Real-world Datasets Research Articles

Variational Bayes latent class analysis for EHR-based phenotyping with large real-world data

IntroductionBayesian approaches to patient phenotyping in clinical observational studies have been limited by the computational challenges associated with applying the Markov Chain Monte Carlo (MCMC) approach to real-world data. Approximate Bayesian inference via optimization of the variational evidence lower bound, variational Bayes (VB), has been successfully demonstrated for other applications.MethodsWe investigate the performance and characteristics of currently available VB and MCMC software to explore the practicability of available approaches and provide guidance for clinical practitioners. Two case studies are used to fully explore the methods covering a variety of real-world data. First, we use the publicly available Pima Indian diabetes data to comprehensively compare VB implementations of logistic regression. Second, a large real-world data set, Optum™ EHR with approximately one million diabetes patients extended the analysis to large, highly unbalanced data containing discrete and continuous variables. A Bayesian patient phenotyping composite model incorporating latent class analysis (LCA) and regression was implemented with the second case study.ResultsWe find that several data characteristics common in clinical data, such as sparsity, significantly affect the posterior accuracy of automatic VB methods compared with conditionally conjugate mean-field methods. We find that for both models, automatic VB approaches require more effort and technical knowledge to set up for accurate posterior estimation and are very sensitive to stopping time compared with closed-form VB methods.DiscussionOur results indicate that the patient phenotyping composite Bayes model is more easily usable for real-world studies if Monte Carlo is replaced with VB. It can potentially become a uniquely useful tool for decision support, especially for rare diseases where gold-standard biomarker data are sparse but prior knowledge can be used to assist model diagnosis and may suggest when biomarker tests are warranted.

Flow prediction via adaptive dynamic graph with spatio-temporal correlations

Flow prediction has shown its significance in optimizing and smoothing resource allocation. However, it still faces challenges due to inherent temporal variability and dynamic spatial correlations of flow data. Although many methods have been introduced to solve the problem, they often rely on predefined graphs to capture shared patterns, which is likely to overlook the critical spatiotemporal dependencies that are necessary for mid-term and long-term prediction tasks. To address the issue, a novel flow prediction model, termed Dynamic Graph Multi-Step Prediction via Long Short-Term Memory Network and Adaptive Learning (DMLSA), is proposed. DMLSA employs Node Adaptive Learning (NAL) to learn node-specific patterns without relying on predefined static adjacency matrices. Moreover, DMLSA utilizes Graph Adaptive Generation (GAG) to extract dynamic features from node attributes and automatically infer interdependencies among flow sequences, thus enhancing the ability to predict complex flow dynamics. Additionally, DMLSA integrates multi-head self-attention mechanisms with gated recurrent units, effectively capturing both long-term and short-term node dependencies, therefore allowing for more precise identification of spatiotemporal patterns at the individual node level. Extensive experiments on five datasets demonstrate that DMLSA outperformed existing baseline models in terms of accuracy and stability for multi-step prediction tasks.

Integrating machine learning models to learn potentially non-monotonic preferences for multi-criteria sorting from large-scale assignment examples

Learning preferences from assignment examples has attracted considerable attention in the field of multi-criteria sorting (MCS). However, traditional MCS methods, designed to infer decision makers’ preferences from small-scale assignment examples, encounter limitations when confronted with large-scale data sets. Additionally, the presence of decision makers’ non-monotonic preferences for certain criteria in MCS problems necessitates accounting for potential non-monotonicity when devising preference learning methods. To address this, this paper proposes some new models to learn potentially non-monotonic preferences for MCS problems from large-scale assignment examples by leveraging machine learning models. Specifically, we first introduce the Piecewise-Linear Neural Network (PLNN) model, which leverages the threshold-based value-driven sorting procedure as the underlying sorting model and integrates a perceptron-based model to establish piecewise-linear marginal value functions to approximate real ones. On this basis, we address MCS problems with criteria interactions and extend the PLNN model to develop the Piecewise-Linear Factorization Machine-based Neural Network (PLFMNN) model by incorporating the factorization machine to factorize interaction coefficients. Training these models allows us to learn potentially non-monotonic preferences of decision makers. To illustrate the proposed models, we apply them to a red wine quality classification problem. Furthermore, we assess the performance of the proposed models through computational experiments on both artificial and real-world data sets. Additionally, we conduct statistical tests to ascertain the significance of the performance differences. Experimental results reveal that the proposed models are comparable to the multilayer perceptron model and outperform other baseline models on most data sets, thus affirming their efficacy. Finally, we conduct some sensitivity analysis to assess the impact of certain parameters on the performance of the proposed models and compare them with existing studies from a theoretical perspective, further demonstrating their effectiveness.

