Streaming classification systems must maintain robustness under the "holy trinity" of non-stationarity: concept drift, extreme class imbalance, and evolving label spaces (class-incremental learning). In these regimes, standard online learners and ensemble methods frequently suffer from catastrophic forgetting of minority concepts or lack mechanisms to accommodate newly emerging classes without retraining. In this paper, we introduce Wave-GA, an evolutionary framework that evolves a transparent linear model through cyclical, class-focused training waves. By rotating optimization pressure across classes and utilizing dynamic matrix expansion, Wave-GA integrates distinct mechanisms for drift adaptation (per-class replay buffers), imbalance handling (minority-aware blended fitness), and emerging labels (architectural growth). Extensive prequential evaluation on six synthetic benchmarks and three real-world streams demonstrates that Wave-GA prevents the minority-class collapse observed in state-of-the-art ensembles like Adaptive Random Forest. The approach improves average balanced accuracy by 6–12 percentage points over standard evolutionary baselines and delivers robust long-horizon stability with constant-time inference.

Recent advancements in artificial intelligence (AI) and the vast data generated by modern clinical systems have driven the development of AI solutions in medical imaging, encompassing image reconstruction, segmentation, diagnosis, and treatment planning. Despite these successes and potential, many stakeholders worry about the risks and ethical implications of imaging AI, viewing it as complex, opaque, and challenging to understand, use, and trust in critical clinical applications. The FUTURE-AI guideline for trustworthy AI in healthcare was established based on six guiding principles: Fairness, Universality, Traceability, Usability, Robustness, and Explainability. Through international consensus, a set of recommendations was defined, covering the entire lifecycle of medical AI tools, from design, development, and validation to regulation, deployment, and monitoring. In this paper, we describe how these specific recommendations can be instantiated in the domain of medical imaging, providing an overview of current best practices along with guidelines and concrete metrics on how those recommendations could be met, offering a valuable resource to the international medical imaging community.

The paper introduces White Rabbit, a novel method for discovering high-quality, meaningful paths between entities in online Knowledge Graphs (KGs). Traditional exploration methods, such as SPARQL endpoints, struggle due to the large size and complexity of KGs. The proposed approach addresses this by introducing the problem of context-aware path finding, ensuring that retrieved paths are coherent and involve highly relevant entities. White Rabbit uses embeddings to score entity neighbors, a queue-based prioritization mechanism, and an iterative refinement process to improve efficiency and relevance. The system is demonstrated live, allowing participants to test it and compare against baseline methods (structural approaches, pretrained embeddings, and large language models). Results show that White Rabbit enhances both the efficiency of exploration and the quality of discovered paths.

International audience

Property Graphs are a versatile and expressive data model that has gained widespread adoption due to their flexibility in supporting labeled and attributed nodes and edges. They are well-established in research communities and are becoming widespread in companies and organizations across various sectors. They have been boosted by a fervent ISO/IEC standardization activity, leading to dedicated query and schema languages. While the current standards are still evolving, opportunities remain to enrich them with features such as composability. The plethora of existing query languages reflects a rich and diverse ecosystem, which ongoing unification efforts aim to align. This tutorial aims to deepen the understanding of Property Graph standards by showcasing their strengths, highlighting recent unification efforts, clarifying the central role of schema constraints, and exploring the rich landscape of research and industrial opportunities shaping the future of graph data management.

Abstract Data streams which are now massively and constantly arriving from Internet of Things devices, sensors and social media, require efficient processing, querying and reasoning within a given timeframe. With this in mind, the RDF data model, the cornerstone of the Web of Data, supports a feature-rich stream processing ecosystem that takes into account the temporal dimension associated with events. These timestamped streams support advanced temporal analysis ranging from time-based queries, temporal anomaly detection to temporal reasoning. This survey is the first to provide a comprehensive overview of the field of RDF stream processing, focusing on (query) languages, systems, and benchmarks. For each of these areas, we present salient dimensions, propose a taxonomy of existing work, detail the concepts at the core of each approach and describe their main technical aspects and implementation. We hope that the survey will help readers understand this scientifically rich field and identify the most relevant method for various usage scenarios.

