Large-scale Electronic Health Record (EHR) databases have become indispensable in supporting clinical decision-making through data-driven treatment recommendations. However, existing medication recommender methods often struggle with a user (i.e., patient) cold-start problem, where recommendations for new patients are usually unreliable due to the lack of sufficient prescription history for patient profiling. While prior studies have utilized medical knowledge graphs to connect medication concepts through pharmacological or chemical relationships, these methods primarily focus on mitigating the item cold-start issue and fall short in providing personalized recommendations that adapt to individual patient characteristics. Meta-learning has shown promise in handling new users with sparse interactions in recommender systems. However, its application to EHRs remains underexplored due to the unique sequential structure of EHR data. To tackle these challenges, we propose MetaDrug, a multi-level, uncertainty-aware meta-learning framework designed to address the patient cold-start problem in medication recommendation. MetaDrug proposes a novel two-level meta-adaptation mechanism, including self-adaptation, which adapts the model to new patients using their own medical events as support sets to capture temporal dependencies; and peer-adaptation, which adapts the model using similar visits from peer patients to enrich new patient representations. Meanwhile, to further improve meta-adaptation outcomes, we introduce an uncertainty quantification module that ranks the support visits and filters out the unrelated information for adaptation consistency. We evaluate our approach on the MIMIC-III and Acute Kidney Injury (AKI) datasets. Experimental results on both datasets demonstrate that MetaDrug consistently outperforms state-of-the-art medication recommendation methods on cold-start patients.

Medical ontology graphs map external knowledge to medical codes in electronic health records (EHRs) via structured relationships. By leveraging domain-approved connections (e.g., parent-child), predictive models can generate richer medical concept representations by incorporating contextual information from related concepts. However, existing literature primarily focuses on incorporating domain knowledge from a single ontology system, or from multiple ontology systems (e.g., diseases, drugs, and procedures) in isolation, without integrating them into a unified learning structure. Consequently, concept representation learning often remains limited to intra-ontology relationships, overlooking cross-ontology connections that could enhance the richness of healthcare representations. In this paper, we propose LINKO, a large language model (LLM)-augmented integrative ontology learning framework that leverages multiple ontology graphs simultaneously by enabling dual-axis knowledge propagation both within and across heterogeneous ontology systems to enhance medical concept representation learning. Specifically, LINKO first employs LLMs to provide a graph-retrieval-augmented initialization for ontology concept embedding, through an engineered prompt that includes concept descriptions, and is further augmented with ontology graph relations and task-specific details. Second, our method jointly learns the medical concepts in diverse ontology graphs by performing knowledge propagation in two axes: (1) intra-ontology vertical propagation across hierarchical ontology levels and (2) inter-ontology horizontal propagation within every level in parallel. Last, through extensive experiments on two public datasets, we validate the superior performance of LINKO over state-of-the-art baselines. As a plug-in encoder compatible with existing EHR predictive models, LINKO further demonstrates enhanced robustness in scenarios involving limited data availability and rare disease prediction.

Iterative feature space optimization includes continuously evaluating and refining the feature space to improve downstream task performance. However, existing methods commonly suffer from three major limitations: 1) ignoring differences between samples leads to evaluation bias; 2) the feature space is overly tailored to specific models, resulting in overfitting and poor generalization; and 3) retraining the evaluator from scratch in each iteration significantly reduces overall efficiency. To bridge these gaps, we introduce EASE (gEneralized Adaptive feature Space Evaluator), a generalized framework for efficient and objective evaluation of iteratively generated feature spaces. This framework includes two key components: Feature-Sample Subspace Generator and Contextual Attention Evaluator. The first component aims to mitigate evaluation bias by decoupling the information distribution within the feature space. To achieve this, based on feedback from the subsequent evaluator, we identify the samples most challenging for evaluation and the features most relevant to prediction tasks. The second component intends to incrementally capture evolving patterns of the feature space for efficient evaluation. Specifically, we propose a weighted-sharing multi-head attention mechanism to encode the feature space into an embedding vector for evaluation, and update the evaluator incrementally to retain prior knowledge while incorporating new information. Extensive experiments on fifteen public datasets demonstrate the effectiveness of EASE. We have released our code and data to the public.

