Given a set of historical, unlabeled patient records, how can we model the relationships between various concepts related to health conditions and treatments? Modeling health data as knowledge graphs can aid in better understanding and mining recurrent and abnormal patterns within massive amounts of records. However, for generic data modeling, the concepts and relationships of the data are often defined manually, which can be laborious and prone to human errors. The lack of structure in textual reports makes modeling challenging, as there is typically no standard in information, terminology, or other elements. This work proposes CliniKG for utilizing Large Language Models (LLMs) to structure patient records, enabling the modeling of health data concepts. First, we employ LLMs with zero-, or few-shot learning to define the graph's relationships of pairwise concepts. Based on the resulting modeling, we extract meaningful features from nodes and define their semantics automatically. With data visualizations, CliniKG highlights key findings, such as recurrent or rare relationships. The experimental evaluation shows CliniKG in action using a real dataset from a public hospital in Brazil. We performed a qualitative analysis with 40 domain experts to evaluate medium-sized LLMs and prompt con-figurations. The study reveals interesting patterns automatically identified within the data. CliniKG exhibits linear performance relative to the number of nodes, and the visual tools can assist specialists in monitoring patients' conditions.

Given a patient’s series of exams conducted over time, how can we identify cases with similar abnormalities or symptoms? Hospitals and medical facilities continuously monitor patients through periodic exams, a crucial practice for assessing their current condition and potential progression, thereby supporting decision-making. However, similarity-based searches often consider several exams of a patient, most times overlooking the temporal aspect, which is crucial for patient monitoring. In this paper, we present: (1) a novel similarity search framework that identifies similar cases based on symptoms while considering the temporal evolution of the patients’ conditions; and (2) a novel similarity function, called GCWei function, which is built upon the traditional Levenshtein similarity and improves the quality of the search by penalizing the similarity between non-related sets of symptoms. To identify relations, GCWei relies on well-established graph community detection procedures using all patients’ historical data. By combining (1) and (2), we obtain a search approach called GCWei-based search, which efficiently retrieves similar cases with similar developments and thus gives the specialist a broader view of the patient’s condition based on past cases of other patients. To demonstrate the value of our approach, we evaluate it both quantitatively and qualitatively using the recent and publicly available MIMIC-IV database.

Given a large, unlabeled set of Electronic Health Records (EHRs) acquired from multiple hospitals, how can we analyze the available entities and identify relationships in the data? Also, how can we perform Exploratory Data Analysis (EDA) over such EHR data? Many medical institutions generate EHRs as tabular data with entities and attributes in common. However, due to a large number of records, attributes, and high cardinality, exploring the different datasets and finding patterns and insights become laborious and prone to errors. In this work, we propose GraF- Eda for EDA over EHR data from different institutions. GraF-EDA models EHRs as time-evolving graphs, allowing the interoperability of such data into a single representation. We extract meaningful features from the graph nodes and provide intuitive visualizations to improve data explainability. We evaluate GraF-EDA with four COVID-19 datasets from hospitals of the São Paulo state, Brazil, resulting in million-scale graphs. Our method identified correlations, similarities and dissimilarities among medical treatments, exams, clinics, and outcomes. With the visual tools provided by GraF-EDA, we were able to spot cases of interest and check more details about them. Our results indicate that GraF-EDA is a fast, effective, open-sourced tool for EDA of EHRs from multiple institutions.

Guaranteeing anatomical shape consistency in cardiac magnetic resonance imaging for left ventricle segmentation is a complex task due to its shape-changing during the cardiac cycle, the low contrast and resolution of images, the size change between apical and basal slices, the similarity with nearby organs, and the presence of cardiomyopathies that can deform the heart. Although producing segmentations very close to the ones produced by experts according to standard evaluation metrics, deep learning networks still often produce anatomically inconsistent segmentations. In this work, we propose a new shape-based loss function that favors shape consistency. The loss function uses shape information extracted from distance maps estimated by the network. We validate our approach with the ACDC and Sunnybrook public datasets by using standard metrics as well as a shape similarity metric. The results indicate that the proposed loss is able to improve shape similarity and demonstrate good generalization ability, while presenting competitive performance in the standard evaluation metrics.

Contextual word embeddings and the Transformers architecture have reached state-of-the-art results in many natural language processing (NLP) tasks and improved the adaptation of models for multiple domains. Despite the improvement in the reuse and construction of models, few resources are still developed for the Portuguese language, especially in the health domain. Furthermore, the clinical models available for the language are not representative enough for all medical specialties. This work explores deep contextual embedding models for the Portuguese language to support clinical NLP tasks. We transferred learned information from electronic health records of a Brazilian tertiary hospital specialized in cardiology diseases and pre-trained multiple clinical BERT-based models. We evaluated the performance of these models in named entity recognition experiments, fine-tuning them in two annotated corpora containing clinical narratives. Our pre-trained models outperformed previous multilingual and Portuguese BERT-based models for cardiology and multi-specialty environments, reaching the state-of-the-art for analyzed corpora, with 5.5% F1 score improvement in TempClinBr (all entities) and 1.7% in SemClinBr (Disorder entity) corpora. Hence, we demonstrate that data representativeness and a high volume of training data can improve the results for clinical tasks, aligned with results for other languages.