Informal care is fundamental in the wellbeing and resilience of elderly and people with chronic conditions. However, solutions for the effective collaboration of healthcare professionals, patients and informal carers are not yet widely available. CareKeeper builds on a state-of-the-art personal health system, augmenting it with Artificial Intelligence and Big Data technologies, to boost informal care coordination. In this paper we report on the design of the platform with the aim of providing a light-weighted communication solution to support practical challenges about sharing the responsibility of caring, such as the frequency of visits, support to routinely activities and timely intervention in case of emergency and need.

The continuous growth of high volumes of biomedical data in healthcare generates significant challenges for their efficient management. This data requires efficient management and analysis in order to derive meaningful and actionable information. Especially in the current situation of the COVID-19 pandemic, complications that might occur after the onset of this disease are important. Such a complication is Acute Respiratory Distress Syndrome (ARDS), which is a serious respiratory condition with high mortality and associated morbidity. A large number of basic and clinical studies demonstrated that early diagnosis and intervention are keys to improve the survival rate of patients with ARDS. Therefore, there is a pressing need for the development and clinical testing of predictive models for ARDS events, which might improve the clinical diagnosis or the management of ARDS. In this paper, we focus on two distinct objectives; namely a) to design a scalable data science platform, built on open source technologies able to streamline the development of such models, and b) to exploit the platform using publicly available big datasets to develop such models. To this direction, we employ random forests and logistic regression algorithmic models for the early prediction and diagnosis of ARDS. Our approach achieves better results in all metrics, when compared to relevant published efforts using the MIMIC III dataset.

Databases are at the core of virtually any software product. Changes to database schemas cannot be made in isolation, as they are intricately coupled with application code. Such couplings enforce collateral evolution, which is a recognised, important research problem. In this demonstration, we show a new dimension to this problem, in software that supports alternative database backends: vendor-specific SQL dialects necessitate a simultaneous evolution of both, database schema and program code, for all supported DB variants. These near-same changes impose substantial manual effort for software developers. We introduce DeBinelle, a novel framework and domain-specific language for semantic patches that abstracts DB-variant schema changes and coupled program code into a single, unified representation. DeBinelle further offers a novel alternative to manually evolving coupled schemas and code. DeBinelle considerably extends established, seminal results in software engineering research, supporting several programming languages, and the many dialects of SQL. It effectively eliminates the need to perform vendor-specific changes, replacing them with intuitive semantic patches. Our demo of DeBinelle is based on real-world use cases from reference systems for schema evolution.

Analytical queries over RDF data are becoming prominent as a result of the proliferation of knowledge graphs. Yet, RDF databases are not optimized to perform such queries efficiently, leading to long processing times. A well known technique to improve the performance of analytical queries is to exploit materialized views.Although popular in relational databases, view materialization for RDF and SPARQL has not yet transitioned into practice, due to the non-trivial application to the RDF graph model. Motivated by a lack of understanding of the impact of view materialization alternatives for RDF data, we demonstrate Sofos, a system that implements and compares several cost models for view materialization. Sofos is, to the best of our knowledge, the first attempt to adapt cost models, initially studied in relational data, to the generic RDF setting, and to propose new ones, analyzing their pitfalls and merits. Sofos takes an RDF dataset and an analytical query for some facet in the data, and compares and evaluates alternative cost models, displaying statistics and insights about time, memory consumption, and query characteristics.

The rapid explosion of linked data has resulted into many weakly structured and incomplete data sources, where typing information might be missing. On the other hand, type information is essential for a number of tasks such as query answering, integration, summarization and partitioning. Existing approaches for type discovery, either completely ignore type declarations available in the dataset (implicit type discovery approaches), or rely only on existing types, in order to complement them (explicit type enrichment approaches). Implicit type discovery approaches are based on instance grouping, which requires an exhaustive comparison between the instances. This process is expensive and not incremental. Explicit type enrichment approaches on the other hand, are not able to identify new types and they can not process data sources that have little or no schema information. In this paper, we present HInT, the first incremental and hybrid type discovery system for RDF datasets, enabling type discovery in datasets where type declarations are missing. To achieve this goal, we incrementally identify the patterns of the various instances, we index and then group them to identify the types. During the processing of an instance, our approach exploits its type information, if available, to improve the quality of the discovered types by guiding the classification of the new instance in the correct group and by refining the groups already built. We analytically and experimentally show that our approach dominates in terms of efficiency, competitors from both worlds, implicit type discovery and explicit type enrichment while outperforming them in most of the cases in terms of quality.

