cells in PSC livers corroborated convergent B and T cell responses against EBV. By analyzing electronic health records of >116 million people, we identified an association between infectious mononucleosis and PSC (odds ratio, 12; 95% confidence interval, 6.3–22.9), suggesting a link between EBV and PSC.
<<<
翻译
44.
孤舟蓑笠翁
(2025-06-12 11:26):
#paper 【doi】10.1016/j.cell.2025.05.019;【发表年份】2025年;【期刊】Cell;【标题】MT-125 inhibits non-muscle myosin IIA and IIB and prolongs survival in glioblastoma。这项研究的目标是开发一种新药MT-125来治疗致命的胶质母细胞瘤(GBM),因为GBM是最具侵袭性的脑肿瘤,当前的治疗效果有限,而MT-125通过特异性地抑制非肌肉肌球蛋白IIA和IIB(NMIIA和NMIIB),这两种蛋白在肿瘤细胞运动、分裂和信号传递中起关键作用,从而阻断肿瘤生长。在方法上,研究人员首先设计了MT-125作为已知抑制剂blebbistatin的衍生物,通过生化实验验证它对NMIIA和NMIIB的高选择性(如抑制常数测定),并评估了其药代动力学、脑渗透性和安全性(包括动物毒性测试);在体外实验中,测试了MT-125对GBM细胞侵袭、细胞分裂的影响(如Transwell入侵实验和细胞核多核化分析),以及它如何增加活性氧(ROS)水平和诱导铁死亡;在体内实验中,使用小鼠GBM模型(包括基因工程模型和患者来源的异种移植模型),给予MT-125单药或与放疗或激酶抑制剂(如sunitinib或paxalisib)联合治疗,监测生存期和肿瘤变化。结果显示,MT-125有效抑制肿瘤细胞侵袭和细胞分裂,导致多核化(如48小时后12%-25%细胞出现多核),并增加ROS和铁死亡,显著延长小鼠生存期(如单药治疗使中位生存期提高);更详细地,MT-125表现出良好的安全性和治疗指数(如耐受剂量高达30 mg/kg),与放疗结合时通过ROS机制增强DNA损伤和细胞死亡,与PDGFR抑制剂sunitinib或mTOR抑制剂联合时产生强协同效应,在动物模型中生存期翻倍,部分小鼠实现长期缓解,这为GBM治疗提供了新的临床策略。
45.
林海onrush
(2025-06-07 13:27):
#paper, Token-Importance Guided Direct Preference Optimization,DOI: https://arxiv.org/abs/2505.19653, share一下个人最新的大模型微调算法工作,我们针对大语言模型(LLMs)如何更好地对齐人类偏好提出了一种新方法——TI-DPO。以往常用的DPO(直接偏好优化)方法虽然省去了显式奖励模型,直接用人类偏好数据优化模型,但忽略了不同token(词/字)在生成内容中的重要性差异,这可能导致模型在关键token上犯错,从而产生不符合人类价值观的输出。
TI-DPO通过两大创新点解决了这一问题:
1. 在token level层面引入基于梯度归因的Token重要性权重,能动态识别和优先优化对人类偏好最关键的token;
2. 加入基于对比学习的Triplet(三元组)损失,不仅区分“好-坏”样本,还引入“中间”输出,使优化更细致,有助于模型生成更接近人类期望、远离不理想响应的内容。
实验表明,TI-DPO在多个任务上(如TruthfulQA、IFEval等)表现优异,准确率和生成多样性均超过DPO及其他对齐方法。消融实验进一步验证了token-importance机制和triplet loss的必要性和有效性。理论分析还证明了TI-DPO在优化上拥有更严格的损失下界,训练过程更加稳定。TI-DPO通过精细化地关注关键token,并结合三元组对齐结构,有效提升了大模型的对齐能力与输出质量,为人机交互中的AI安全和有用性提供了新的解决方案。
arXiv,
2025-05-26T08:11:24Z.
DOI: 10.48550/arXiv.2505.19653
Abstract:
Ensuring that large language models (LLMs) generate outputs aligned withhuman preferences is important for safe and effective AI interactions. WhileDirect Preference Optimization (DPO) employs an implicit reward function tooptimize the …
>>>
Ensuring that large language models (LLMs) generate outputs aligned withhuman preferences is important for safe and effective AI interactions. WhileDirect Preference Optimization (DPO) employs an implicit reward function tooptimize the policy model, however, it and its related variants overlook thedifferential importance of individual tokens and are sensitive to judgmentnoise in preference datasets during generation. Although recent methods attemptto assess the important weight of tokens via probability prediction orsimplistic weighting schemes, these evaluation methods are prone to biases andstill cannot fully address these issues. To solve this problem, we propose theToken-Importance Guided Direct Preference Optimization (TI-DPO), whichintroduces two key innovations: the gradient-based token-importance weightsthat dynamically prioritize critical tokens, and a triple loss that explicitlyguides model outputs to approach human-preferred responses and stay away fromnon-preferred responses. Experimental results show that TI-DPO achieves higheraccuracy and stronger generative diversity, providing more stable andcomputationally efficient solutions compared with DPO and other RLHF methods.
