Assessing the possibility of using large language models in ocular surface diseases

AIM:To assess the possibility of using different large language models(LLMs)in ocular surface diseases by selecting five different LLMS to test their accuracy in answering specialized questions related to ocular surface diseases:ChatGPT-4,ChatGPT-3.5,Claude 2,PaLM2,and SenseNova.METHODS:A group of experienced ophthalmology professors were asked to develop a 100-question singlechoice question on ocular surface diseases designed to assess the performance of LLMs and human participants in answering ophthalmology specialty exam questions.The exam includes questions on the following topics:keratitis disease(20 questions),keratoconus,keratomalaciac,corneal dystrophy,corneal degeneration,erosive corneal ulcers,and corneal lesions associated with systemic diseases(20 questions),conjunctivitis disease(20 questions),trachoma,pterygoid and conjunctival tumor diseases(20 questions),and dry eye disease(20 questions).Then the total score of each LLMs and compared their mean score,mean correlation,variance,and confidence were calculated.RESULTS:GPT-4 exhibited the highest performance in terms of LLMs.Comparing the average scores of the LLMs group with the four human groups,chief physician,attending physician,regular trainee,and graduate student,it was found that except for ChatGPT-4,the total score of the rest of the LLMs is lower than that of the graduate student group,which had the lowest score in the human group.Both ChatGPT-4 and PaLM2 were more likely to give exact and correct answers,giving very little chance of an incorrect answer.ChatGPT-4 showed higher credibility when answering questions,with a success rate of 59%,but gave the wrong answer to the question 28% of the time.CONCLUSION:GPT-4 model exhibits excellent performance in both answer relevance and confidence.PaLM2 shows a positive correlation(up to 0.8)in terms of answer accuracy during the exam.In terms of answer confidence,PaLM2 is second only to GPT4 and surpasses Claude 2,SenseNova,and GPT-3.5.Despite the fact that ocular surface disease is a highly specialized discipline,GPT-4 still exhibits superior performance,suggesting that its potential and ability to be applied in this field is enormous,perhaps with the potential to be a valuable resource for medical students and clinicians in the future.

Evaluating large language models as patient education tools for inflammatory bowel disease:A comparative study

BACKGROUND Inflammatory bowel disease(IBD)is a global health burden that affects millions of individuals worldwide,necessitating extensive patient education.Large language models(LLMs)hold promise for addressing patient information needs.However,LLM use to deliver accurate and comprehensible IBD-related medical information has yet to be thoroughly investigated.AIM To assess the utility of three LLMs(ChatGPT-4.0,Claude-3-Opus,and Gemini-1.5-Pro)as a reference point for patients with IBD.METHODS In this comparative study,two gastroenterology experts generated 15 IBD-related questions that reflected common patient concerns.These questions were used to evaluate the performance of the three LLMs.The answers provided by each model were independently assessed by three IBD-related medical experts using a Likert scale focusing on accuracy,comprehensibility,and correlation.Simultaneously,three patients were invited to evaluate the comprehensibility of their answers.Finally,a readability assessment was performed.RESULTS Overall,each of the LLMs achieved satisfactory levels of accuracy,comprehensibility,and completeness when answering IBD-related questions,although their performance varies.All of the investigated models demonstrated strengths in providing basic disease information such as IBD definition as well as its common symptoms and diagnostic methods.Nevertheless,when dealing with more complex medical advice,such as medication side effects,dietary adjustments,and complication risks,the quality of answers was inconsistent between the LLMs.Notably,Claude-3-Opus generated answers with better readability than the other two models.CONCLUSION LLMs have the potential as educational tools for patients with IBD;however,there are discrepancies between the models.Further optimization and the development of specialized models are necessary to ensure the accuracy and safety of the information provided.