Semantic summaries try to extract compact information from the original RDF graph, while reducing its size. State of the art structural semantic summaries, focus primarily on the graph structure of the data, trying to maximize the summary’s utility for a specific purpose, such as indexing, query answering and source selection. In this paper, we present an approach that is able to construct high quality summaries, exploiting a small part of the query workload, maximizing their utility for query answering, i.e. the query coverage. We demonstrate our approach using two real world datasets and the corresponding query workloads and we show that we strictly dominates current state of the art in terms of query coverage.

During the burst of the coronavirus pandemic, in early-midst 2020, public health authorities worldwide considered appropriate identification, isolation and contact tracing as the most appropriate strategy for infection containment. This work presents an outbreak response tool, designed for public health authorities to effectively track suspect, probable and confirmed incidence cases in a pandemic by means of a mobile app used by citizens to provide immediate feedback. It is developed based on an already existing personal health record app, which has been extended to properly accommodate specific needs that emerged during the crisis. The aim is to better support human tracers and should not be confused with proximity tracking apps. It respects safety and security regulations, while at the same time it conforms to international standards and widely accepted medical protocols. Issues relevant to privacy concerns, and interoperability with available patient registries and data analytics tools are also examined to better support public healthcare delivery and contain the spread of the infection.

Coping with breast cancer and its consequences has now become a major socioeconomic challenge. The BOUNCE EU H2020 project aims at building a quantitative mathematical model of factors associated with optimal adjustment capacity to cancer. This paper gives an overview of the project targets and on the algorithmic methods focusing on modeling the psychological resilience trajectories during breast cancer treatment.

Breast cancer is the most common cancer disease in women and is rapidly becoming a chronic illness due recent advances in treatment methods. As such, coping with cancer has become a major socio-economic challenge leading to an increasing need for predicting resilience of women to the variety of stressful experiences and practical challenges they face. In this paper, we present the data infrastructure developed for this purpose, demonstrating the various components that will contribute to the developing the resilience trajectory predictor. Special emphasis is given to the semantic tier, presenting the project solution already implemented for effectively collecting, ingesting, cleaning, modelling and processing data that will be used throughout the lifetime of the project.

The widespread usage of electronic health record (EHR) systems and the prevalence of personal health apps has led to the digitization of huge quantities of health and medical data. However, the lack of plug and play interoperability between them is a major bottleneck, significantly restricting their exploitation potential. This paper presents a novel infrastructure developed for enabling the seamless integration between such systems, focusing on the use of state-of-the-art, widely accepted, interfaces. The infrastructure effectively supports end-user needs for integrated care, bringing linked EHR information from multiple providers at the point of care by means of personal health apps. Integration is provided through cross-enterprise document sharing (XDS) and fast healthcare interoperability resources (FHIR) services. The described infrastructure facilitates interoperability by enabling citizens to use a personal health app of their choosing for accessing own medical record information stored at multiple sites, by combining the benefits and potential of both XDS and FHIR.

Many modern applications produce massive streams of data series and maintain them in indexes to be able to explore them through nearest neighbor search. Existing data series indexes, however, are expensive to operate as they issue many random I/Os to storage. To address this problem, we recently proposed Coconut, a new infrastructure that organizes data series based on a new sortable format. In this way, Coconut is able to leverage state-of-the-art indexing techniques that rely on sorting for the first time to build, maintain and query data series indexes using fast sequential I/Os. In this demonstration, we present Coconut Palm, a new exploration tool that allows to interactively combine different indexing techniques from within the Coconut infrastructure and to thereby seamlessly explore data series from across various scientific domains. We highlight the rich indexing design choices that Coconut opens up, and we present a new recommender tool that allows users to intelligently navigate them for both static and streaming data exploration scenarios.

