Action Of Agents Research Articles

A Dynamic-Bayesian-Network-Based Approach to Predict Immediate Future Action of an Intelligent Agent

Predicting immediate future actions taken by an intelligent agent is considered an essential problem inhuman-autonomy teaming (HAT) in many fields, such as industries and transportation, particularly toimprove human comprehension of the agent as their non-human counterpart. Moreover, the results of suchpredictions can shorten the human response time to gain control back from their non-human counterpartwhen it is required. An example case of HAT that can be benefitted from the action predictor is partiallyautomated driving with the autopilot agent as the intelligent agent. Hence, this research aims to develop anapproach to predict the immediate future actions of an intelligent agent with partially automated drivingas the experimental case. The proposed approach relies on a machine learning method called naive Bayesto develop an action classifier, and the Dynamic Bayesian Network (DBN) as the action predictor. Theautonomous driving simulation software called Carla is used for the simulation. The results show that theproposed approach is applicable to predict an intelligent agent’s three-second time-window immediate futureaction.

Synthesis, characterization, biological applications of novel coumarine azo-dye ligand and its transition metal complexes: Simultaneous electrochemical detection of paracetamol and uric acid

Coumarin is a well-known pharmacophore and structure with wide-ranging applications. It is a member of the heterocyclic family. Furthermore, metal complexes can improve the drug action and efficacy of organic therapeutic agents. Various physical and spectroscopic procedures are used to synthesize and characterize the novel azo dye ligand (HDM) and its transition metal complexes for their geometrical composition. The well diffusion method and DPPH technique were used to measure the antimicrobial potency and antioxidant levels, respectively. MTT analysis was performed to assess anticancer activity for Hela and P53 cell lines. CV and DPV were used to confirm the electro catalytic ability of [Co(HDM)2(H2O)2]/GCE (modified electrode) towards various concentrations of paracetamol and uric acid. The modified electrode performed well as a sensor with paracetamol and uric acid concentration of CV 10 to 80 µmol/L and 10 to 100 µmol/L; DPV 5 to 20 µmol/L and 5 to 20 µmol/L respectively and LOD values CV 3.333, 3.333 µmol/L; DPV 1.666, 1.666 µmol/L. This electrode has a sensitivity of CV 0.02287; DPV 0.0137 µAµM−1 for paracetamol and CV 7.558; DPV 0.01674 µAµM−1 for uric acid. The results showed that, when compared to bare glassy carbon electrode, the modified electrode had a good electro catalytic activities, high selectivity and sensitivity to the redox reaction of paracetamol and uric acid.

Rule-based shields embedded safe reinforcement learning approach for electric vehicle charging control

The optimal and secure electric vehicles charging control can be formulated as a large-scale partially-observable constrained Markov decision process (PO-CMDP) problem with high levels of security risks and uncertainties. Such a problem is very difficult to be handled by optimization-theory-based methods. Data-driven methods, especially reinforcement learning (RL), are suitable for handling uncertainties but are poor in ensuring safety, which is unacceptable in power systems. Motivated by the fact that humans could be supervised by an expert when learning a new skill involving risk, this paper proposes a safe RL framework that embeds rule-based local and global shields in the loop of RL for supervising the actions of agents. The proposed framework not only strictly guarantees local and global security in the training and execution phases, but also helps the agent to find a more near-optimal policy. The effectiveness and efficiency are demonstrated by comparison with multiple baseline methods in the IEEE-33 node system.

Religious Differences and Imperial Pragmatism in a Polemical Arena: A Privileged Law for Muslims, Hindus, and Jains in Diu (1557)

Abstract In 1557 Francisco Barreto, the Portuguese governor of the State of India, issued a decree forbidding the destruction of temples and books belonging to the Gujarati Muslims, Hindus, and Jains of Diu and allowed them to practice their religion freely. This and other reforms introduced by the crown and the Goan archbishop created a special situation for Muslims and Hindus living in Diu, in contrast to the policies designed to establish monolithic confessionalism that the Portuguese authorities had been attempting to impose in most part of Asia since around 1540. This article aims to reconstruct and explain the specific nature of the religious policies pursued in Diu, adopting a holistic approach that compares them with developments in other areas of Portuguese Asia. It will be argued that the policies adopted made Diu a multiconfessional city, one that accommodated interreligious encounters though it was less supportive of mobility between different religions. This was the outcome of pragmatism, an essential feature of the actions of the different Portuguese agents who intervened at different levels, creating dynamics in which the role of the native population was also decisive in shaping imperial societies.