<<<
翻译
46.
少颖-focus reverse aging
(2025-06-06 14:30):
#paper Socioeconomic Status and Cardiovascular Outcomes: Challenges and Interventions doi: 10.1161/CIRCULATIONAHA.117.029652 推荐原因:能学到一些规避心血管病的方法; 论文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC5958918 论文重点内容:穷人更容易得心血管病,原因:
1.没钱买优质食物,便宜的食物通常高油高盐;
2.没条件运动,没公园没健身房;
3.没钱体检和买药;
4.压力大长期压抑;
5.保养意识健康意识差;
6. 穷人区环境污染大。
Circulation,
2018-5-15.
DOI: 10.1161/CIRCULATIONAHA.117.029652
Abstract:
Socioeconomic status (SES) has a measurable and significant effect on cardiovascular health. Biological, behavioral, and psychosocial risk factors prevalent in disadvantaged individuals accentuate the link between SES and cardiovascular disease …
>>>
Socioeconomic status (SES) has a measurable and significant effect on cardiovascular health. Biological, behavioral, and psychosocial risk factors prevalent in disadvantaged individuals accentuate the link between SES and cardiovascular disease (CVD). Four measures have been consistently associated with CVD in high-income countries: income level, educational attainment, employment status, and neighborhood socioeconomic factors. In addition, disparities based on sex have been shown in several studies. Interventions targeting patients with low SES have predominantly focused on modification of traditional CVD risk factors. Promising approaches are emerging that can be implemented on an individual, community, or population basis to reduce disparities in outcomes. Structured physical activity has demonstrated effectiveness in low-SES populations, and geomapping may be used to identify targets for large-scale programs. Task shifting, the redistribution of healthcare management from physician to nonphysician providers in an effort to improve access to health care, may have a role in select areas. Integration of SES into the traditional CVD risk prediction models may allow improved management of individuals with high risk, but cultural and regional differences in SES make generalized implementation challenging. Future research is required to better understand the underlying mechanisms of CVD risk that affect individuals of low SES and to determine effective interventions for patients with high risk. We review the current state of knowledge on the impact of SES on the incidence, treatment, and outcomes of CVD in high-income societies and suggest future research directions aimed at the elimination of these adverse factors, and the integration of measures of SES into the customization of cardiovascular treatment.
<<<
翻译
47.
刘昊辰
(2025-06-03 16:34):
#paper AlphaZero-Edu Making AlphaZero Accessible to Everyone. AlphaZero-Edu 是基于 AlphaZero 数学框架开发的轻量化强化学习框架,专为教育场景和五子棋设计,具有模块化架构(解耦蒙特卡洛树搜索、自我对弈训练、策略价值网络)、资源高效训练(单块 NVIDIA RTX 3090 GPU 即可运行)和高度并行自我对弈数据生成(8 进程实现 3.2 倍加速)等特点。其状态特征采用 21 层张量(含当前状态和 20 层历史状态),输出包含策略概率分布和价值评估标量,并通过旋转 / 翻转数据增强提升泛化能力。训练中结合循环学习率调度器,使策略损失和价值损失均收敛,且在与 4 名人类玩家的对战中实现最高 100% 胜率,最低 60% 胜率(含 20% 平局)。该框架已开源,为学术研究和工业应用提供了可访问的基准。下载地址:https://arxiv.org/pdf/2504.14636
arXiv,
2025-04-20T14:29:39Z.