Multi-Way adaptive Time Aware LSTM for irregularly collected sequential ICU data

In personalized predictive medicine, accurately modeling a patient’s illness and care processes is essential, given their inherent long-term temporal dependencies. However, electronic medical records contain episodic and irregularly timed data due to patients visiting hospitals based on treatment needs, resulting in unique patterns for each hospital stay. Consequently, when constructing a personalized predictive model, it is crucial to consider these factors in order to accurately capture the patient’s health journey.To address this challenge, we present a novel deep dynamic memory neural network called Multi-Way adaptive Time Aware LSTM (MWTA-LSTM). The primary objective of MWTA-LSTM is to leverage medical records, memorize illness trajectories and care processes, estimate current illness states, and predict future risks, thereby providing a high level of precision and predictive power. To enhance its capabilities, MWTA-LSTM extends the conventional Long Short-Term Memory (LSTM) model in two key ways. Firstly, it incorporates frequency measurement and the most recent observation(last observation) to enhance personalized predictive modeling of patient illnesses, enabling a more accurate understanding of the patient’s condition. Secondly, it parameterizes the cell state to handle irregular timing effectively, utilizing both frequency measurement and elapsed times. Furthermore, the model capitalizes on both to comprehend the impact of interventions on the course of illness on the cell state, facilitating the memorization of illness courses and improving its ability to capture the temporal dynamics of healthcare data, accommodating variations and irregularities in event and observation timing. Lastly, we introduce a novel adaptive pooling strategy that specifically targets and resolves outlier issues that can potentially occur during the analysis of EHR data. By incorporating these features, MWTA-LSTM significantly improves its ability to capture the temporal dynamics of healthcare data, accommodating variations and irregularities in event and observation timing.We validate the effectiveness of our proposed model through empirical experiments conducted on two real-world clinical datasets and three real-world time series datasets. Our results demonstrate the superiority of MWTA-LSTM over current state-of-the-art models and other robust baselines. This showcases the potential of MWTA-LSTM in advancing personalized predictive medicine by offering a more accurate and comprehensive approach to modeling patient health trajectories.

Multidimensional dynamic attention for multivariate time series forecasting

Attention-based models have been very effective in identifying important lagged variables for multivariate time series (MTS) forecasting applications. However, current attention-based models only provide static weights and do not consider the dynamic nature of predictions for multistep predictions of heterogeneous MTS. To address these limitations, this paper proposes a novel multidimensional dynamic attention (MDA) model for computing lagged variable importance. It incorporates a dynamic representation learner unit and considers multiple attention calculations to account for prediction dynamics, temporal information, and variable relations. Extensive experiments with both synthetic and real-world data demonstrate the effectiveness of the MDA model. It outperforms existing methods in the literature for sequence-to-sequence prediction of heterogeneous MTS in most cases and accurately identifies important features. MDA demonstrates enhancements up to 33% with real-world datasets. These findings demonstrate that the MDA model is a promising approach to MTS forecasting. The proposed attention mechanism can be utilised for other tasks related to MTS analysis beyond just forecasting, potentially enhancing the performance and interpretability of various MTS applications. The code is publicly available on GitHub, promoting widespread adoption and further research in this field [ Note: The GitHub link will be shared based on the paper’s acceptance].