Many modern applications produce massive amounts of data series that need to be analyzed, requiring efficient similarity search operations. However, the state-of-the-art data series indexes that are used for this purpose do not scale well for massive datasets in terms of performance, or storage costs. We pinpoint the problem to the fact that existing summarizations of data series used for indexing cannot be sorted while keeping similar data series close to each other in the sorted order. This leads to two design problems. First, traditional bulk-loading algorithms based on sorting cannot be used. Instead, index construction takes place through slow top-down insertions, which create a non-contiguous index that results in many random I/Os. Second, data series cannot be sorted and split across nodes evenly based on their median value; thus, most leaf nodes are in practice nearly empty. This further slows down query speed and amplifies storage costs. To address these problems, we present Coconut. The first innovation in Coconut is an inverted, sortable data series summarization that organizes data series based on a z-order curve, keeping similar series close to each other in the sorted order. As a result, Coconut is able to use bulk-loading techniques that rely on sorting to quickly build a contiguous index using large sequential disk I/Os. We then explore prefix-based and median-based splitting policies for bottom-up bulk-loading, showing that median-based splitting outperforms the state of the art, ensuring that all nodes are densely populated. Overall, we show analytically and empirically that Coconut dominates the state-of-the-art data series indexes in terms of construction speed, query speed, and storage costs.

As the number of clinical guidelines and rules for effective management of cancer therapy is rapidly increasing decision support systems are more and more required. To this direction, in this paper, we present a collaborative knowledge management system for cancer diseases leading to decision support and intelligent diagnosis. Clinicians can specify a variety of knowledge rules in a collaborative fashion. Then, those rules are applied on top of patient data collected within a personal health record. The generated knowledge is formulated as a free text and returned back to the clinicians to support them and enhance the communication with their patients.

Cancer research has led to more cancer patients being cured, and many more enabled to live with their cancer. As such, some cancers are now considered a chronic disease, where patients and their families face the challenge to take an active role in their own care and in some cases in their treatment. To this direction the iManageCancer project aims to provide a cancer specific self-management platform designed according to the needs of patient groups while focusing, in parallel, on the wellbeing of the cancer patient. In this paper, we present the use-case requirements collected using a survey, a workshop and the analysis of three white papers and then we explain the corresponding system architecture. We describe in detail the main technological components of the designed platform, show the current status of development and we discuss further directions of research.

This visual paper aims at proposing a framework for detecting depression in cancer patients using prosodic and statistical features extracted by speech, while chatting with a virtual coach.

As knowledge bases are constantly evolving, there is a clear need for monitoring and analyzing the changes that occur on them. Traditional approaches for studying the evolution of data focus on providing humans with deltas that include loads of information. In this work, we envision a processing model that recommends evolution measures taking into account particular challenges, such as relatedness, transparency, diversity, fairness and anonymity. We target at supporting humans with complementary measures that offer high-level overviews of the changes to help them understand how data of interest evolves.

During the last decade, the number of users who look for health-related information has impressively increased. On the other hand, health professionals have less and less time to recommend useful sources of such information online to their patients. To this direction, we target at streamlining the process of providing useful online information to patients by their caregivers and improving as such the opportunities that patients have to inform themselves online about diseases and possible treatments. Using our system, relevant and high quality information is delivered to patients based on their profile, as represented in their personal healthcare record data, facilitating an easy interaction by minimizing the necessary manual effort. Specifically, in this paper, we propose a model for group recommendations following the collaborative filtering approach. Since in collaborative filtering is crucial to identify the correct set of similar users for a user in question, in addition to the traditional ratings, we pay particular attention on how to exploit healthrelated information for computing similarities between users. Our special focus is on providing valuable suggestions to a caregiver who is responsible for a group of users. We interpret valuable suggestions as suggestions that are both highly related and fair to the users of the group. In this line, we propose an algorithm for identifying the top-z most valuable recommendations, and present its implementation in MapReduce.