Whole-body magnetic resonance imaging provides accurate staging of diffuse large B-cell lymphoma, but is less preferred by patients

BACKGROUND Whole-body magnetic resonance imaging(wbMRI)allows general assessment of systemic cancers including lymphomas without radiation burden.AIM To evaluate the diagnostic performance of wbMRI in the staging of diffuse large B-cell lymphoma(DLBCL),determine the value of individual MRI sequences,and assess patients’concerns with wbMRI.METHODS In this single-center prospective study,adult patients newly diagnosed with systemic DLBCL underwent wbMRI on a 3T scanner[diffusion weighted images with background suppression(DWIBS),T2,short tau inversion recovery(STIR),contrast-enhanced T1]and fluorodeoxyglucose(18F-FDG)positron emission tomo-graphy/computed tomography(PET/CT)(reference standard).The involvement of 12 nodal regions and extranodal sites was evaluated on wbMRI and PET/CT.The utility of wbMRI sequences was rated on a five-point scale(0=not useful,4=very useful).Patients received a questionnaire regarding wbMRI.RESULTS Of 60 eligible patients,14(23%)were enrolled and completed the study.The sensitivity of wbMRI in the nodal involvement(182 nodal sites)was 0.84,with 0.99 specificity,positive predictive value of 0.96,negative predictive value of 0.97,and 0.97 accuracy.PET/CT and wbMRI were concordant both in extranodal involvement(13 instances)and staging(κ=1.0).The mean scores of the utility of MRI sequences were 3.71±0.73 for DWIBS,2.64±0.84 for T1,2.14±0.77 for STIR,and 1.29±0.73 for T2(P<0.0001).Patients were mostly concerned about the enclosed environment and duration of the MRI examination(27%of patients).CONCLUSION The wbMRI exhibited excellent sensitivity and specificity in staging DLBCL.DWIBS and contrast-enhanced T1 were rated as the most useful sequences.Patients were less willing to undergo wbMRI as a second examination parallel to PET/CT,especially owing to the long duration and the enclosed environment.

Evolution and Prospects of Foundation Models: From Large Language Models to Large Multimodal Models

Since the 1950s,when the Turing Test was introduced,there has been notable progress in machine language intelligence.Language modeling,crucial for AI development,has evolved from statistical to neural models over the last two decades.Recently,transformer-based Pre-trained Language Models(PLM)have excelled in Natural Language Processing(NLP)tasks by leveraging large-scale training corpora.Increasing the scale of these models enhances performance significantly,introducing abilities like context learning that smaller models lack.The advancement in Large Language Models,exemplified by the development of ChatGPT,has made significant impacts both academically and industrially,capturing widespread societal interest.This survey provides an overview of the development and prospects from Large Language Models(LLM)to Large Multimodal Models(LMM).It first discusses the contributions and technological advancements of LLMs in the field of natural language processing,especially in text generation and language understanding.Then,it turns to the discussion of LMMs,which integrates various data modalities such as text,images,and sound,demonstrating advanced capabilities in understanding and generating cross-modal content,paving new pathways for the adaptability and flexibility of AI systems.Finally,the survey highlights the prospects of LMMs in terms of technological development and application potential,while also pointing out challenges in data integration,cross-modal understanding accuracy,providing a comprehensive perspective on the latest developments in this field.

Frontogenesis and Frontolysis of a Cold Filament Driven by the Cross-Filament Wind and Wave Fields Simulated by a Large Eddy Simulation

The variations of the frontogenetic trend of a cold filament induced by the cross-filament wind and wave fields are studied by a non-hydrostatic large eddy simulation. Five cases with different strengths of wind and wave fields are studied.The results show that the intense wind and wave fields further break the symmetries of submesoscale flow fields and suppress the levels of filament frontogenesis. The changes of secondary circulation directions—that is, the conversion between the convergence and divergence of the surface cross-filament currents with the downwelling and upwelling jets in the filament center—are associated with the inertial oscillation. The filament frontogenesis and frontolysis caused by the changes of secondary circulation directions may periodically sharpen and smooth the gradient of submesoscale flow fields.The lifecycle of the cold filament may include multiple stages of filament frontogenesis and frontolysis.