Feature selection removes redundant features to enhance both performance and computational efficiency in downstream tasks. Existing methods often struggle to capture complex feature interactions and adapt to diverse scenarios. Recent advances in this domain have incorporated generative intelligence to address these drawbacks by uncovering intricate relationships between features. However, two key limitations remain: 1) embedding feature subsets in a continuous space is challenging due to permutation sensitivity, as changes in feature order can introduce biases and weaken the embedding learning process; 2) gradient-based search in the embedding space assumes convexity, which is rarely guaranteed, leading to reduced search effectiveness and suboptimal subsets. To address these limitations, we propose a new framework that can: 1) preserve feature subset knowledge in a continuous embedding space while ensuring permutation invariance; 2) effectively explore the embedding space without relying on strong convex assumptions. For the first objective, we develop an encoder-decoder paradigm to preserve feature selection knowledge into a continuous embedding space. This paradigm captures feature interactions through pairwise relationships within the subset, removing the influence of feature order on the embedding. Moreover, an inducing point mechanism is introduced to accelerate pairwise relationship computations. For the second objective, we employ a policy-based reinforcement learning (RL) approach to guide the exploration of the embedding space. The RL agent effectively navigates the space by balancing multiple objectives. By prioritizing high-potential regions adaptively and eliminating the reliance on convexity assumptions, this search strategy effectively reduces the risk of converging to local optima. Finally, we conduct extensive experiments to demonstrate the effectiveness, efficiency, robustness and explicitness of our model. Our code and dataset are publicly accessible on GitHub. https://github.com/RayLiu1103/CAPS.

Over the past decade, significant advancements in mining electronic health records (EHRs) have enabled a broad range of decision-support applications and offered an unprecedented capacity for predicting critical events such as disease prognosis and mortality in healthcare. Despite the availability of comprehensive coding systems in EHRs (e.g., ICD-9), which are designed to record diverse information on diseases, procedures, and medications over time, the complex and dynamic dependencies among the recorded data are usually not captured. This limitation often hinders the contextual understanding of medical observations for effective EHR representation learning. Therefore, there is a compelling need to discover a hidden “EHR graph” that represents the medical relationship between the observed features according to a patient’s history. These hidden graphs consisting of the medical codes from the same visits can offer a comprehensive insight derived from disease-to-disease, disease-to-drug, and drug-to-drug dependencies. However, it is still unclear how to address the challenge that the dependencies may vary from patient to patient, and they can dynamically evolve from one visit to another. To this end, we propose Time-aware Personalized Graph Transformer (TPGT), a novel attention-based time-aware hidden graph model, that captures the personalized graphical structures among observed medical codes and summarizes the temporal code dependencies over time to improve patient representation for outcome prediction. Built upon an intra-visit and an inter-visit dual-attention mechanism to model patients’ EHR graphs, our model offers an interpretability of what diagnosis or medication in a patient’s history can interact, and how those interactions may change over time. We conduct extensive experiments on two real-world EHR datasets for different healthcare predictive tasks: acute kidney injury (AKI) prediction and ICU mortality prediction. The experimental results demonstrate a significant performance improvement of the proposed model over baselines through multi-aspect quantitative evaluation. Furthermore, we perform various qualitative studies to validate the interpretability of the model which highlights the application of the proposed method in the context of personalized medicine.

Recent advancements in sequential modeling applied to Electronic Health Records (EHR) have greatly influenced prescription recommender systems. While the recent literature on drug recommendation has shown promising performance, the study of discovering a diversity of coexisting temporal relationships at the level of medical codes over consecutive visits remains less explored. The goal of this study can be motivated from two perspectives. First, there is a need to develop a sophisticated sequential model capable of disentangling the complex relationships across sequential visits. Second, it is crucial to establish multiple and diverse health profiles for the same patient to ensure a comprehensive consideration of different medical intents in drug recommendation. To achieve this goal, we introduce Attentive Recommendation with Contrasted Intents (ARCI), a multi-level transformer-based method designed to capture the different but coexisting temporal paths across a shared sequence of visits. Specifically, we propose a novel intent-aware method with contrastive learning, that links specialized medical intents of the patients to the transformer heads for extracting distinct temporal paths associated with different health profiles. We conducted experiments on two real-world datasets for the prescription recommendation task using both ranking and classification metrics. Our results demonstrate that ARCI has outperformed the state-of-the-art prescription recommendation methods and is capable of providing interpretable insights for healthcare practitioners.