DOI: 10.48550/arXiv.2504.14636
Abstract:
Recent years have witnessed significant progress in reinforcement learning,especially with Zero-like paradigms, which have greatly boosted thegeneralization and reasoning abilities of large-scale language models.Nevertheless, existing frameworks are often plagued by …
>>>
Recent years have witnessed significant progress in reinforcement learning,especially with Zero-like paradigms, which have greatly boosted thegeneralization and reasoning abilities of large-scale language models.Nevertheless, existing frameworks are often plagued by high implementationcomplexity and poor reproducibility. To tackle these challenges, we presentAlphaZero-Edu, a lightweight, education-focused implementation built upon themathematical framework of AlphaZero. It boasts a modular architecture thatdisentangles key components, enabling transparent visualization of thealgorithmic processes. Additionally, it is optimized for resource-efficienttraining on a single NVIDIA RTX 3090 GPU and features highly parallelizedself-play data generation, achieving a 3.2-fold speedup with 8 processes. InGomoku matches, the framework has demonstrated exceptional performance,achieving a consistently high win rate against human opponents. AlphaZero-Eduhas been open-sourced at https://github.com/StarLight1212/AlphaZero_Edu,providing an accessible and practical benchmark for both academic research andindustrial applications.
<<<
翻译
48.
少颖-focus reverse aging
(2025-06-02 00:22):
#paper Cardiovascular aging: from cellular and molecular changes to therapeutic interventions doi: 10.20517/jca.2023.09 论文网址:https://pmc.ncbi.nlm.nih.gov/articles/PMC10238104/#S16 公众号阅读笔记链接:https://mp.weixin.qq.com/s/O9IqSEJpAwSGIlaAMap3mQ
Abstract:
Progressive age-induced deterioration in the structure and function of the cardiovascular system involves cardiac hypertrophy, diastolic dysfunction, myocardial fibrosis, arterial stiffness, and endothelial dysfunction. These changes are driven by complex …
>>>
Progressive age-induced deterioration in the structure and function of the cardiovascular system involves cardiac hypertrophy, diastolic dysfunction, myocardial fibrosis, arterial stiffness, and endothelial dysfunction. These changes are driven by complex processes that are interconnected, such as oxidative stress, mitochondrial dysfunction, autophagy, inflammation, fibrosis, and telomere dysfunction. In recent years, the advances in research of cardiovascular aging, including the wide use of animal models of cardiovascular aging, elucidated an abundance of cell signaling pathways involved in these processes and brought into sight possible interventions, which span from pharmacological agents, such as metformin, sodium-glucose cotransporter 2-inhibitors, rapamycin, dasatinib and quercetin, to lifestyle changes.
<<<
翻译
49.
DeDe宝
(2025-06-01 17:15):
#paper https://doi.org/10.1101/2024.07.02.601741 bioRxiv.2025. 3D directional tuning in the orofacial sensorimotor cortex during natural feeding and drinking
这篇文献研究了在自然摄食和饮水行为中,口腔面部感觉运动皮层(OSMCx)对舌头三维方向的编码特性,以及口腔触觉感觉丧失对这些编码特性的影响。实验对象为两只成年雄性恒河猴,实验任务分为摄食任务和饮水任务,目标脑区是OSMCx(OSMCx是大脑皮层中负责口腔面部感觉和运动功能的区域,包括初级运动皮层(MIo)、初级体感皮层(SIo)和皮层咀嚼区(CMA))。为了研究感觉反馈在控制舌头运动中的作用,研究者使用局部注射布比卡因(Bupivacaine HCL)和肾上腺素(Epinephrine)混合液,阻断三叉神经的感觉分支,以消除口腔触觉感觉。研究者在猴子的颅骨、下颌骨和舌头上植入直径为1毫米的钽珠,用于标记,并使用双平面视频放射照相术(biplanar videoradiography)以200 Hz的频率记录标记物的三维位置数据。同时,通过慢性植入的微电极阵列记录OSMCx中初级运动皮层(MIo)和初级体感皮层(SIo)的神经元放电活动。
实验结果表明,在摄食和饮水任务中,大多数MIo和SIo神经元表现出对三维舌运动方向的调谐,其中MIo神经元的方向调谐更为显著(尤其是在摄食任务中)。感觉神经阻断后,MIo和SIo中方向调谐的神经元比例显著下降,表明触觉反馈在控制舌运动方向中起着关键作用。。
bioRxiv,
2025-4-24.