Stability of Cytoplasmic Membrane of Escherichia coli Bacteria in Aqueous and Ethanolic Environment.

The mechanism of action of any antibacterial agent or disinfectant depends largely on their interaction with the bacterial membrane. Herein, we use the SPICA (surface property fitting coarse graining) force-field and develop a coarse-grained (CG) model for the structure of the cytoplasmic membrane of Escherichia coli (E. coli) and its interaction with water and ethanol. We elucidate the impact of different concentrations of ethanol on the cytoplasmic membrane bilayers and vesicles of E. coli using the CG molecular dynamics (CG MD) simulations. Our modeling approach first focuses on the parametrization of the required force-field for POPG lipid and its interaction with water, ethanol, and POPE lipid. Subsequently, the structural stability of the E. coli bacterial membrane in the presence of high and low concentrations of ethanol is delineated. Both flat bilayers as well as vesicles of E. coli membrane were considered for the CG MD. Our results reveal that, at low ethanol concentrations (<30 mol %), the size of the E. coli vesicles increases with discernible deformations in their shapes. Because of ethanol-induced interdigitation, thinning of the E. coli vesicular membrane is also observed. However, at higher ethanol concentrations (>30 mol %), the integrity of the vesicles is lost because of deteriorating invasion of ethanol molecules into the vesicle bilayer and significant weakening of lipid-lipid interactions. At higher ethanol concentrations (40 and 70 mol %), both the multivesicle and single-vesicle bacterial membranes exhibit a similar rupturing pattern wherein the extraction of lipids from the membrane and formation of aggregates of the component lipids are observed. These aggregates consist of polar head groups of 3-5 POPE/POPG lipids with intertwined nonpolar tails.

Abstract A066: Machine learning elucidates biology of response within and outside the mechanisms of action of therapeutic agents in the I-SPY2 breast cancer TRIAL

Abstract Background: Machine learning (ML) in translational medicine has led to prediction of clinical outcomes and identification of new biomarkers. We employ ML in prediction of pathologic complete response (pCR) in high-risk breast cancer patients in the neoadjuvant I-SPY2 TRIAL where not all novel agents have strong predictive biomarkers. Leveraging a ML approach using progressively expanded candidate genes, we explore the limitations of using only known mechanisms of action in predicting pCR, and the extent to which biology outside known drug action improves response prediction in the first 10 arms of the trial. Methods: ML random forest models were developed in I-SPY2 patients (n=982) with pre-treatment gene expression and pCR data across 10 treatment arms (PMID: 35623341), including inhibitors of HER2: neratinib (N), pertuzumab (P), TDM1/P; AKT (MK-2206); IGF1R (ganitumab); HSP90 (ganetespib); PARP/DNA repair (veliparib/carboplatin, VC); ANG1/2 (trebananib, T); immune checkpoints (PD1-inh); and Control (Ctr). Each HR/HER2 receptor/treatment arm subset (m=27) was evaluated independently. We employed a three-pronged feature-selection approach using (1) genes restricted to known mechanism of action of individual I-SPY2 agents (k=10 to 88 genes); (2) genes expanded to include targeted pathways for all 10 agents/combinations (k=282); and (3) an unbiased whole genome approach (k=17,990). Samples were partitioned with 75% used for training and cross-validation, and 25% held out as test sets. Predictive ML models were defined as those with performance ≥ 0.90 based on different performance metrics (e.g., AUC, sensitivity, specificity). Results: For each of the 27 subtype-treatment subsets, at least one high performing model was identified. In 6 subtype-treatment subsets, mechanism of action genes were sufficient to predict pCR: AKT/PI3K/HER genes in HR+HER2- N and HR-HER2+ P; DNA repair genes in HR+HER2- VC; angiogenesis-associated genes in HR+HER2+ T; and immune-associated genes in both HR+HER2- and HR-HER2- PD1-inh subsets. Expanded targeted pathway models were required to identify predictive models in 8 additional subtype-treatment pairs from the N, T-DM1/P, MK-2206, VC, T, and HER2+ Ctr arms, with significant contribution of DNA repair, immune, and HSP90 genes for multiple arms. A genome-wide approach was required for the remaining 13 subtype-treatment pairs with no previous models from the N, P, MK-2206, ganitumab, ganetespib, T, and HER2- Ctr arms. Even for subtype-treatment pairs where mechanism of action gene sets was sufficient for reasonable models, expanded gene sets resulted in improved performance. For instance, metabolism genes improved model performance for HR-HER2+ in N and Ctr, and for HR+HER2- in the PD1-inh arm; and mitochondrial and protein folding dysfunction genes improved response prediction in HR-HER2- in the ganetespib arm. Conclusion: Our study identifies mechanism of action biomarkers associated with response to each drug and elucidates possible off-target effects contributing to observed drug sensitivity and resistance. Citation Format: Rosalyn W. Sayaman, Denise M. Wolf, Christina Yau, Julia Wulfkhule, Emanuel F. Petricoin, Lamorna Brown-Swigart, Tam Binh Bui, Gillian L. Hirst, Diane Heditsian, W. Fraser Symmans, Angela DeMichele, Mark LaBarge, Laura J. Esserman, Laura van ‘t Veer. Machine learning elucidates biology of response within and outside the mechanisms of action of therapeutic agents in the I-SPY2 breast cancer TRIAL [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Breast Cancer Research; 2023 Oct 19-22; San Diego, California. Philadelphia (PA): AACR; Cancer Res 2024;84(3 Suppl_1):Abstract nr A066.