ACGAN: adaptive conditional generative adversarial network architecture predicting skin lesion using collaboration of transfer learning models

Skin cancer has become a serious disease which has the potential to scale up if it is not identified earlier. It is imperative to detect and give treatment to skin cancer promptly. Diagnosing skin cancer manually takes a lot of time and it is costly, and the probability of false diagnosis has increased due to the outstanding resemblances among various skin lesions. Enhancing the classification of multi-class lesions of skin needs the development of investigative systems which should be automated. Data augmentation with GANs and Adaptive Conditional Generative Adversarial Network strategies improves performance. The performance is tested using balanced and unbalanced datasets. Using a proper process of augmentation of data, the suggested system attains a 94% accuracy for the VGG16, 93% for the ResNet50 and 94.25% for ResNet101. The process of collaboration of all such methods improves accuracy further to 95%. In summary, the novelty of the work lies in its holistic approach to automated skin lesion classification, incorporating advanced deep learning models, novel data augmentation techniques and comprehensive performance evaluation on real-world datasets. These contributions collectively advance the field of computer-aided diagnosis for the detection of skin cancer and treatment.

CLlight: Enhancing representation of multi-agent reinforcement learning with contrastive learning for cooperative traffic signal control

Multi-agent reinforcement learning has shown great potential for coordinating multi-intersection traffic signals due to its powerful adaptive capabilities, treating each intersection as an agent. However, in the real world, different intersections possess differentiating characteristics such as unique vehicle distributions and traffic patterns. Most existing methods directly add neighboring intersection states to local intersections and optimize the cooperative policy network based on synthesized global features. This indirect optimization approach makes it difficult to thoroughly explore the mutual interactions among different intersection agents, preventing agents from truly learning features with cooperative awareness. To resolve these challenges, we introduce contrastive learning as representation task to the multi-intersection traffic signal control approach named CLlight for two-stage policy network updating. In the first stage, we utilize policy-based or actor-critic-based reinforcement learning methods such as A2C, SAC, and PPO to train policy networks with certain representational capabilities. In the second stage, by extracting pre- and post-masked features and reconstructing the post-masked features, the agents are encouraged to learn the similarities and differences between different intersection policies, which in turn enhances the cooperative and individual representation capabilities of the policy network. To the best of our knowledge, this is the first application of contrastive learning in the field of traffic signal control. Experimental results demonstrate, compared to other state-of-the-art traffic signal control methods, superior average travel time and average waiting time performance under various scenarios, tested on synthetic and real-world datasets.

FedADMM-InSa: An inexact and self-adaptive ADMM for federated learning

Federated learning (FL) is a promising framework for learning from distributed data while maintaining privacy. The development of efficient FL algorithms encounters various challenges, including heterogeneous data and systems, limited communication capacities, and constrained local computational resources. Recently developed FedADMM methods show great resilience to both data and system heterogeneity. However, they still suffer from performance deterioration if the hyperparameters are not carefully tuned. To address this issue, we propose an inexact and self-adaptive FedADMM algorithm, termed FedADMM-InSa. First, we design an inexactness criterion for the clients’ local updates to eliminate the need for empirically setting the local training accuracy. This inexactness criterion can be assessed by each client independently based on its unique condition, thereby reducing the local computational cost and mitigating the undesirable straggle effect. The convergence of the resulting inexact ADMM is proved under the assumption of strongly convex loss functions. Additionally, we present a self-adaptive scheme that dynamically adjusts each client’s penalty parameter, enhancing algorithm robustness by mitigating the need for empirical penalty parameter choices for each client. Extensive numerical experiments on both synthetic and real-world datasets have been conducted. As validated by some tests, our FedADMM-InSa algorithm improves model accuracy by 7.8% while reducing clients’ local workloads by 55.7% compared to benchmark algorithms.