DOI: 10.1101/2024.07.02.601741
Abstract:
AbstractDirectional tongue movements are crucial for feeding and speech, ensuring proper food positioning for chewing and swallowing, as well as accurate sound production. While directional tuning in the arm region …
>>>
AbstractDirectional tongue movements are crucial for feeding and speech, ensuring proper food positioning for chewing and swallowing, as well as accurate sound production. While directional tuning in the arm region of the sensorimotor cortex during reaching tasks is well-studied, little is known about how 3D tongue direction is encoded in the orofacial sensorimotor cortex (OSMCx) during natural behaviors. Understanding this neural representation has important implications for rehabilitating individuals with orolingual dysfunctions. This study examines the directional tuning and population dynamics in OSMCx during naturalistic feeding and drinking, and how these are affected by sensory loss. Using biplanar videoradiography, we tracked implanted tongue markers in behaving rhesus macaques (Macaca mulatta) and simultaneously recorded 3D positional data with spiking activity from chronically implanted microelectrode arrays in primary motor (MIo) and somatosensory (SIo) areas of the orofacial cortex. In some sessions, tasks were preceded by bilateral nerve block injections to the sensory branches of the trigeminal nerve. Modulation to 3D tongue direction during feeding and drinking was found in most MIo and SIo neurons. Directional information in both individual- and population-level was higher in feeding and was more robust in MIo. Following sensory loss, alterations in tongue kinematics were accompanied by changes in directional information in MIo and SIo, manifesting as modifications in both individual neuron tuning characteristics and the broader dynamics of population-level neural activity. Overall, this study advances our understanding of how OSMCx contributes to complex, coordinated control of naturalistic tongue movements. It expands our current knowledge of orofacial control to three dimensions and demonstrates the specificity and adaptability of population activity in MIo and SIo in response to different behavioral contexts.
<<<
翻译
50.
龙海晨
(2025-06-01 11:58):
#paper Xu Y, Lan F, Bi Q, Li X, Wang Z, Li Y, Li P, Long H, Du L. Comprehensive analysis of the prognosis and tumor immune microenvironment of ubiquitin-conjugating enzyme transport-related gene UBE2C in hepatocellular carcinoma. Discov Oncol. 2025 May 23;16(1):884. doi: 10.1007/s12672-025-02675-0. PMID: 40410642; PMCID: PMC12102447.这是我发的第一篇纯生物信息的sci,是研究肝细胞癌中泛素结合酶E2 C(UBE2C)预后意义和肿瘤免疫反应的机制的。研究发现细胞周期相关蛋白与UBE2C基因表达之间存在很高的相关性,肝癌样品中的UBE2C基因表达水平明显高于正常样本,UBE2C表达高的肝癌患者的存活率低于低表达患者。高表达的UBE2C基因与免疫抑制分子数量增加有关。 以上研究结果因为实验条件的限制,基于纯生信分析,未能进行分子生物学实验有一定的局限性。
51.
ZĒNG Yíngzhū (Zoo) 曾莹珠
(2025-05-31 23:51):
#paper
doi:10.1136/bmjopen-2024-086602
BMJ Open
2024
Graded Exposure Therapy for Fear Avoidance Behaviour After Concussion (GET FAB): protocol for a multisite Canadian randomised controlled trial.
该研究针对轻度创伤性脑损伤(mTBI)后恐惧回避行为(fear avoidance behaviour)这一可干预的风险因素,提出了分级暴露疗法(GET)的随机对照试验方案,通过行为实验修正对症状的灾难化认知,与我做过的恐惧消退的实验室研究机制相关。
BMJ Open,
2024-6.
DOI: 10.1136/bmjopen-2024-086602
Abstract:
IntroductionPersistent symptoms after mild traumatic brain injury (mTBI) negatively affect daily functioning and quality of life. Fear avoidance behaviour, a coping style in which people avoid or escape from activities …
>>>
IntroductionPersistent symptoms after mild traumatic brain injury (mTBI) negatively affect daily functioning and quality of life. Fear avoidance behaviour, a coping style in which people avoid or escape from activities or situations that they expect will exacerbate their symptoms, maybe a particularly potent and modifiable risk factor for chronic disability after mTBI. This study will evaluate the efficacy of graded exposure therapy (GET) for reducing persistent symptoms following mTBI, with two primary aims: (1) To determine whether GET is more effective than usual care; (2) to identify for whom GET is the most effective treatment option, by evaluating whether baseline fear avoidance moderates differences between GET and an active comparator (prescribed aerobic exercise). Our findings will guide evidence-based care after mTBI and enable better matching of mTBI patients to treatments.Methods and analysisWe will conduct a multisite randomised controlled trial with three arms. Participants (n=220) will be recruited from concussion clinics and emergency departments in three Canadian provinces and randomly assigned (1:2:2 ratio) to receive enhanced usual care, GET or prescribed aerobic exercise. The outcome assessment will occur remotely 14–18 weeks following baseline assessment, after completing the 12-week treatment phase. The primary outcome will be symptom severity (Rivermead Post-concussion Symptoms Questionnaire).Ethics and disseminationInformed consent will be obtained from all participants. All study procedures were approved by the local research ethics boards (University of British Columbia Clinical Research Ethics Board, University of Calgary Conjoint Health Research Ethics Board, University Health Network Research Ethics Board—Panel D). Operational approvals were obtained for Vancouver Coastal Health Research Institute and Provincial Health Services Authority. If GET proves effective, we will disseminate the GET treatment manual and present instructional workshops for clinicians.Trial registration numberClinicalTrials.gov #NCT05365776
<<<
翻译
52.