Study of Q-learning and deep Q-network learning control for a rotary inverted pendulum system

The rotary inverted pendulum system (RIPS) is an underactuated mechanical system with highly nonlinear dynamics and it is difficult to control a RIPS using the classic control models. In the last few years, reinforcement learning (RL) has become a popular nonlinear control method. RL has a powerful potential to control systems with high non-linearity and complex dynamics, such as RIPS. Nevertheless, RL control for RIPS has not been well studied and there is limited research on the development and evaluation of this control method. In this paper, RL control algorithms are developed for the swing-up and stabilization control of a single-link rotary inverted pendulum (SLRIP) and compared with classic control methods such as PID and LQR. A physical model of the SLRIP system is created using the MATLAB/Simscape Toolbox, the model is used as a dynamic simulation in MATLAB/Simulink to train the RL agents. An agent trainer system with Q-learning (QL) and deep Q-network learning (DQNL) is proposed for the data training. Furthermore, agent actions are actuating the horizontal arm of the system and states are the angles and velocities of the pendulum and the horizontal arm. The reward is computed according to the angles of the pendulum and horizontal arm. The reward is zero when the pendulum attends the upright position. The RL algorithms are used without a deep understanding of the classical controllers and are used to implement the agent. Finally, the outcome indicates the effectiveness of the QL and DQNL algorithms compared to the conventional PID and LQR controllers.

Exploring pollution control in transboundary watersheds under the perspective of ecological compensation based on aggregate game.

Pollution control in inter-provincial river basins involves many complex subjects, so it is difficult to effectively implement ecological compensation policies. To clarify the interest relationship among pollution control subjects and stimulate their willingness to cooperate in collaborative governance, this paper builds a multi-agent coordinated pollution control model based on the theory of aggregate game; explores the change of equilibrium action of a single pollution control agent and multi-actors, symmetric, and asymmetric situations under the influence of synergistic benefits; and studies the incentive mechanism design to maintain the cooperation of various agents. The research results show that an increase in the number of upstream firms would lead to a reduction in the incentive effects of downstream government compensation and an increase in the likelihood of "free-riding" and "coordination failure." Synergy benefits vary positively with the degree of cooperation between government and enterprises, and higher synergy benefits can effectively compensate for the high transaction costs caused by multiple entities, alleviate the financial pressure on downstream governments, and increase the willingness of upstream and downstream entities to cooperate. In addition, focusing on wastewater reduction from core enterprises, such as heavy polluters can help improve the efficiency of regional emissions reduction, while having a catalytic effect on small enterprises.