Survival analysis plays a crucial role in many healthcare decisions, where the risk prediction for the events of interest can support an informative outlook for a patient's medical journey. Given the existence of data censoring, an effective way of survival analysis is to enforce the pairwise temporal concordance between censored and observed data, aiming to utilize the time interval before censoring as partially observed time-to-event labels for supervised learning. Although existing studies mostly employed ranking methods to pursue an ordering objective, contrastive methods which learn a discriminative embedding by having data contrast against each other, have not been explored thoroughly for survival analysis. Therefore, in this paper, we propose a novel Ontology-aware Temporality-based Contrastive Survival (OTCSurv) analysis framework that utilizes survival durations from both censored and observed data to define temporal distinctiveness and construct negative sample pairs with adjustable hardness for contrastive learning. Specifically, we first use an ontological encoder and a sequential self-attention encoder to represent the longitudinal EHR data with rich contexts. Second, we design a temporal contrastive loss to capture varying survival durations in a supervised setting through a hardness-aware negative sampling mechanism. Last, we incorporate the contrastive task into the time-to-event predictive task with multiple loss components. We conduct extensive experiments using a large EHR dataset to forecast the risk of hospitalized patients who are in danger of developing acute kidney injury (AKI), a critical and urgent medical condition. The effectiveness and explainability of the proposed model are validated through comprehensive quantitative and qualitative studies.

For care of chronic diseases (e.g., depression, diabetes, hypertension), it is critical to identify effective treatment pathways that aim to promptly update the medication following the change of patient state and disease progression. This task is challenging because the optimal treatment pathway for each patient needs to be personalized due to the significant heterogeneity among individuals. Therefore, it is naturally promising to investigate how to use the abundant electronic health records to recommend effective and safe prescriptions. However, prescription recommendation needs to consider multiple aspects of life-critical evidence, such as the information relevance in terms of medical concepts, the health condition in terms of diagnosis history, and the further constraint in terms of side information (e.g., patient demographics and drug side effects). To this end, in this article, we propose a novel prescription recommendation framework named OntoPath to predict the next drug in disease treatment pathways, by building an ontology-aware hierarchical-attention model that integrates multiple medical evidence from domain knowledge guidance, medical history profiling, and side information utilization. Specifically, our method can be characterized from three aspects: (1) by incorporating the longitudinal diagnosis history, we enrich the profiling of patients in terms of comprehensive health conditions, which can largely influence a drug’s outcome on individual patients; (2) using the hierarchical disease and drug ontology structures, we are able to model the domain-specific relevance between patients and drugs at multiple levels of granularity and achieve in-depth collaborative filtering; (3) we introduce a pre-training stage to enhance the discriminativeness of network representations, which helps us obtain a premium model initialization to further boost the final recommendation training. We perform extensive experiments on a large-scale depression cohort with over 37,000 patients from a real-world medical claims database. The quantitative and qualitative results demonstrate the effectiveness of OntoPath through the consistent outperformance over state-of-the-art prescription recommendation baselines and the interpretation of model mechanism in case studies.

