Toxicity risk assessment plays a crucial role in determining the clinical success and market potential of drug candidates. Traditional animal-based testing is costly, time-consuming, and ethically controversial, which has led to the rapid development of computational toxicology. This review surveys over 20 ADMET prediction platforms, categorizing them into rule/statistical-based methods, machine learning (ML) methods, and graph-based methods. We also summarize major toxicological databases into four types: chemical toxicity, environmental toxicology, alternative toxicology, and biological toxin databases, highlighting their roles in model training and validation. Furthermore, we review recent advancements in ML and artificial intelligence (AI) applied to toxicity prediction, covering acute toxicity, organ-specific toxicities, and carcinogenicity. The field is transitioning from single-endpoint predictions to multi-endpoint joint modeling, incorporating multimodal features. We also explore the application of generative modeling techniques and interpretability frameworks to improve the accuracy and credibility of predictions. Additionally, we discuss the use of network toxicology in evaluating the safety of traditional Chinese medicines (TCMs) and the potential of large language models (LLMs) in literature mining, knowledge integration, and molecular toxicity prediction. Finally, we address current challenges, including data quality, model interpretability, and causal inference, and propose future directions such as multi-omics integration, interpretable AI models, and domain-specific LLMs, aiming to provide more efficient and precise technical support for preclinical toxicity assessments in drug development.

In this paper, the problem of emergency power supply for electric vehicles in road traffic networks after earthquake disasters is studied. In emergency response, it is important for evs to reach the target shelter as quickly as possible while considering the path dynamic changes caused by secondary disasters. Therefore, in order to solve this problem, this paper classifies it as the Dynamic Electric Vehicle Routing Problem with Time Windows (DEVRPTW), formulates it as a mixed integer linear programming model, and uses the Deep Reinforcement Learning (DRL) algorithm to solve it. The DRL algorithm adopted a dual-end merging Attention (DEMA) mechanism based on Multi-head Attention (MHA) modification. The DEMA model is subsequently trained using a reinforcement algorithm with rollout baselines. Finally, the MHA model, DEMA model and different training algorithms were compared and analyzed by using randomly generated geographic information data. The simulation results show that DEMA performs well when combined with an enhancement algorithm with rollout baseline.

This paper investigates an energy management problem of parallel hybrid electric vehicles (HEVs), which can be modeled as a finite horizon optimal control problem for the discrete dynamical systems. Taking the essential characteristics of plug-in HEVs into account, a logical-based optimization approach is applied to realize the equivalent energy cost minimization of the powertrain system. Then, based on semi-tensor product, an effective algorithm for obtaining an approximate optimal solution is proposed by using the logical network-based approach. Finally, simulation results are presented to illustrate and show the effectiveness of the proposed optimal control scheme and the corresponding algorithm.