Contemporary Paradigm of Anticoagulant and Antiplatelet Therapy: Revolutionary Approaches to Assessing Blood Coagulation in the Laboratory and Using Instrumental Methods

Statement of the Problem: Impairment of the blood coagulation system remains a relevant issue in clinical practice. A striking example of this is the fact that despite the administration of anticoagulant therapy, up to 35% of COVID-19 patients were hospitalized in the intensive care unit with thromboembolic complications. Prophylactic and therapeutic methods for the hemostatic system, recommended by various protocols, have a general nature. Moreover, they introduce dissonance into the clinical clarity of appropriate anticoagulant and antiplatelet use and fail to provide practitioners with the necessary fundamental knowledge to understand the aspects of clinical tactics for correcting the hemostatic system. Uncertainties remain regarding how to correctly choose a specific anticoagulant and determine its effective dose based on individual clinical and laboratory data, which laboratory markers of the coagulation system should be investigated, and whether they always reflect the true picture of hemostasis. Methodology and Theoretical Orientation: This article briefly presents the main aspects of the functioning of the blood coagulation system, points of application for pharmacological agents, and provides objective information for clinicians about the mechanisms of action of major anticoagulants and antiplatelets. Effective and practical methods for assessing the hemostatic system are recommended. Findings: The proper correction of the hemostatic system can only be carried out by a highly qualified specialist - a clinical transfusiologist, who possesses a comprehensive understanding of the functioning of the coagulation and anticoagulation systems, the platelet and fibrinolysis systems, knows the points of application and mechanisms of action of the pharmacological agents and blood components used, and takes into account internal and external factors that influence hemostasis. Conclusion and Significance: Empirical prescription of anticoagulant and antiplatelet therapies, based solely on instructions and clinical protocols without objective comprehensive analysis of the hemostatic components, is not only ineffective but also life-threatening for the patient. Hemostasiogram (coagulogram) and D-dimer measurements cannot serve as the sole determining markers for monitoring the coagulation system. Evaluation of the blood coagulation system should only be carried out based on a comprehensive analysis of hemostasiogram data, complete blood count, thromboelastography (TEG), and data from all relevant systems.

Exploring the dynamics of innovation: patterns of growth and contraction in the local food industry

PurposeThe purpose of this study is shed light on the underlying forces behind entrepreneurship within a regional innovation system (RIS) in a remote rural region. The authors examine the following questions: Which are the main underlying forces behind the entrepreneurial process in a rural RIS characterized by traditionally low-tech, small-scale businesses? How can the development of a low-tech regional innovation system be conceptualized?Design/methodology/approachThe design of the study is based on entrepreneurship theory. Data analysis followed practices used in phenomenography, a research approach used to analyse and identify commonalities and variations in populations' perceptions of a certain phenomenon. Data are composed using semi-structured interviews and a database composed of company information of all firms in the population.FindingsA proactive mobilization of regional stakeholders and resources can be an important driving force behind the entrepreneurial process and generation of a rural RIS. Innovation can be generated within low-tech industries turning the rural context into an asset. An RIS in a remote rural context can be initiated and orchestrated by regional authorities, but knowledge brokering and orchestration can also be managed by networks of small-scale businesses brought together by mutual benefit and common interests.Research limitations/implicationsRegional innovation systems theory is most often used to study high-tech industries. But by combining regional innovation systems with rural entrepreneurship and entrepreneurship context theory is a fruitful avenue to understand the role of rural entrepreneurship in regional development, even in remote and peripheral regions. Innovation does not need to entail high-tech international environments; it can appear as the result of efforts in low-tech industries in rural and remote environments. The authors’ findings need to be scrutinized; therefore, the authors call for more research on regional innovation systems in rural environments.Practical implicationsIt is possible for regional authorities to orchestrate a development process through the actions of a strong regional agent but also by supporting the creation of networks of small businesses that are built on trust and common interests.Originality/valueThis study contributes to the literature with a new perspective to the study of entrepreneurship and of regional innovation systems. Entrepreneurship research with focus on rural contexts most often highlight limits to entrepreneurship and see entrepreneurship as “just running a business”. A perspective that starts from innovation and innovative behaviour, despite the rural context and embedded resources, helps to generate new knowledge that can enrich the understanding of entrepreneurship and also be the foundation for more precise business development policies in rural settings.