TeST: Temporal-spatial separated transformer for temporal action localization

Temporal action localization is a fundamental task in video understanding. Existing methods fall into three categories: anchor-based, actionness-guided, and anchor-free. Anchor-based and actionness-guided models need huge computation resources to process redundant proposals or enumerate every possible proposal. Anchor-free models with lighter parameters become a more attractive option as temporal actions become more complex. However, they typically struggle to achieve high performance due to the need to aggregate global temporal-spatial features at every time step. To overcome this limitation, we design three efficient transformer-based architectures, bringing two advantages: (i) the global receptive field of transformers enables models to aggregate spatial and temporal at each time step, and (ii) the transformers could capture the moment-level feature, enhancing localization performance. Our designed architectures are adapted to any framework, thus we propose a simple but effective anchor-free framework named TeST. Compared to strong baselines, TeST achieves 0.96% to 3.20% improvement on two real-world datasets. Meanwhile, it improves time efficiency by 1.36 times and space efficiency by 1.08 times. Further experiments prove the effectiveness of TeST’s modules. Implementation of our work is available at https://github.com/whr000001/TeST.

Integrating granular computing with density estimation for anomaly detection in high-dimensional heterogeneous data

Detecting anomalies in complex data is crucial for knowledge discovery and data mining across a wide range of applications. While density-based methods are effective for handling varying data densities and diverse distributions, they often struggle with accurately estimating densities in heterogeneous, uncertain data and capturing interdependencies among features in high-dimensional spaces. This paper proposes a fuzzy granule density-based anomaly detection algorithm (GDAD) for heterogeneous data. Specifically, GDAD first partitions high-dimensional attributes into subspaces based on their interdependencies and employs fuzzy information granules to represent data. The core of the method is the definition of fuzzy granule density, which leverages local neighborhood information alongside global density patterns and effectively characterizes anomalies in data. Each object is then assigned a fuzzy granule density-based anomaly factor, reflecting its likelihood of being anomalous. Through extensive experimentation on various real-world datasets, GDAD has demonstrated superior performance, matching or surpassing existing state-of-the-art methods. GDAD's integration of granular computing with density estimation provides a practical framework for anomaly detection in high-dimensional heterogeneous data.

A Machine learning-based framework and open-source software for Non Intrusive Water Monitoring

Recent research highlights the potential of consumption-based feedback for water conservation, emphasizing the need for Non Intrusive Water Monitoring (NIWM). However, existing studies often rely on small datasets, a pre-selected class of models, and inaccessible software. Here, we introduce PyNIWM, a machine learning-based open-source Python framework for NIWM. PyNIWM enables water end-use classification via (i) data characterization and feature engineering, (ii) water end-use event classification with four machine learning classifiers, and (iii) performance assessment. We demonstrate PyNIWM on a real-world dataset containing around 800,000 labelled end-use events from 762 homes across the USA and Canada. The four PyNIWM classifiers achieve F1 scores above 0.85, indicating high suitability for water end-use classification. However, a tradeoff between accuracy and computational cost exists. Finally, data balancing through oversampling enhances classification of low-represented end-use classes, but does not improve overall classification. We release PyNIWM as an open-source software, aiming for collaborative and reproducible research.

Contrastive Clustering Learning for Multi-Behavior Recommendation

Increasing multiple behavior recommendation models have achieved great successes. However, many models do not consider commonalities and differences between behaviors and data sparsity of the target behavior. This paper proposes a novel Multi-behavior Recommendation Model (MBRCC) based on Contrastive Clustering Learning. Specifically, the graph convolutional network (GCN) is employed to obtain the embeddings of users and items, respectively. Then, three kinds of tasks (including behavior-level embedding, instance-level embedding and cluster-level embedding) are designed to optimize the embeddings of users and items. In behavior-level embedding, we design an adaptive parameter learning strategy to analyze the impact of auxiliary behaviors on the target behavior. Then, the embeddings of users for each behavior are weighted to obtain the final embeddings of users. In instance-level embedding, we employ contrastive learning to analyze the instances of user and item for mitigating the issue of data sparsity. In cluster-level embedding, we design a new cluster contrastive learning method to capture the similarity between groups of user and item. Finally, we combine these three tasks to improve the quality of the embeddings of users and items. We conduct extensive experiments on three real-world datasets and experimental results indicate that the MBRCC remarkably outperforms numerous existing recommendation models.