Large animal models for Huntington's disease research

Huntington'sdisease(HD)isahereditary neurodegenerative disorder for which there is currently no effectivetreatmentavailable.Consequently,the development of appropriate disease models is critical to thoroughly investigate disease progression.The genetic basis of HD involves the abnormal expansion of CAG repeats in the huntingtin(HTT)gene,leading to the expansion of a polyglutamine repeat in the HTT protein.Mutant HTT carrying the expanded polyglutamine repeat undergoes misfolding and forms aggregates in the brain,which precipitate selective neuronal loss in specific brain regions.Animal models play an important role in elucidating the pathogenesis of neurodegenerative disorders such as HD and in identifying potential therapeutic targets.Due to the marked species differences between rodents and larger animals,substantial efforts have been directed toward establishing large animal models for HD research.These models are pivotal for advancing the discovery of novel therapeutic targets,enhancing effective drug delivery methods,and improving treatment outcomes.We have explored the advantages of utilizing large animal models,particularly pigs,in previous reviews.Since then,however,significant progress has been made in developing more sophisticated animal models that faithfully replicate the typical pathology of HD.In the current review,we provide a comprehensive overview of large animal models of HD,incorporating recent findings regarding the establishment of HD knock-in(KI)pigs and their genetic therapy.We also explore the utilization of large animal models in HD research,with a focus on sheep,non-human primates(NHPs),and pigs.Our objective is to provide valuable insights into the application of these large animal models for the investigation and treatment of neurodegenerative disorders.

Evaluating Privacy Leakage and Memorization Attacks on Large Language Models (LLMs) in Generative AI Applications

The recent interest in the deployment of Generative AI applications that use large language models (LLMs) has brought to the forefront significant privacy concerns, notably the leakage of Personally Identifiable Information (PII) and other confidential or protected information that may have been memorized during training, specifically during a fine-tuning or customization process. We describe different black-box attacks from potential adversaries and study their impact on the amount and type of information that may be recovered from commonly used and deployed LLMs. Our research investigates the relationship between PII leakage, memorization, and factors such as model size, architecture, and the nature of attacks employed. The study utilizes two broad categories of attacks: PII leakage-focused attacks (auto-completion and extraction attacks) and memorization-focused attacks (various membership inference attacks). The findings from these investigations are quantified using an array of evaluative metrics, providing a detailed understanding of LLM vulnerabilities and the effectiveness of different attacks.

The influence of inter-band rock on rib spalling in longwall panel with large mining height

In order to improve rib stability,failure criteria and instability mode of a thick coal seam with inter-band rock layer are analysed in this study.A three-dimensional mechanical model is established for the rib by considering the rock layer.A safety factor is defined foy the rib,and it is observed that the safety factor exhibits a positive correlation with the thickness and strength of the inter-band rock.A calculation method for determining critical parameters of the rock layer is presented to ensure the rib stability.It is revealed that incomplete propagation of the fracture at the hard rock constitutes a fundamental prerequisite for ensuring the rib stability.The influence of the position of the inter-band rock in the coal seam on failure mechanism of the rib was thoroughly investigated by developing a series of physical models for the rib at the face area.The best position for the inter-band rock in the coal seam is at a height of 1.5 m away from the roof line,which tends to provide a good stability state for the rib.For different inter-band rock positions,two ways of controlling rib by increasing supports stiffness and flexible grouting reinforcement are proposed.

Plume-lithosphere interaction in the Comei Large Igneous Province: Evidence from two types of mafic dykes in Gyangze, south Tibet, China

Two suites of mafic dykes,T1193-A and T1194-A,outcrop in Gyangze area,southeast Tibet.They are in the area of Comei LIP and have indistinguishable field occurrences with two other dykes in Gyangze,T0902 dyke with 137.7±1.3 Ma zircon age and T0907 dyke with 142±1.4 Ma zircon age reported by Wang YY et al.(2016),indicating coeval formation time.Taking all the four diabase dykes into consideration,two different types,OIB-type and weak enriched-type,can be summarized.The“OIB-type”samples,including T1193-A and T0907 dykes,show OIB-like geochemical features and have initial Sr-Nd isotopic values similar with most mafic products in Comei Large Igneous Provinces(LIP),suggesting that they represent melts directly generated from the Kerguelen mantle plume.The“weak enriched-type”samples,including T1194-A and T0902 dykes,have REEs and trace element patterns showing withinplate affinity but have obvious Nb-Ta-Ti negative anomalies.They show uniform lowerεNd(t)values(−6‒−2)and higher 87Sr/86Sr(t)values(0.706‒0.709)independent of their MgO variation,indicating one enriched mantle source.Considering their closely spatial and temporal relationship with the widespread Comei LIP magmatic products in Tethyan Himalaya,these“weak enriched-type”samples are consistent with mixing of melts from mantle plume and the above ancient Tethyan Himalaya subcontinental lithospheric mantle(SCLM)in different proportions.These weak enriched mafic rocks in Comei LIP form one special rock group and most likely suggest large scale hot mantle plume-continental lithosphere interaction.This process may lead to strong modification of the Tethyan Himalaya lithosphere in the Early Cretaceous.