Microbial eczema: modern possibilities of consistent topical therapy

At the present stage, there is an increase in the prevalence of allergic spectrum diseases. Epidemiological studies and the active introduction of genomic technologies in recent years have made it possible to accumulate data indicating the significant role of bacteria and fungi in the development of sensitization, which is one of the triggering and supporting factors for the course of microbial eczema. Dermatosis has a chronic relapsing course, in the formation of which delayed-type hypersensitivity, free-radical processes, and violations of the epidermal barrier play a significant role. A number of studies have noted the relationship between the development of microbial eczema and genetic factors, neuroendocrine and vegetative-vascular disorders. Features of etiopathogenesis and the recurrent nature of the course of microbial eczema necessitate an integrated approach to the management of patients with this dermatosis. Treatment of the disease includes elimination of the action of bacterial and fungal agents, anti-inflammatory therapy and restoration of the epidermal barrier. To suppress the infection, the appointment of antibacterial and antifungal drugs is indicated. However, their use as monotherapy in patients with eczema can provoke an exacerbation of the allergic process. In this regard, the article describes the current possibilities of using topical glucocorticosteroid drugs in combination with antibiotics and antimycotics. Attention is focused on ways to prevent complications and a staged approach to the management of patients with microbial eczema, the important components of which, after stopping the infectious process, are the suppression of inflammation and the restoration of the epidermal barrier. We present our own observations of patients with microbial eczema treated with preparations containing combinations of glucocorticosteroid with antibiotics and antimycotics, glucocorticosteroid with ceramides.

Unraveling the transcriptional features and gene expression networks of pathogenic and saprotrophic Ophiostoma species during the infection of Ulmus americana.

American elm (Ulmus americana), highly prized for its ornamental value, has suffered two successive outbreaks of Dutch elm disease (DED) caused by ascomycete fungi belonging to the genus Ophiostoma. To identify the genes linked to the pathogenicity of different species and lineages of Ophiostoma, we inoculated 2-year-old U. americana saplings with six strains representing three species of DED fungi, and one strain of the saprotroph Ophiostoma quercus. Differential expression analyses were performed following RNA sequencing of fungal transcripts recovered at 3- and 10-days post-infection. Based on a total of 8,640 Ophiostoma genes, we observed a difference in fungal gene expression depending on the strain inoculated and the time of incubation in host tissue. Some genes overexpressed in the more virulent strains of Ophiostoma encode hydrolases that possibly act synergistically. A mutant of Ophiostoma novo-ulmi in which the gene encoding the ogf1 transcription factor had been deleted did not produce transcripts for the gene encoding the hydrophobin cerato-ulmin and was less virulent. Weighted gene correlation network analyses identified several candidate pathogenicity genes distributed among 13 modules of interconnected genes.IMPORTANCEOphiostoma is a genus of cosmopolitan fungi that belongs to the family Ophiostomataceae and includes the pathogens responsible for two devastating pandemics of Dutch elm disease (DED). As the mechanisms of action of DED agents remain unclear, we carried out the first comparative transcriptomic study including representative strains of the three Ophiostoma species causing DED, along with the phylogenetically close saprotrophic species Ophiostoma quercus. Statistical analyses of the fungal transcriptomes recovered at 3 and 10 days following infection of Ulmus americana saplings highlighted several candidate genes associated with virulence and host-pathogen interactions wherein each strain showed a distinct transcriptome. The results of this research underscore the importance of investigating the transcriptional behavior of different fungal taxa to understand their pathogenicity and virulence in relation to the timeline of infection.

Novel Activity of Oral Hypoglycemic Agents Linked with Decreased Formation of Tryptophan Metabolite, Kynurenic Acid.