Identifying the dynamic functions of different urban zones enables a variety of smart city applications, such as intelligent urban planning, real-time traffic scheduling, and community precision management. Traditional urban function research using government administrative zoning systems is often conducted in a coarse resolution with fixed split, and ignore the reshaping of zones by city growth. To solve this problem, we propose a two-stage framework in order to represent the high-definition distribution of urban function across the city, by analyzing continuous human traces extracted from the dense, widespread, and full-time cellular data. At the representation stage, we embed the locations of base stations by modeling the user movements with staying and transfer events, along with the consideration of dynamic trip purposes in continuous human traces. At the annotation stage, we first divide the city into the finest unit zones and each covers at least one base station. By clustering the base stations, we further group the unit zones into functional zones. Last, we annotate functional zones based on the local point-of-interest (POI) information. In experiments, we evaluate the proposed high-definition function study in two tasks: (i) in-zone crowd flow prediction, and (ii) zone-enhanced POI recommendation. The results demonstrate the advantage of the proposed method with both the effectiveness of city split and the high-quality function annotation.

Spatial co-location patterns are subsets of spatial features usually located together in geographic space. Recent literature has provided different approaches to discover co-location patterns over point spatial data. However, most approaches consider the neighborhood relationship among spatial objects as binary and are mainly designed for point spatial features, thus are not appropriate for extended spatial features such as line strings and polygons, the neighborhood relationship among which is naturally continuous. This paper adopts a buffer-based model for measuring the spatial relationship of extended objects and mining co-location patterns. While the buffer-based model has several advantages for extended spatial features, it involves high computational complexity due to the expensive buffer-level overlay operation. To tackle this challenge, we introduce a coarse-level co-location mining framework, which follows a filter-and-refine paradigm. Within the framework, we develop a serious of rigorous upper bounds based on geometric property and progressively prune search space with these upper bounds. Moreover, we develop a join-less schema to further reduce computation cost of size-k(k > 2) co-location patterns. Finally, we conduct experiments with large-scale spatial data to validate the efficiency of the developed algorithms against several state-of-art methods. All experimental results demonstrate the superiority of our methods.

Point-of-interest (POI) recommendation, which provides personalized recommendation of places to mobile users, is an important task in location-based social networks (LBSNs). Unlike traditional interest-oriented merchandise recommendation, POI recommendation is more complex due to the timing effects: we need to examine whether the POI fits a user»s availability. While there are some prior studies which consider temporal effects by solely using check-in timestamps for modeling, they suffer from check-in data sparsity. Recent years, the advent in positioning technology has accumulated a variety of urban data related to human mobility. There is a potential to exploit human mobility patterns from heterogeneous information sources for improving POI recommendation. To this end, we propose a novel method which incorporates the degree of temporal matching between users and POIs into personalized POI recommendations. Specifically, we profile the temporal popularity of POIs, learn the latent regularity to characterize users, and conduct comprehensive experiments with real-world data. Evaluation results demonstrate the effectiveness of the proposed method.

Word evolution refers to the changing meanings and associations of words throughout time, as a byproduct of human language evolution. By studying word evolution, we can infer social trends and language constructs over different periods of human history. However, traditional techniques such as word representation learning do not adequately capture the evolving language structure and vocabulary. In this paper, we develop a dynamic statistical model to learn time-aware word vector representation. We propose a model that simultaneously learns time-aware embeddings and solves the resulting alignment problem. This model is trained on a crawled NYTimes dataset. Additionally, we develop multiple intuitive evaluation strategies of temporal word embeddings. Our qualitative and quantitative tests indicate that our method not only reliably captures this evolution over time, but also consistently outperforms state-of-the-art temporal embedding approaches on both semantic accuracy and alignment quality.

Point of interest (POI) recommendation, which provides personalized recommendation of places to mobile users, is an important task in location-based social networks (LBSNs). However, quite different from traditional interest-oriented merchandise recommendation, POI recommendation is more complex due to the timing effects: we need to examine whether the POI fits a user's availability. While there are some prior studies which included the temporal effect into POI recommendations, they overlooked the compatibility between time-varying popularity of POIs and regular availability of users, which we believe has a non-negligible impact on user decision-making. To this end, in this paper, we present a novel method which incorporates the degree of temporal matching between users and POIs into personalized POI recommendations. Specifically, we first profile the temporal popularity of POIs to show when a POI is popular for visit by mining the spatio-temporal human mobility and POI category data. Secondly, we propose latent user regularities to characterize when a user is regularly available for exploring POIs, which is learned with a user-POI temporal matching function. Finally, results of extensive experiments with real-world POI check-in and human mobility data demonstrate that our proposed user-POI temporal matching method delivers substantial advantages over baseline models for POI recommendation tasks.