Physicians work at a very tight schedule and need decision-making support tools to help on improving and doing their work in a timely and dependable manner. Examining piles of sheets with test results and using systems with little visualization support to provide diagnostics is daunting, but that is still the usual way for the physicians' daily procedure, especially in developing countries. Electronic Health Records systems have been designed to keep the patients' history and reduce the time spent analyzing the patient's data. However, better tools to support decision-making are still needed. In this article, we propose ClinicalPath, a visualization tool for users to track a patient's clinical path through a series of tests and data, which can aid in treatments and diagnoses. Our proposal is focused on patient's data analysis, presenting the test results and clinical history longitudinally. Both the visualization design and the system functionality were developed in close collaboration with experts in the medical domain to ensure a right fit of the technical solutions and the real needs of the professionals. We validated the proposed visualization based on case studies and user assessments through tasks based on the physician's daily activities. Our results show that our proposed system improves the physicians' experience in decision-making tasks, made with more confidence and better usage of the physicians' time, allowing them to take other needed care for the patients.

Machine learning (ML) algorithms have been used in many areas of activity, and their results can often be applied without further human intervention. The ML algorithms have also been widely used in medical contexts, but in this area, the result needs to be thoroughly confirmed by a specialist, who needs explanatory information on how the results were obtained. Aimed at such scenarios, we propose the Multilevel Clustering Explainer (MCE), a method capable of providing explanatory information to health professionals about the knowledge discovery process. The MCE was developed for the analysis of medical data, providing a synthesis of explanatory information for the specialist to quickly and clearly understand how the results were obtained.

The current COVID-19 pandemic has promoted the periodic release of several health databases aimed at discovering relationships in the data, detecting similar problems in patients, and studying the evolution of the disease. A way to exploit the data is to use visualization techniques, which can lead to the discovery of insights and patterns, as well as to guide analysis procedures to understand the data. In this paper, we present I-CovidVis, a visualization tool to explore data from interoperable healthcare systems, able to compare and navigate in both global and local perspectives. Our approach is to model data as a graph and explore its structural and temporal views. Our proposal facilitates the perception of patterns, trends, periodicity, and anomalies, resulting in faster decision making.

Several studies have been performed worldwide to improve health services using data generated by digital medical systems. The increasing volume of data generated by these systems is making the use of knowledge discovery and data analysis techniques essential to improve the quality of the health services, which are offered by the medical facilities. However, it is possible to observe a gap, in the literature, about generic and flexible vector space models (VSM) that are well adapted to handle electronic health records (EHR), requiring that each knowledge discovery effort develop their own VSM or other representation model. This restriction can turn a knowledge discovery task over clinical pathways nonviable for comparative evaluations among different methods. Targeting such scenario, we propose the Bag-of-Attributes Representation (BOAR). BOAR represents an EHR as an n-dimensional vector space. Since BOAR takes advantage of the OMOP (Observational Medical Outcomes Partnership) standard, BOAR is able to represent records retrieved from different data models. The experimental results show that BOAR is flexible and robust to representing EHR from several sources, and allows the execution and evaluation of several clustering algorithms.

Availability of low cost Internet connections and specialized hardware, like webcams and headsets, makes it possible to develop solutions for remote collaborative work. These solutions can be advantageous compared to face meetings for several reasons: they allow real-time presence of experts on remote locations without the cost of displacement, discussions can be recorded for educative and legal reasons, it is possible to create online didactic material (e.g. video-classes), technology also permits richer ways of interaction between participants. The Heart Institute of São Paulo with the support of the Brazilian Health Ministry have developed and implemented a second opinion pilot system for emergency cases in cardiology in the metropolitan area of São Paulo on two remote locations. The pilot project will be extended to comprise two hundred points across the country. In this paper we present the results of the project so far.

A computer-aided diagnostic (CAD) method of calculating lumen and wall thickness of carotid vessels is presented. The CAD is able to measure the geometry of the lumen and plaque surfaces in ultrasound carotid images using a least-square fitting of the active contours obtained automatically from the vessels border. To evaluate the approach, ultrasound image sequences from 30 patients were submitted to the procedure. The images were stored on an international data grid repository that consists of three international sites: IPI Laboratory at University of Southern California, USA; Heart Institute at University Sao Paulo, Brazil, and Hong Kong Polytechnic University, Hong Kong. The three chosen sites are connected with high speed international networks including the Internet, and the Brazilian 'National Research and Education Network (RNP2). The Data Grid was used to store, backup, and share the ultrasound images and analysis results, which provided a large-scale and a virtual data system.

Patients usually get medical assistance in several clinics and hospitals during their lifetime, archiving vital information in a dispersed way. Clearly, a proper patient care should take into account that information in order to check for incompatibilities, avoid unnecessary exams, and get relevant clinical history. The Heart Institute (InCor) of São Paulo, Brazil, has been committed to the goal of integrating all exams and clinical information within the institution and other hospitals. Since InCor is one of the six institutes of the University of São Paulo Medical School and each institute has its own information system, exchanging information among the institutes is also a very important aspect that has been considered. In the last few years, a system for transmission, archiving, retrieval, processing, and visualization of medical images integrated with a hospital information system has been successfully created and constitutes the InCor's electronic patient record (EPR). This work describes the experience in the effort to develop a functional and comprehensive EPR, which includes laboratory exams, images (static, dynamic, and three dimensional), clinical reports, documents, and even real-time vital signals. A security policy based on a contextual role-based access control model was implemented to regulate user's access to EPR. Currently, more than 10 TB of digital imaging and communications in medicine (DICOM) images have been stored using the proposed architecture and the EPR stores daily more than 11 GB of integrated data. The proposed storage subsystem allows 6 months of visibility for rapid retrieval and more than two years for automatic retrieval using a jukebox. This paper addresses also a prototype for the integration of distributed and heterogeneous EPR.