Kynurenic acid is a tryptophan (Trp) metabolite formed along the kynurenine (KYN) pathway in the brain and in peripheral tissues. The disturbed formation of kynurenic acid, which targets glutamate-mediated neurotransmission, GPR35, and aryl hydrocarbon receptors of immune or redox status, was implicated in the development of neuropsychiatric and metabolic disorders among others. Kynurenic acid exerts neuroprotective and immunomodulatory effects, yet its high brain levels may negatively impact cognition. Changes in the Trp-KYN pathway are also linked with the pathogenesis of diabetes mellitus, which is an established risk factor for cardiovascular and neurological diseases or cognitive deficits. Here, the effects of metformin and glibenclamide on the brain synthesis of kynurenic acid were evaluated. Acute exposure of rat cortical slices in vitro to either of the drugs reduced kynurenic acid production de novo. Glibenclamide, but not metformin, inhibited the activity of kynurenic acid biosynthetic enzymes, kynurenine aminotransferases (KATs) I and II, in semi-purified cortical homogenates. The reduced availability of kynurenic acid may be regarded as an unwanted effect, possibly alleviating the neuroprotective action of oral hypoglycemic agents. On the other hand, considering that both compounds ameliorate the cognitive deficits in animal and human studies and that high brain kynurenic acid may hamper learning and memory, its diminished synthesis may improve cognition.

Crowd evacuation with human-level intelligence via neuro-symbolic approach

Understanding human response to crowd emergencies is extremely complex, and it plays a significant role in engineering construction designs and crowd safety. Individual choices, reasoning, and behaviours cannot be fully described by equations or rule-based methods. Accordingly, this research proposes a neuro-symbolic approach for modelling agents with human-level capabilities of reasoning and performance in an emergency evacuation. The proposed neuro-symbolic approach combines deep reinforcement learning with evaluative fuzzy logic to address the challenges of large amounts of required data, time, and trials-and-errors for policy optimization and to handle the assumption of reward function that may not be practical in real scenarios. This neuro-symbolic model has the potential to deal with the complexity of the environment and decision-making process via deep reinforcement learning and enhances the cognitive and visual intelligence via an evaluative fuzzy function, which continuously evaluates agent actions during the training process to boost pedestrian active response to their surroundings, with full awareness of time, thereby, the human-level capacity of reasoning. Moreover, this proposed model optimizes the computational demands of deep reinforcement learning and enables faster learning of new situations. The findings indicate that the proposed model can produce behavioural patterns that align with real observations of crowd evacuation, such as laminar flow, stop-and-go flow, and crowd turbulence. On top of that, a new evacuation behaviour is observed, as some pedestrians avoid congestion at the exit until the density reduces which reflects a level of human reasoning. The proposed model illustrates a higher accuracy and much faster converge than the pure proximal policy optimization model with substantially minimal training timesteps of as little percentage as 2 to 8. Meanwhile, the reliability study records an increase of the mean and standard deviation of evacuation time from 39.7 s, 1.06 to 155.09 s, 7.39 as crowd size increases from 15 to 200 pedestrians, which implies a rise of uncertainty. Therefore, we perceive that this work can provide crowd authorities and construction engineers with insights into complex behaviour and critical conditions to make better evacuation plans and sustainable designs to ensure crowd safety. It also provides a promising alternative to the evident lack of data on critical crowd conditions.

Abstract B020: Targeting the DNA damage response sensor Replication Protein A for first in class cancer therapy