Double-Structured Sparsity Guided Flexible Embedding Learning for Unsupervised Feature Selection.

In this article, we propose a novel unsupervised feature selection model combined with clustering, named double-structured sparsity guided flexible embedding learning (DSFEL) for unsupervised feature selection. DSFEL includes a module for learning a block-diagonal structural sparse graph that represents the clustering structure and another module for learning a completely row-sparse projection matrix using the l2,0 -norm constraint to select distinctive features. Compared with the commonly used l2,1 -norm regularization term, the l2,0 -norm constraint can avoid the drawbacks of sparsity limitation and parameter tuning. The optimization of the l2,0 -norm constraint problem, which is a nonconvex and nonsmooth problem, is a formidable challenge, and previous optimization algorithms have only been able to provide approximate solutions. In order to address this issue, this article proposes an efficient optimization strategy that yields a closed-form solution. Eventually, through comprehensive experimentation on nine real-world datasets, it is demonstrated that the proposed method outperforms existing state-of-the-art unsupervised feature selection methods.

MinoritySalMix and adaptive semantic weight compensation for long-tailed classification

In real-world datasets, the widespread presence of a long-tailed distribution often leads models to become overly biased towards majority class samples while ignoring minority class samples. We propose a strategy called MASW (MinoritySalMix and adaptive semantic weight) to improve this problem. First, we propose a data augmentation method called MinoritySalMix (minority-saliency-mixing), which uses significance detection techniques to select significant regions from minority class samples as cropping regions and paste them into the same regions of majority class samples to generate brand new samples, thereby amplifying images containing important regions of minority class samples. Second, in order to make the label value information of the newly generated samples more consistent with the image content of the newly generated samples, we propose an adaptive semantic compensation factor. This factor provides more label value compensation for minority samples based on the different cropping areas, thereby making the new label values closer to the content of the newly generated samples. Improve model performance by generating more accurate new label value information. Finally, considering that some current re-sampling strategies generally lack flexibility in handling class sampling weight allocation and frequently require manual adjustment. We designed an adaptive weight function and incorporated it into the re-sampling strategy to achieve better sampling. The experimental results on three long-tailed datasets show that our method can effectively improve the performance of the model and is superior to most advanced long-tailed methods. Furthermore, we extended MinoritySalMix’s strategy to three balanced datasets for experimentation, and the results indicated that our method surpassed several advanced data augmentation techniques.

GEMimp: An accurate and robust imputation method for microbiome data using graph embedding neural network

Microbiome research has increasingly underscored the profound link between microbial compositions and human health, with numerous studies establishing a strong correlation between microbiome characteristics and various diseases. However, the analysis of microbiome data is frequently compromised by inherent sparsity issues, characterized by a substantial presence of observed zeros. These zeros not only skew the abundance distribution of microbial species but also undermine the reliability of scientific conclusions drawn from such data. Addressing this challenge, we introduce GEMimp, an innovative imputation method designed to infuse robustness into microbiome data analysis. GEMimp leverages the node2vec algorithm, which incorporates both Breadth-First Search (BFS) and Depth-First Search (DFS) strategies in its random walks sampling process. This approach enables GEMimp to learn nuanced, low-dimensional representations of each taxonomic unit, facilitating the reconstruction of their similarity networks with unprecedented accuracy.Our comparative analysis pits GEMimp against state-of-the-art imputation methods including SAVER, MAGIC and mbImpute. The results unequivocally demonstrate that GEMimp outperforms its counterparts by achieving the highest Pearson correlation coefficient when compared to the original raw dataset. Furthermore, GEMimp shows notable proficiency in identifying significant taxa, enhancing the detection of disease-related taxa and effectively mitigating the impact of sparsity on both simulated and real-world datasets, such as those pertaining to Type 2 Diabetes (T2D) and Colorectal Cancer (CRC). These findings collectively highlight the strong effectiveness of GEMimp, allowing for better analysis on microbial data. With alleviation of sparsity issues, it could be greatly facilitated in downstream analyses and even in the field of microbiology.