It is well recognized that community safety which affects people's right to live without fear of crime has considerable impacts on housing investments. Housing investors can make more informed decisions if they are fully aware of safety related factors. To this end, we develop a safety-aware house ranking method by incorporating community safety into house assessment. Specifically, we first propose a novel framework to infer community safety level by mining community crime evidences from rich spatio-temporal historical crime data. Then we develop a ranking model which fuses multiply community safety features to rank house value based on the degree of community safety. Finally, we conduct a comprehensive evaluation of the proposed method with real-world crime and house data. The experimental results show that the proposed method substantially outperforms the baseline methods for house ranking.

It is traditionally a challenge for home buyers to understand, compare, and contrast the investment value of real estate. Although a number of appraisal methods have been developed to value real properties, the performances of these methods have been limited by traditional data sources for real estate appraisal. With the development of new ways of collecting estate-related mobile data, there is a potential to leverage geographic dependencies of real estate for enhancing real estate appraisal. Indeed, the geographic dependencies of the investment value of an estate can be from the characteristics of its own neighborhood (individual), the values of its nearby estates (peer), and the prosperity of the affiliated latent business area (zone). To this end, in this paper, we propose a geographic method, named ClusRanking, for real estate appraisal by leveraging the mutual enforcement of ranking and clustering power. ClusRanking is able to exploit geographic individual, peer, and zone dependencies in a probabilistic ranking model. Specifically, we first extract the geographic utility of estates from geography data, estimate the neighborhood popularity of estates by mining taxicab trajectory data, and model the influence of latent business areas. Also, we fuse these three influential factors and predict real estate investment value. Moreover, we simultaneously consider individual, peer and zone dependencies, and derive an estate-specific ranking likelihood as the objective function. Furthermore, we propose an improved method named CR-ClusRanking by incorporating checkin information as a regularization term which reduces the performance volatility of real estate ranking system. Finally, we conduct a comprehensive evaluation with the real estate-related data of Beijing, and the experimental results demonstrate the effectiveness of our proposed methods.

The problem of point of interest (POI) recommendation is to provide personalized recommendations of places, such as restaurants and movie theaters. The increasing prevalence of mobile devices and of location based social networks (LBSNs) poses significant new opportunities as well as challenges, which we address. The decision process for a user to choose a POI is complex and can be influenced by numerous factors, such as personal preferences, geographical considerations, and user mobility behaviors. This is further complicated by the connection LBSNs and mobile devices. While there are some studies on POI recommendations, they lack an integrated analysis of the joint effect of multiple factors. Meanwhile, although latent factor models have been proved effective and are thus widely used for recommendations, adopting them to POI recommendations requires delicate consideration of the unique characteristics of LBSNs. To this end, in this paper, we propose a general geographical probabilistic factor model ($\sf{Geo}$-PFM) framework which strategically takes various factors into consideration. Specifically, this framework allows to capture the geographical influences on a user’s check-in behavior. Also, user mobility behaviors can be effectively leveraged in the recommendation model. Moreover, based our$\sf{Geo}$-PFM framework, we further develop a Poisson$\sf{Geo}$-PFM which provides a more rigorous probabilistic generative process for the entire model and is effective in modeling the skewed user check-in count data as implicit feedback for better POI recommendations. Finally, extensive experimental results on three real-world LBSN datasets (which differ in terms of user mobility, POI geographical distribution, implicit response data skewness, and user-POI observation sparsity), show that the proposed recommendation methods outperform state-of-the-art latent factor models by a significant margin.