半面阳光
(2025-05-31 23:50):
#paper https://doi.org/10.1186/s12920-025-02086-8. BMC Medical Genomics. 2025. Visualization using NIPTviewer support the clinical interpretation of noninvasive prenatal testing results. 这篇文章主要是开发了一个临床使用友好的NIPT交互软件,NIPTviewer。这是一款基于网页的应用程序,用于可视化并指导NIPT数据结果的解读,方便使用。NIPTviewer具备数据库功能以存储NIPT结果,并提供用户交互与可视化的网页界面。NIPT技术已经达到比较成熟的阶段,更便捷地应用优化似乎是当前和未来一段时间IVD研发可竞争的方向之一。
BMC Medical Genomics,
2025-1-20.
DOI: 10.1186/s12920-025-02086-8
Abstract:
Abstract Background Noninvasive prenatal testing (NIPT) is increasingly used to screen for fetal chromosomal aneuploidy by analyzing cell-free DNA (cfDNA) in peripheral maternal blood. The method provides an opportunity for …
>>>
Abstract Background Noninvasive prenatal testing (NIPT) is increasingly used to screen for fetal chromosomal aneuploidy by analyzing cell-free DNA (cfDNA) in peripheral maternal blood. The method provides an opportunity for early detection of large genetic abnormalities without an increased risk of miscarriage due to invasive procedures. Commercial applications for use at clinical laboratories often take advantage of DNA sequencing technologies and include the bioinformatic workup of the sequence data. The interpretation of the test results and the clinical report writing, however, remains the responsibility of the diagnostic laboratory. In order to facilitate this step, we developed NIPTviewer, a web-based application to visualize and guide the interpretation of NIPT data results. Results NIPTviewer has a database functionality to store the NIPT results and a web interface for user interaction and visualization. The application has been implemented as part of a novel analysis pipeline for NIPT in a diagnostic laboratory at Uppsala University Hospital. The validation data set included 84 previously analyzed plasma samples with known results regarding chromosomes 13, 18, 21, X and Y. They were sequenced in six different experiments, uploaded to NIPTviewer and assigned to a clinical laboratory geneticist for interpretation. The results of all previously analyzed samples were replicated. Conclusion NIPTviewer facilitates NIPT results interpretation and has been implemented as part of a NIPT analysis routine that was accredited by the national accreditation body for Sweden (Swedac).
<<<
翻译
53.
符毓
(2025-05-31 22:59):
#paper doi: 10.48550/arXiv.2505.21906, 2025, Vision-Language-Action Model with Open-World Embodied Reasoning from Pretrained Knowledge.
视觉-语言-动作 (VLA) 模型已成为机器人领域的下一代模型。然而,尽管现有的端到端 VLA 系统利用了强大的预训练视觉-语言模型 (VLM),但在微调过程中,由于模型需要适应特定的机器人任务,它们往往会丢失关键功能。我们认为,一个可泛化的 VLA 模型应该保留并扩展 VLM 的核心能力:1)开放世界具身推理——VLA 应该继承 VLM 的知识,即识别 VLM 能够识别的任何事物,能够解决数学问题,并具备视觉空间智能;2)推理跟随——有效地将开放世界推理转化为机器人可执行的步骤。
本文推出ChatVLA-2,通过端到端利用预训练视觉语言模型所获得的先天推理和理解能力,赋予视觉-语言-动作 (VLA) 模型执行各种任务的能力。核心贡献是在预训练的视觉语言主干之上集成了一个dynamic Mixture-of-Experts (MoE)模块。该模块可以有效地管理不同的任务需求,其中一些专家共识共享普遍的多模态特征,而其他专家则专注于特定任务的表征。此外,提出了一种两阶段训练策略:首先,引导 VLA 模型建立预训练多模态知识与机器人动作之间的联系;随后,引入推理跟踪阶段,使模型能够理解推理输出并有效地将其转化为相应的动作。
arXiv,
2025-05-28T02:48:42Z.