A geographic-semantic context-aware urban commuting flow prediction model using graph neural network

Urban commuting flow prediction is crucial for urban planning, transportation optimization, and supply chain management. Traditional models and machine learning solutions often fall short in capturing the complex dynamics of urban mobility, overlooking factors like spatial correlation and human mobility patterns. Recent deep learning advancements have shown promise by integrating spatial correlation through incorporating geographical adjacency. However, this approach may not fully capture spatial correlation, particularly the semantic adjacency effect of zones connected via public transportation lines, such as metro networks—a novel dimension overlooked in prior research. The metro station serves as a major transit hub, attracting passengers from various origins and destinations and correlating the flow between the connected zones and other zones within the metro network. Devising an optimized model to encapsulate these spatial correlations is another crucial area of research where scientists are actively seeking the most effective solutions. We propose a novel sequential hybrid approach leveraging graph-based learning, focusing on the underexplored impact of metro networks on commuting flow prediction. Our model integrates two types of adjacency—geographic and semantic—using Graph Convolutional Networks (GCNs) to embed geographic correlations among proximate zones and Graph Attention Networks (GATs) to incorporate semantic adjacency defined by metro line connectivity. Furthermore, our approach innovatively incorporates features directly from online maps, eliminating reliance on third-party resources and enhancing scalability and efficiency. We validate our model using real-world datasets from Montreal, Canada, comprising travel surveys and GPS trajectory types, totaling nearly 900,000 trip records. Through evaluation using established performance metrics such as MAE, RMSE, MAPE, and CPC, we demonstrate significant improvement in predictive accuracy compared to existing state-of-the-art models. Our research has practical implications for urban planning and infrastructure development, aiding policymakers and urban planners in analyzing the impact of new urban infrastructure on commuting flow dynamics and facilitating informed decisions regarding transportation and infrastructure planning.

Diversifying Collaborative Filtering via Graph Spreading Network and Selective Sampling.

Graph neural network (GNN) is a robust model for processing non-Euclidean data, such as graphs, by extracting structural information and learning high-level representations. GNN has achieved state-of-the-art recommendation performance on collaborative filtering (CF) for accuracy. Nevertheless, the diversity of the recommendations has not received good attention. Existing work using GNN for recommendation suffers from the accuracy-diversity dilemma, where slightly increases diversity while accuracy drops significantly. Furthermore, GNN-based recommendation models lack the flexibility to adapt to different scenarios' demands concerning the accuracy-diversity ratio of their recommendation lists. In this work, we endeavor to address the above problems from the perspective of aggregate diversity, which modifies the propagation rule and develops a new sampling strategy. We propose graph spreading network (GSN), a novel model that leverages only neighborhood aggregation for CF. Specifically, GSN learns user and item embeddings by propagating them over the graph structure, utilizing both diversity-oriented and accuracy-oriented aggregations. The final representations are obtained by taking the weighted sum of the embeddings learned at all layers. We also present a new sampling strategy that selects potentially accurate and diverse items as negative samples to assist model training. GSN effectively addresses the accuracy-diversity dilemma and achieves improved diversity while maintaining accuracy with the help of a selective sampler. Moreover, a hyper-parameter in GSN allows for adjustment of the accuracy-diversity ratio of recommendation lists to satisfy the diverse demands. Compared to the state-of-the-art model, GSN improved R @20 by 1.62%, N @20 by 0.67%, G @20 by 3.59%, and E @20 by 4.15% on average over three real-world datasets, verifying the effectiveness of our proposed model in diversifying overall collaborative recommendations.