Abstract The DNA Damage Response (DDR) is a DNA repair and cell signaling pathway crucial in maintaining genomic stability. The most successful DDR targeted therapies inhibit the DNA damage sensor, PARP, a protein that recognizes single strand DNA (ssDNA) breaks as well as other DNA secondary structures induced by single strand gaps that can initiate chromosome instability and cell death. Building on the initial clinical success of PARP inhibitors, development of DDR targeted therapeutics has become increasingly popular with the majority of new agents targeting protein kinases downstream of the DNA damage sensors. The clinical outcomes with these therapeutics have not met expectations and thus a new approach to targeting the DDR pathway is needed. NERx Biosciences has developed a novel strategy of targeting the DDR by intervening upstream of the DDR kinases and targeting specific DDR sensors. The human ssDNA binding protein, replication protein A (RPA), is a critical sensor for the DDR, serving to sense ssDNA at replication forks that arise from replication stress (RS) and is a novel target for cancer therapy. We have discovered, developed, and characterized a novel small molecule RPA inhibitor (RPAi) NERx-329 that blocks the RPA-DNA interaction and elicits a state of chemical RPA exhaustion that results in in vivo anticancer activity. Chemical optimization enabled in vitro and in vivo analyses to elucidate mechanism of action, cellular engagement, and therapeutic activity of RPA-targeted agents. Biochemical reconstitution of the ATR signaling pathway shows that NERx-329 disrupts ATR kinase activity, further suggesting a novel mechanism of action driven by RPA inhibition induced replication catastrophe. NERx-329 elicits single agent in vitro anticancer activity across a broad spectrum of cancers and the cellular response suggests existence of a threshold before chemical RPA exhaustion induces cell death. In vivo analysis reveals activity in a series of lung and ovarian PDX models of human cancer. Single agent activity results in a tumor growth delay, which can be enhanced in combination therapy with DDR inhibitors as well as traditional DNA damaging cancer therapeutics. Data demonstrate that tumor growth observed in vivo is a function of NERx329 bioavailability and not development of a specific resistance mechanism. In addition, specific genetic predictors of RPAi sensitivity have been identified and results demonstrate specific genetic alterations increase RPAi activity in vitro and in vivo. NERx-329 represents the first clinically viable agent to target the DDR pathway by disrupting the RPA-DNA interaction and holds potential for significant impact in cancer treatment. This is supported by data from a retrospective clinical assessment of RPA expression to assess treatment outcomes in lung cancer patients, revealing that high RPA expression is a negative prognostic biomarker correlating with worse overall patient survival. NERx 329 offers a novel therapeutic approach for these cancers that are not effectively treated with existing therapeutics. Citation Format: Katherine Pawelczak, Pamela VanderVere-Carozza, Matthew Jordon, Navnath Gavande, John Turchi. Targeting the DNA damage response sensor Replication Protein A for first in class cancer therapy [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: DNA Damage Repair: From Basic Science to Future Clinical Application; 2024 Jan 9-11; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2024;84(1 Suppl):Abstract nr B020.

Codelivery of TGFβ and Cox2 siRNA inhibits HCC by promoting T-cell penetration into the tumor and improves response to Immune Checkpoint Inhibitors.

Upregulation of TGFβ and Cox2 in the tumor microenvironment results in blockade of T-cell penetration into the tumor. Without access to tumor antigens, the T-cell response will not benefit from administration of the immune checkpoint antibodies. We created an intravenous polypeptide nanoparticle that can deliver two siRNAs (silencing TGFβ and Cox2). Systemic administration in mice, bearing a syngeneic orthotopic hepatocellular carcinoma (HCC), delivers the siRNAs to various cells in the liver, and significantly reduces the tumor. At 2mg/kg (BIW) the nanoparticle demonstrated a single agent action and induced tumor growth inhibition to undetectable levels after five doses. Reducing the siRNAs to 1mg/kg BIW demonstrated greater inhibition in the presence of PD-L1 mAbs. After only three doses BIW, we could still recover a smaller tumor and, in tumor sections, showed an increase in penetration of CD4+and CD8+T-cells deeper into the remaining tumor that was not evident in animals treated with non-silencing siRNA. The combination of TGFβ and Cox2 siRNA co-administered in a polypeptide nanoparticle can act as a novel therapeutic alone against HCC and may augment the activity of the immune checkpoint antibodies. Silencing TGFβ and Cox2 converts an immune excluded (cold) tumor into a T-cell inflamed (hot) tumor.

АГЕНТНЕ МОДЕЛЮВАННЯ ПРИЙНЯТТЯ РІШЕНЬ НА ФІНАНСОВИХ РИНКАХ

The work deals with the topical issue of researching the behavior of financial market agents and identifying the motives of participants in unstable, extremely risky and practically unmanaged markets in the modern conditions of their functioning. This article presents the principles of creating financial models, defines the principles of decision-making in financial markets, and analyzes the impact of herd behavior on decision-making. In the work, the concept of agent modeling of decision-making in financial markets is built, mathematical and statistical models, multi-agent systems, models of decisionmaking in financial markets are considered. The available tools for the development and implementation of simulation agent models were analyzed and a simulation agent model of decision-making in the financial market was developed in the AnyLogic Professional modeling environment, with the aim of analyzing the influence of agent behavior on the dynamics of prices in the stock market. Based on a reflexive approach to the analysis of the stock market, within the framework of this work, situations of decision-making by agents regarding the purchase or sale of securities on the virtual stock market were simulated using the Ising model – modeling the impact of different levels of the herding coefficient on the stock market. In the process of modeling, the set goals and objectives were achieved, namely: determination of the principles of functioning and structure of the stock market; the influence of information value on the efficiency of the stock market was investigated; the process of emergence of bubbles on the stock market is investigated. As a result of the computer experiments carried out as part of this work, two virtual stock markets were investigated – a virtual stock market with the number of 1000 agents and the presence of intuitive agents, and a dynamic virtual market with agents under the influence of authoritative agents. So, the scientific article considers methodological approaches to the analysis of decision-making by agents in financial markets and modeling of decision-making processes by economic agents based on the agent model of group behavior, which allows to identify the presence of herd behavior in the actions of agents and to form management measures to coordinate the decision-making process.