Over the last few years we witness an explosion on the development of data management solutions for big data applications. To this direction NoSQL databases provide new opportunities by enabling elastic scaling, fault tolerance, high availability and schema flexibility. Despite these benefits, their limitations in the flexibility of query mechanisms impose a real barrier for any application that has not predetermined access use-cases. One of the main reasons for this bottleneck is that NoSQL databases do not support joins. In this poster we present a solution that efficiently supports joins over such databases. More specifically, we present a query optimization and execution module placed on top of Cassandra clusters that is able to efficiently combine information stored in different columnfamilies. Our preliminary evaluation demonstrates the feasibility of our solution and the advantages gained when compared to a recent commercial solution by DataStax. To the best of our knowledge our approach is the first and the only available open source solution allowing joins over NoSQL Cassandra databases.

The digital avatar is a vision for the digital representation of personal health status in body centric views. It is designed as an integrated facility that allows collection of, access to and sharing to life-long and consistent data. A number of Virtual Physiological Human (VPH) communities have started the movement to this direction by creating a digital patient road-map and by supporting data sharing infrastructures. As an innovative concept, the impact of digital patient and avatar to personalized medicine and treatment is yet to be clear. This requires a focused and concerted effort in addressing various questions regarding user perspective, use cases and scenarios. This paper presents use cases and future scenarios realizing the vision for the digital avatar as well as architectural consideration for the envisaged platform.

Personalized medicine refers to the tailoring of treatment to the individual characteristics of a patient. Part of the personalized medicine is the patient profiling and the communicative relation between physician and patient. The ways of exchanging information, the nature of the information itself and the information assimilation capabilities of the patient can assist the physicians to have a better understanding. Taking advantage of these information sources, a smart environment could be implemented. This environment will be able to act as a decision support infrastructure to support the communication, interaction and information delivery process from the doctor to the patient. A prerequisite of personalized delivery of information and intelligent guidance of the patient into his/her treatment plans is our ability to develop an appropriate and accurate profile of the patient. In this paper we present a collaborative platform which will empower patient with knowledge about his/her health condition and at the same time it will assist the physician to have a better understanding about the patient's unique psychological profile. We also introduce the p-medicine project and its vision in the field of personalized medicine and show project's approach on patient empowerment.

The evolution of ontologies is an undisputed necessity in ontology-based data integration. Yet, few research efforts have focused on addressing the need to reflect ontology evolution onto the underlying data integration systems. We present Exelixis, a web platform that enables query answering over evolving ontologies without mapping redefinition. This is achieved by rewriting queries among ontology versions. First, changes between ontologies are automatically detected and described using a high level language of changes. Those changes are interpreted as sound global-as-view (GAV) mappings. Then query expansion is applied in order to consider constraints from the ontology and unfolding to apply the GAV mappings. Whenever equivalent rewritings cannot be produced we a) guide query redefinition and/or b) provide the best "over-approximations", i.e. the minimally-containing and minimally-generalized rewritings. For the demonstration we will use four versions of the CIDOC-CRM ontology and real user queries to show the functionality of the system. Then we will allow conference participants to directly interact with the system to test its capabilities.

The aim of this study is to identify a gene expression signature which is characteristic of ER status in breast cancer patients. To our knowledge, this is the first microarray study in Greece involving clinical samples. We identified 97 genes that are characteristic for ER status and can well distinguish the ER+ from the ER- samples. We shrank our list to a 11-gene list correlating to the same patient ER status. We found a significant overlap of these genes with published ER status characteristic signatures like the ones of West et.al. and of Vanpsilat Veer et. al.. This fact is very important given the minimal overlap of such genes reported by others . In order to obtain a molecular insight into how the expression of estrogen receptor activates cancer cells, we found associations with biological pathways. Interestingly, the vast majority of these genes are highly related to breast cancer.

Detecting proteins in human blood holds the promise of a revolution in cancer diagnosis. Also, the ability to perform laboratory operations on small scales using miniaturized (lab-on-a-chip) devices has many benefits. Designing and fabricating such systems is extremely challenging, but physicists and engineers are beginning to construct such highly integrated and compact labs on chips with exciting functionality. This paper focuses on the presentation of the information technology layer in such an integrated platform that has been developed in the LOCCANDIA project. LOCCANDIA ultimate objective is to develop an innovative nano-technology based (lab-on-a-chip) platform for the medical-proteomics field. The paper presents the main engineering aspects and the architecture of the integrated Clinico-Proteomic environment.