Summary Reports Optimization in the Privacy Sandbox Attribution Reporting API

The Privacy Sandbox Attribution Reporting API has been recently deployed by Google Chrome to support the basic advertising functionality of attribution reporting (aka conversion measurement) after deprecation of third-party cookies. The API implements a collection of privacy-enhancing guardrails including contribution bounding and noise injection. It also offers flexibility for the analyst to allocate the contribution budget. In this work, we present algorithms for optimizing the allocation of the contribution budget for summary reports from the Attribution Reporting API. We develop a synthetic data model that we find to accurately capture real-world conversion data, and extensively evaluate our method on two real-world datasets and two synthetic datasets. We show that optimizing the parameters that can be set by the analyst can significantly improve the utility achieved by querying the API while satisfying the same privacy bounds.

TS-DM: A Time Segmentation-Based Data Stream Learning Method for Concept Drift Adaptation.

Concept drift arises from the uncertainty of data distribution over time and is common in data stream. While numerous methods have been developed to assist machine learning models in adapting to such changeable data, the problem of improperly keeping or discarding data samples remains. This may results in the loss of valuable knowledge that could be utilized in subsequent time points, ultimately affecting the model's accuracy. To address this issue, a novel method called time segmentation-based data stream learning method (TS-DM) is developed to help segment and learn the streaming data for concept drift adaptation. First, a chunk-based segmentation strategy is given to segment normal and drift chunks. Building upon this, a chunk-based evolving segmentation (CES) strategy is proposed to mine and segment the data chunk when both old and new concepts coexist. Furthermore, a warning level data segmentation process (CES-W) and a high-low-drift tradeoff handling process are developed to enhance the generalization and robustness. To evaluate the performance and efficiency of our proposed method, we conduct experiments on both synthetic and real-world datasets. By comparing the results with several state-of-the-art data stream learning methods, the experimental findings demonstrate the efficiency of the proposed method.

Large Language Model as Parking Planning Agent in the Context of Mixed Period of Autonomous Vehicles and Human-Driven Vehicles

Autonomous vehicles (AVs) are anticipated to revolutionize future transportation, necessitating updates to traffic infrastructure, particularly parking facilities, due to the unique characteristics of AVs compared to Human-Driven Vehicles (HDVs). During the transition period in which AVs and HDVs coexist, adaptable infrastructure is essential to accommodate both vehicle types. Traditional research, typically reliant on complex mathematical models and simulations, faces challenges in adapting to diverse urban settings, requiring substantial time and resources. To address these challenges, a government-level framework was developed, enabling urban planners to quickly and accurately evaluate and optimize existing parking facilities for future AV and HDV coexistence scenarios. The framework integrates a Large Language Model (LLM) to enhance flexibility and efficiency in parking planning throughout the transitional period. Structured guidance is incorporated to enhance decision-making precision and reduce LLM hallucination risks. The flexibility, robustness, and accuracy of the framework were validated through step-by-step and end-to-end testing using real-world datasets. Specifically, the framework achieved 91.1% comprehensiveness and 70.2% consistency in Indicator Selection Module testing, a 68.9% success rate in the Single Indicator Calculation Module, and a 66.7% success rate in end-to-end testing, demonstrating its practical value in supporting cities during AV integration. Finally, the success rates of different LLM agent modules were further explored, along with a comparison of multiple LLMs and an analysis of key issues related to LLM trustworthiness in urban planning applications. The research highlights the potential of LLMs in advancing urban planning processes and optimizing existing infrastructure, contributing to smarter and more adaptable urban environments.