Ranking residential real estates based on investment values can provide decision making support for home buyers and thus plays an important role in estate marketplace. In this paper, we aim to develop methods for ranking estates based on investment values by mining users' opinions about estates from online user reviews and offline moving behaviors (e.g., Taxi traces, smart card transactions, check-ins). While a variety of features could be extracted from these data, these features are Interco related and redundant. Thus, selecting good features and integrating the feature selection into the fitting of a ranking model are essential. To this end, in this paper, we first strategically mine the fine-grained discrminative features from user reviews and moving behaviors, and then propose a probabilistic sparse pair wise ranking method for estates. Specifically, we first extract the explicit features from online user reviews which express users' opinions about point of interests (POIs) near an estate. We also mine the implicit features from offline moving behaviors from multiple perspectives (e.g., Direction, volume, velocity, heterogeneity, topic, popularity, etc.). Then we learn an estate ranking predictor by combining a pair wise ranking objective and a sparsity regularization in a unified probabilistic framework. And we develop an effective solution for the optimization problem. Finally, we conduct a comprehensive performance evaluation with real world estate related data, and the experimental results demonstrate the competitive performance of both features and the proposed model.

It is traditionally a challenge for home buyers to understand, compare and contrast the investment values of real estates. While a number of estate appraisal methods have been developed to value real property, the performances of these methods have been limited by the traditional data sources for estate appraisal. However, with the development of new ways of collecting estate-related mobile data, there is a potential to leverage geographic dependencies of estates for enhancing estate appraisal. Indeed, the geographic dependencies of the value of an estate can be from the characteristics of its own neighborhood (individual), the values of its nearby estates (peer), and the prosperity of the affiliated latent business area (zone). To this end, in this paper, we propose a geographic method, named ClusRanking, for estate appraisal by leveraging the mutual enforcement of ranking and clustering power. ClusRanking is able to exploit geographic individual, peer, and zone dependencies in a probabilistic ranking model. Specifically, we first extract the geographic utility of estates from geography data, estimate the neighborhood popularity of estates by mining taxicab trajectory data, and model the influence of latent business areas via ClusRanking. Also, we use a linear model to fuse these three influential factors and predict estate investment values. Moreover, we simultaneously consider individual, peer and zone dependencies, and derive an estate-specific ranking likelihood as the objective function. Finally, we conduct a comprehensive evaluation with real-world estate related data, and the experimental results demonstrate the effectiveness of our method.

If properly analyzed, the multi-aspect rating data could be a source of rich intelligence for providing personalized restaurant recommendations. Indeed, while recommender systems have been studied for various applications and many recommendation techniques have been developed for general or specific recommendation tasks, there are few studies for restaurant recommendation by addressing the unique challenges of the multi-aspect restaurant reviews. As we know, traditional collaborative filtering methods are typically developed for single aspect ratings. However, multi-aspect ratings are often collected from the restaurant customers. These ratings can reflect multiple aspects of the service quality of the restaurant. Also, geographic factors play an important role in restaurant recommendation. To this end, in this paper, we develop a generative probabilistic model to exploit the multi-aspect ratings of restaurants for restaurant recommendation. Also, the geographic proximity is integrated into the probabilistic model to capture the geographic influence. Moreover, the profile information, which contains customer/restaurant-independent features and the shared features, is also integrated into the model. Finally, we conduct a comprehensive experimental study on a real-world data set. The experimental results clearly demonstrate the benefit of exploiting multi-aspect ratings and the improvement of the developed generative probabilistic model.

The problem of point of interest (POI) recommendation is to provide personalized recommendations of places of interests, such as restaurants, for mobile users. Due to its complexity and its connection to location based social networks (LBSNs), the decision process of a user choose a POI is complex and can be influenced by various factors, such as user preferences, geographical influences, and user mobility behaviors. While there are some studies on POI recommendations, it lacks of integrated analysis of the joint effect of multiple factors. To this end, in this paper, we propose a novel geographical probabilistic factor analysis framework which strategically takes various factors into consideration. Specifically, this framework allows to capture the geographical influences on a user's check-in behavior. Also, the user mobility behaviors can be effectively exploited in the recommendation model. Moreover, the recommendation model can effectively make use of user check-in count data as implicity user feedback for modeling user preferences. Finally, experimental results on real-world LBSNs data show that the proposed recommendation method outperforms state-of-the-art latent factor models with a significant margin.