Research status and application of artificial intelligence large models in the oil and gas industry

This article elucidates the concept of large model technology,summarizes the research status of large model technology both domestically and internationally,provides an overview of the application status of large models in vertical industries,outlines the challenges and issues confronted in applying large models in the oil and gas sector,and offers prospects for the application of large models in the oil and gas industry.The existing large models can be briefly divided into three categories:large language models,visual large models,and multimodal large models.The application of large models in the oil and gas industry is still in its infancy.Based on open-source large language models,some oil and gas enterprises have released large language model products using methods like fine-tuning and retrieval augmented generation.Scholars have attempted to develop scenario-specific models for oil and gas operations by using visual/multimodal foundation models.A few researchers have constructed pre-trained foundation models for seismic data processing and interpretation,as well as core analysis.The application of large models in the oil and gas industry faces challenges such as current data quantity and quality being difficult to support the training of large models,high research and development costs,and poor algorithm autonomy and control.The application of large models should be guided by the needs of oil and gas business,taking the application of large models as an opportunity to improve data lifecycle management,enhance data governance capabilities,promote the construction of computing power,strengthen the construction of“artificial intelligence+energy”composite teams,and boost the autonomy and control of large model technology.

Construction and preliminary application of large language model for reservoir performance analysis

A large language model(LLM)is constructed to address the sophisticated demands of data retrieval and analysis,detailed well profiling,computation of key technical indicators,and the solutions to complex problems in reservoir performance analysis(RPA).The LLM is constructed for RPA scenarios with incremental pre-training,fine-tuning,and functional subsystems coupling.Functional subsystem and efficient coupling methods are proposed based on named entity recognition(NER),tool invocation,and Text-to-SQL construction,all aimed at resolving pivotal challenges in developing the specific application of LLMs for RDA.This study conducted a detailed accuracy test on feature extraction models,tool classification models,data retrieval models and analysis recommendation models.The results indicate that these models have demonstrated good performance in various key aspects of reservoir dynamic analysis.The research takes some injection and production well groups in the PK3 Block of the Daqing Oilfield as an example for testing.Testing results show that our model has significant potential and practical value in assisting reservoir engineers with RDA.The research results provide a powerful support to the application of LLM in reservoir performance analysis.

Gas encapsulation technology for large volume press

For samples in the gaseous state at room temperature and ambient pressure,mature technology has been developed to encapsulate them in a diamond anvil cell(DAC).However,the large volume press(LVP)can only treat samples with starting materials in solid or liquid form.We have achieved stable encapsulation and reaction treatment of carbon dioxide in a centimeter sized sample chamber for a long time(over 10 min)under conditions of temperature higher than 1200℃ and pressure over 5 GPa through the use of integrated low-temperature freezing and rapid compression sealing method for LVP cell assemblies.This technology can also be applied to the packaging of other gaseous or liquid samples,such as ammonia,sulfur dioxide,water,etc.in LVP devices.

ASYMPTOTIC BEHAVIOR NEAR THE BOUNDARY OF A LARGE SOLUTION TO SEMILINEAR POISSON EQUATION WITH DOUBLE-POWER NONLINEARITY

We deal with a large solution to the semilinear Poisson equation with doublepower nonlinearityΔ^(u)=u^(p)+αu^(q)in a bounded smooth domain D■R^(n),where p>1,-1<q<p andα∈R.We obtain the asymptotic behavior of a solution u near the boundary OD up to the third or higher term.