DOI: 10.48550/arXiv.2505.21906
Abstract:
Vision-language-action (VLA) models have emerged as the next generation ofmodels in robotics. However, despite leveraging powerful pre-trainedVision-Language Models (VLMs), existing end-to-end VLA systems often lose keycapabilities during fine-tuning as the …
>>>
Vision-language-action (VLA) models have emerged as the next generation ofmodels in robotics. However, despite leveraging powerful pre-trainedVision-Language Models (VLMs), existing end-to-end VLA systems often lose keycapabilities during fine-tuning as the model adapts to specific robotic tasks.We argue that a generalizable VLA model should retain and expand upon the VLM'score competencies: 1) Open-world embodied reasoning - the VLA should inheritthe knowledge from VLM, i.e., recognize anything that the VLM can recognize,capable of solving math problems, possessing visual-spatial intelligence, 2)Reasoning following - effectively translating the open-world reasoning intoactionable steps for the robot. In this work, we introduce ChatVLA-2, a novelmixture-of-expert VLA model coupled with a specialized three-stage trainingpipeline designed to preserve the VLM's original strengths while enablingactionable reasoning. To validate our approach, we design a math-matching taskwherein a robot interprets math problems written on a whiteboard and pickscorresponding number cards from a table to solve equations. Remarkably, ourmethod exhibits exceptional mathematical reasoning and OCR capabilities,despite these abilities not being explicitly trained within the VLA.Furthermore, we demonstrate that the VLA possesses strong spatial reasoningskills, enabling it to interpret novel directional instructions involvingpreviously unseen objects. Overall, our method showcases reasoning andcomprehension abilities that significantly surpass state-of-the-art imitationlearning methods such as OpenVLA, DexVLA, and pi-zero. This work represents asubstantial advancement toward developing truly generalizable roboticfoundation models endowed with robust reasoning capacities.
<<<
翻译
54.
庞庞
(2025-05-31 22:52):
#paper doi:https://doi.org/10.1016/j.transm.2025.100095 Evaluation of scalp-based targeting methods of DLPFC for TMS therapy。该研究系统比较四种用于抑郁症TMS治疗的颅表定位方法(EEG Cap、Beam F3、CPC F3和adjusted Beam F3),结果表明CPC F3方法展现出较为优越的综合性能,为临床治疗定位方案的选择提供了重要科学依据。
Abstract:
No abstract available.
55.
钟鸣
(2025-05-31 21:32):
#paper doi:10.1016/j.stem.2023.05.012 Organoid cultures for cancer modeling
这是一篇综述。笔记:2D肿瘤细胞培养,由于基因组的畸变和扭曲以及表观基因组的修饰,它们已经失去了对微环境的内在依赖;PDX可以弥补2D培养的不足,但成本和技术难度都较高。ASC 和 PSC 类器官因生长因子的添加方式不同。
有必要通过外显子测序等方式验证用做类器官培养的肿瘤细胞纯度,因为组织中正常细胞形成的类器官在外观上与肿瘤类器官难以区分,而前者会逐步挤占后者的生存空间,除非培养条件对后者有足够多的优势。正常细胞的污染也可通过精细挑选取样部位、增减细胞因子、传代时挑选肿瘤类器官等方式缓解。
ASC 癌症类器官和 2D 癌细胞系之间的一个主要区别是,类器官仍然依赖于利基因子和细胞-细胞/细胞-基质相互作用来生存。器官在体外有失去基础状态的趋势,而这种状态可以通过撤回培养基中的必需微环境因子来恢复,肿瘤亚型受微环境的影响大于仅受遗传学的影响。
与微生物共培养时,可观察到类器官具有与体内相似的突变,因此类器官也可在探索微生物在肿瘤发生中的作用提供方法。
类器官的另一应用在于药物筛选,包括预测药敏、体内药效,以及发现新的靶点。
Cell Stem Cell,
2023-7.
DOI: 10.1016/j.stem.2023.05.012
Abstract:
No abstract available.
56.
刘馨云
(2025-05-31 21:32):
#paper https://arxiv.org/pdf/2505.20290
人类通过观察他人来学习新任务。受到这一点启发,我们提出了 EgoZero 框架,它可以从人类佩戴智能眼镜拍摄的第三人称视频中学习闭环机器人策略。智能眼镜能够捕捉人类交互的丰富多模态第一人称视角:RGB 视频记录周围场景,IMU(惯性测量单元)提供头部运动信息,麦克风则记录对话和环境声音。我们的方法仅通过观察这些第一人称视频来学习如何行动,无需任何机器人演示。当给定一个人类完成任务的视频时,EgoZero 能预测一系列中间目标和语言子目标,并据此在真实机器人上以闭环方式执行任务。EgoZero 将人类观察压缩为与机器人形态无关的状态表示,这些表示可用于决策和闭环控制。所学策略在不同的机器人形态、环境和任务之间表现出良好的泛化能力。我们在真实的 Franka Panda 机械臂上进行了验证,结果表明 EgoZero 能以 70% 的零样本成功率完成多种具有挑战性的操控任务,每项任务仅需 20 分钟的数据采集时间。
arXiv,
2025-05-26T17:59:17Z.