Decision-Making With Speculative Opponent Models.

Opponent modeling has proven effective in enhancing the decision-making of the controlled agent by constructing models of opponent agents. However, existing methods often rely on access to the observations and actions of opponents, a requirement that is infeasible when such information is either unobservable or challenging to obtain. To address this issue, we introduce distributional opponent-aided multiagent actor-critic (DOMAC), the first speculative opponent modeling algorithm that relies solely on local information (i.e., the controlled agent's observations, actions, and rewards). Specifically, the actor maintains a speculated belief about the opponents using the tailored speculative opponent models that predict the opponents' actions using only local information. Moreover, DOMAC features distributional critic models that estimate the return distribution of the actor's policy, yielding a more fine-grained assessment of the actor's quality. This thus more effectively guides the training of the speculative opponent models that the actor depends upon. Furthermore, we formally derive a policy gradient theorem with the proposed opponent models. Extensive experiments under eight different challenging multiagent benchmark tasks within the MPE, Pommerman, and starcraft multiagent challenge (SMAC) demonstrate that our DOMAC successfully models opponents' behaviors and delivers superior performance against state-of-the-art (SOTA) methods with a faster convergence speed.

Learning-Based Reconfiguration of Charged Spacecraft Formation in Geomagnetic Field.

This article introduces a novel approach for spacecraft formation flying utilizing Lorentz-augmented techniques. It demonstrates that the relative motion among spacecraft, driven by the Lorentz force, possesses equilibrium states beneficial for formation maintenance. However, for effective formation reconfiguration, reliance solely on the Lorentz force is insufficient; low thrust is also necessary. To address this, this article proposes an optimal control framework based on reinforcement learning (RL). It derives the nonlinear dynamics of relative motion within the geomagnetic field, considering intersatellite Lorentz force, atmospheric drag, and Earth's gravitational harmonics. The study employs Lagrangian coherent structure analysis to identify relative equilibrium configurations and develops an RL-based optimal control strategy for real-time formation reconfiguration. By leveraging optimal demonstrations, the framework guides the agent's actions to match these demonstrations over time, especially when encountering out-of-distribution states. Numerical simulations confirm the method's optimality, robustness, and real-time performance, highlighting its potential in achieving optimal control and adapting to varying environment in future space missions.

Masked and Inverse Dynamics Modeling for Data-Efficient Reinforcement Learning.

In pixel-based deep reinforcement learning (DRL), learning representations of states that change because of an agent's action or interaction with the environment poses a critical challenge in improving data efficiency. Recent data-efficient DRL studies have integrated DRL with self-supervised learning (SSL) and data augmentation to learn state representations from given interactions. However, some methods have difficulties in explicitly capturing evolving state representations or in selecting data augmentations for appropriate reward signals. Our goal is to explicitly learn the inherent dynamics that change with an agent's intervention and interaction with the environment. We propose masked and inverse dynamics modeling (MIND), which uses masking augmentation and fewer hyperparameters to learn agent-controllable representations in changing states. Our method is comprised of a self-supervised multitask learning that leverages a transformer architecture, which captures the spatiotemporal information underlying in the highly correlated consecutive frames. MIND uses two tasks to perform self-supervised multitask learning: masked modeling and inverse dynamics modeling. Masked modeling learns the static visual representation required for control in the state, and inverse dynamics modeling learns the rapidly evolving state representation with agent intervention. By integrating inverse dynamics modeling as a complementary component to masked modeling, our method effectively learns evolving state representations. We evaluate our method by using discrete and continuous control environments with limited interactions. MIND outperforms previous methods across benchmarks and significantly improves data efficiency. The code is available at https://github.com/dudwojae/MIND.