Optimization of Fixture Number in Large Thin-Walled Parts Assembly Based on IPSO

There are lots of researches on fixture layout optimization for large thin-walled parts.Current researches focus on the positioning problem,i.e.,optimizing the positions of a constant number of fixtures.However,how to determine the number of fixtures is ignored.In most cases,the number of fixtures located on large thin-walled parts is determined based on engineering experience,which leads to huge fixture number and extra waste.Therefore,this paper constructs an optimization model to minimize the number of fixtures.The constraints are set in the optimization model to ensure that the part deformation is within the surface profile tolerance.In addition,the assembly gap between two parts is also controlled.To conduct the optimization,this paper develops an improved particle swarm optimization(IPSO)algorithm by integrating the shrinkage factor and adaptive inertia weight.In the algorithm,particles are encoded according to the fixture position.Each dimension of the particle is assigned to a sub-region by constraining the optional position range of each fixture to improve the optimization efficiency.Finally,a case study on ship curved panel assembly is provided to prove that our method can optimize the number of fixtures while meeting the assembly quality requirements.This research proposes a method to optimize the number of fixtures,which can reduce the number of fixtures and achieve deformation control at the same time.

Improvement in Tol2 transposon for efficient large-cargo capacity transgene applications in cultured cells and zebrafish(Danio rerio)

Most viruses and transposons serve as effective carriers for the introduction of foreign DNA up to 11 kb into vertebrate genomes.However,their activity markedly diminishes with payloads exceeding 11 kb.Expanding the payload capacity of transposons could facilitate more sophisticated cargo designs,improving the regulation of expression and minimizing mutagenic risks associated with molecular therapeutics,metabolic engineering,and transgenic animal production.In this study,we improved the Tol2 transposon by increasing protein expression levels using a translational enhancer(QBI SP163,ST)and enhanced the nuclear targeting ability using the nuclear localization protein H2B(SHT).The modified Tol2 and ST transposon efficiently integrated large DNA cargos into human cell cultures(H1299),comparable to the well-established super PiggyBac system.Furthermore,mRNA from ST and SHT showed a significant increase in transgene delivery efficiency of large DNA payloads(8 kb,14 kb,and 24 kb)into zebrafish(Danio rerio).This study presents a modified Tol2 transposon as an enhanced nonviral vector for the delivery of large DNA payloads in transgenic applications.

Thin-walled and large-sized magnesium alloy die castings for passenger car cockpit:Application,materials,and manufacture

In order to effectively reduce energy consumption and increase range mile,new energy vehicles represented by Tesla have greatly aroused the application of integrated magnesium(Mg)alloy die casting technology in automobiles.Previously,the application of Mg alloys in automobiles,especially in automotive cockpit components,is quite extensive,while it has almost disappeared for a period of time due to its relatively high cost,causing a certain degree of information loss in the application technology of Mg alloy parts in automobiles.The rapid development of automotive technology has led to a higher requirement for the automotive components compared with those traditional one.Therefore,whatever the components themselves,or the Mg alloy materials and die casting process have to face an increasing challenge,needing to be upgraded.In addition,owing to its high integration characteristics,the application of Mg alloy die casting technology in large-sized and thin-walled automotive parts has inherent advantages and needs to be expanded urgently.Indeed,it necessitates exploring advance Mg alloys and new product structures and optimizing die casting processes.This article summarizes and analyzes the development status of thin-walled and large-sized die casting Mg alloy parts in passenger car cockpit and corresponding material selection methods,die casting processes as well as mold design techniques.Furthermore,this work will aid researchers in establishing a comprehensive understanding of the manufacture of thin-walled and large-sized die casting Mg alloy parts in automobile cockpit.It will also assist them in developing new Mg alloys with improved comprehensive performance and new processes to meet the high requirements for die casting automotive components.

Shallow Convection Dataset Simulated by Three Different Large Eddy Models

Shallow convection plays an important role in transporting heat and moisture from the near-surface to higher altitudes,yet its parameterization in numerical models remains a great challenge,partly due to the lack of high-resolution observations.This study describes a large eddy simulation(LES)dataset for four shallow convection cases that differ primarily in inversion strength,which can be used as a surrogate for real data.To reduce the uncertainty in LES modeling,three different large eddy models were used,including SAM(System for Atmospheric Modeling),WRF(Weather Research and Forecasting model),and UCLA-LES.Results show that the different models generally exhibit similar behavior for each shallow convection case,despite some differences in the details of the convective structure.In addition to grid-averaged fields,conditionally sampled variables,such as in-cloud moisture and vertical velocity,are also provided,which are indispensable for calculation of the entrainment/detrainment rate.Considering the essentiality of the entraining/detraining process in the parameterization of cumulus convection,the dataset presented in this study is potentially useful for validation and improvement of the parameterization of shallow convection.