DOI: 10.48550/arXiv.2505.20290
Abstract:
Despite recent progress in general purpose robotics, robot policies still lagfar behind basic human capabilities in the real world. Humans interactconstantly with the physical world, yet this rich data resource …
>>>
Despite recent progress in general purpose robotics, robot policies still lagfar behind basic human capabilities in the real world. Humans interactconstantly with the physical world, yet this rich data resource remains largelyuntapped in robot learning. We propose EgoZero, a minimal system that learnsrobust manipulation policies from human demonstrations captured with ProjectAria smart glasses, $\textbf{and zero robot data}$. EgoZero enables: (1)extraction of complete, robot-executable actions from in-the-wild, egocentric,human demonstrations, (2) compression of human visual observations intomorphology-agnostic state representations, and (3) closed-loop policy learningthat generalizes morphologically, spatially, and semantically. We deployEgoZero policies on a gripper Franka Panda robot and demonstrate zero-shottransfer with 70% success rate over 7 manipulation tasks and only 20 minutes ofdata collection per task. Our results suggest that in-the-wild human data canserve as a scalable foundation for real-world robot learning - paving the waytoward a future of abundant, diverse, and naturalistic training data forrobots. Code and videos are available at https://egozero-robot.github.io.
<<<
翻译
57.
尹志
(2025-05-31 21:23):
#paper https://doi.org/10.48550/arXiv.2012.07436 Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting。这是AAAI2021上的一篇关于长序列时序建模的经典工作。文章对传统Transformer进行了改进,提出了一类新的模型Informer,通过对self attention的改进和蒸馏,以及generative style decoder的构建,在时间复杂度、空间复杂度上都改善了传统Transformer存在的问题。该工作在多个数据集上取得了良好的性能。上述的几个思路在后续的时序建模中被频繁使用,非常具有启发性。
arXiv,
2020-12-14T11:43:09Z.
DOI: 10.48550/arXiv.2012.07436
Abstract:
Many real-world applications require the prediction of long sequencetime-series, such as electricity consumption planning. Long sequencetime-series forecasting (LSTF) demands a high prediction capacity of the model,which is the ability to …
>>>
Many real-world applications require the prediction of long sequencetime-series, such as electricity consumption planning. Long sequencetime-series forecasting (LSTF) demands a high prediction capacity of the model,which is the ability to capture precise long-range dependency coupling betweenoutput and input efficiently. Recent studies have shown the potential ofTransformer to increase the prediction capacity. However, there are severalsevere issues with Transformer that prevent it from being directly applicableto LSTF, including quadratic time complexity, high memory usage, and inherentlimitation of the encoder-decoder architecture. To address these issues, wedesign an efficient transformer-based model for LSTF, named Informer, withthree distinctive characteristics: (i) a $ProbSparse$ self-attention mechanism,which achieves $O(L \log L)$ in time complexity and memory usage, and hascomparable performance on sequences' dependency alignment. (ii) theself-attention distilling highlights dominating attention by halving cascadinglayer input, and efficiently handles extreme long input sequences. (iii) thegenerative style decoder, while conceptually simple, predicts the longtime-series sequences at one forward operation rather than a step-by-step way,which drastically improves the inference speed of long-sequence predictions.Extensive experiments on four large-scale datasets demonstrate that Informersignificantly outperforms existing methods and provides a new solution to theLSTF problem.
<<<
翻译
58.
哪有情可长
(2025-05-31 17:15):
#paper Abscisic acid receptors functionally converge across 500 million years of land plant evolution, Current Biology, 24 February 2025,https://doi.org/10.1016/j.cub.2024.12.043. 在陆地植物早期,由于祖先型脱落酸ABA受体PYRABACTIN RESISTANCE 1-like蛋白(PYLs)具有较高的本底活性可直接和2C型蛋白磷酸酶(PP2Cs)互作,限制了植物对ABA感知范围;在被子植物中,ABA受体家族呈现显著的多样性,形成的二聚体实现了对ABA激素的低背景响应。作者发现苔藓类植物中的ABA受体PYL源于一个独立的扩张,产生了三个亚科。后续利用酵母双杂交和体外PP2C抑制试验证明,藓类植物的PYL受体具有较低的基础活性。色谱分析和相关的生化实验证据表明,藓类植物PYL受体的蛋白主要以单体形式存在。突变分析与生理生化分析相结合表明,藓类植物的PYL的活性调节可能是由于门环结构中的关键氨基酸变异引起。这一研究揭示了苔藓植物在进化中,PYL结构上表现出差异;藓类植物进化出了单体、低信号活性的ABA感知装置。因此,苔藓植物经历了独立的进化,进化出了相同的低背景下的ABA激素的感知能力,只在需要时才能激活相关通路从而优化了对ABA感知范围幅度。
59.