The Immense Impact of Reverse Edges on Large Hierarchical Networks

Hierarchical networks are frequently encountered in animal groups,gene networks,and artificial engineering systems such as multiple robots,unmanned vehicle systems,smart grids,wind farm networks,and so forth.The structure of a large directed hierarchical network is often strongly influenced by reverse edges from lower-to higher-level nodes,such as lagging birds’howl in a flock or the opinions of lowerlevel individuals feeding back to higher-level ones in a social group.This study reveals that,for most large-scale real hierarchical networks,the majority of the reverse edges do not affect the synchronization process of the entire network;the synchronization process is influenced only by a small part of these reverse edges along specific paths.More surprisingly,a single effective reverse edge can slow down the synchronization of a huge hierarchical network by over 60%.The effect of such edges depends not on the network size but only on the average in-degree of the involved subnetwork.The overwhelming majority of active reverse edges turn out to have some kind of“bunching”effect on the information flows of hierarchical networks,which slows down synchronization processes.This finding refines the current understanding of the role of reverse edges in many natural,social,and engineering hierarchical networks,which might be beneficial for precisely tuning the synchronization rhythms of these networks.Our study also proposes an effective way to attack a hierarchical network by adding a malicious reverse edge to it and provides some guidance for protecting a network by screening out the specific small proportion of vulnerable nodes.

Acetylacetone-TiO_(2) Promoted Large Area Compatible Cascade Electron Transport Bilayer for Efficient Perovskite Solar Cells

In designing efficient perovskite solar cells(PSCs),the selection of suitable electron transport layers(ETLs)is critical to the final device performance as they determine the driving force for selective charge extraction.SnO_(2)nanoparticles(NPs)based ETLs have been a popular choice for PSCs due to superior electron mobility,but their relatively deep-lying conduction band energy levels(ECB)result in substantial potential loss.Meanwhile,TiO_(2)NPs establish favorable band alignment owing to shallower ECB,but their low intrinsic mobility and abundant surface trap sites impede the final performance.For this reason,constructing a cascaded bilayer ETL is highly desirable for efficient PSCs,as it can rearrange energy levels and exploit on advantages of an individual ETL.In this study,we prepare SnO_(2)NPs and acetylacetone-modified TiO_(2)(Acac-TiO_(2))NPs and implement them as bilayer SnO_(2)/Acac-TiO_(2)(BST)ETL,to assemble cascaded energy band structure.SnO_(2)contributes to rapid charge carrier transport from high electron mobility while Acac-TiO_(2)minimizes band-offset and effectively suppresses interfacial recombination.Accordingly,the optimized BST ETL generates synergistic influence and delivers power conversion efficiency(PCE)as high as 23.14%with open-circuit voltage(V_(oc))reaching 1.14 V.Furthermore,the BST ETL is transferred to a large scale and the corresponding mini module demonstrates peak performance of 18.39%PCE from 25 cm^(2)aperture area.Finally,the BST-based mini module exhibit excellent stability,maintaining 83.1%of its initial efficiency after 1000 h under simultaneous 1 Sun light-soaking and damp heat(85℃/RH 85%)environment.

THE OPTIMAL LARGE TIME BEHAVIOR OF3D QUASILINEAR HYPERBOLIC EQUATIONS WITH NONLINEAR DAMPING

We are concerned with the large-time behavior of 3D quasilinear hyperbolic equations with nonlinear damping.The main novelty of this paper is two-fold.First,we prove the optimal decay rates of the second and third order spatial derivatives of the solution,which are the same as those of the heat equation,and in particular,are faster than ones of previous related works.Second,for well-chosen initial data,we also show that the lower optimal L^(2) convergence rate of the k(∈[0,3])-order spatial derivatives of the solution is(1+t)^(-(2+2k)/4).Therefore,our decay rates are optimal in this sense.The proofs are based on the Fourier splitting method,low-frequency and high-frequency decomposition,and delicate energy estimates.