孤舟蓑笠翁
(2025-05-30 22:24):
#paper 【doi】10.1038/s41586-025-09053-4;【发表年份】2025年;【期刊】Nature;【标题】Cross-tissue multicellular coordination and its rewiring in cancer。这篇论文研究了人体组织中不同细胞类型如何协同工作维持健康,以及这种协同在癌症中如何被破坏。科学家们整合了35种健康组织和29种癌症类型的单细胞数据,开发了CoVarNet计算工具,发现了12种跨组织的"细胞模块"(CMs)——这些是经常一起出现的细胞组合。比如在脾脏中,某些免疫细胞模块会随着年龄增长而改变;在乳腺中,特定成纤维细胞模块与绝经相关。更惊人的是,癌症会破坏健康组织的细胞模块,同时形成新的、促进肿瘤生长的模块。通过分析这些模块,研究者还找到了可能用于癌症早期诊断的分子标志物。简单来说,这项工作就像绘制了一幅"细胞社交网络"地图,展示了健康时细胞如何团队合作,而癌症时这些团队如何重组——这为理解疾病机制和开发新疗法提供了重要线索。
Abstract:
The multicellular coordination that underlies tissue homeostasis and disease progression is of fundamental interest. However, how diverse cell types are organized within tissue niches for cohesive functioning remains largely unknown. …
>>>
The multicellular coordination that underlies tissue homeostasis and disease progression is of fundamental interest. However, how diverse cell types are organized within tissue niches for cohesive functioning remains largely unknown. Here we systematically characterized cross-tissue coordinated cellular modules in healthy tissues, uncovering their spatiotemporal dynamics and phenotypic associations, and examined their rewiring in cancer. We first compiled a comprehensive single-cell transcriptomic atlas from 35 human tissues, revealing substantial inter-tissue variability in cellular composition. By leveraging covariance in cellular abundance, we identified 12 cellular modules with distinct cellular compositions, tissue prevalences and spatial organizations, and demonstrated coordinated intercellular communication within cellular modules using in situ spatial and in vivo perturbation data. Among them, two immune cellular modules in the spleen showed contrasting chronological dynamics with ageing. Analysis of multicellular changes in the breast revealed a menopausal trajectory associated with fibroblast dynamics. Furthermore, interrogation across cancer types uncovered simultaneous rewiring of two types of multicellular ecosystem during tumour progression, including the loss of tissue-specific healthy organization and the emergence of a convergent cancerous ecosystem. These findings reveal fundamental organizing principles of multicellular ecosystems in health and cancer, laying a foundation for further investigations into tissue-level functional coordination across diverse contexts.
<<<
翻译
60.
李翛然
(2025-05-30 21:13):
#paper NMRExtractor: leveraging large language models to
construct an experimental NMR database from
open-source scienti c publications† DOI: 10.1039/D4SC08802F
中国科学院上海药物研究所郑明月团队开发了NMRExtractor工具,基于微调的大型语言模型(Mistral-7b)从570万篇PubMed开源文献中自动提取实验核磁共振(NMR)数据,构建了目前规模最大的公开NMR数据库NMRBank,包含225,809条高质量记录。
1. 高效提取流程
◦ 通过正则表达式筛选含NMR的段落(380,220条),利用LLM精准提取化合物IUPAC名称、1H/13C NMR化学位移及实验条件。
◦ 引入置信度评分机制(0-1分),高置信度(>0.8)数据准确率达97%,媲美人工标注水平。
最近在看各种仪器数据处理
Chemical Science,
2025.
DOI: 10.1039/D4SC08802F
Abstract:
NMRExtractor is a large language model-powered pipeline that automatically extracts experimental NMR data from massive open-access publications, resulting in the construction of NMRBank—the largest open-access NMR dataset available to date.
>>>
NMRExtractor is a large language model-powered pipeline that automatically extracts experimental NMR data from massive open-access publications, resulting in the construction of NMRBank—the largest open-access NMR dataset available to date.
<<<
翻译