Bringing Bayesian networks to bedside: a web-based framework

Raphael Oliveira1, Joana Ferreira1, Diogo Libânio1, Claudia Camila Dias1, Pedro Pereira Rodrigues1

1CINTESIS – Center for Health Technology and Services Research, Faculty of Medicine of the University of Porto Porto, Portugal

The use of statistical methods and quantitative analysis in the clinical decision-making process, is one of the ways of using the latest scientific evidence. Despite its advantages, the availability of inference software in clinical settings is still limited. Typically, in prognostic or diagnostic models, support systems are based in logistic or linear regression. These techniques have the advantage of being easily interpretable from the clinical point of view, having nonetheless a poor graphical representation. However, the nature of biomedical data requires the application of techniques that go beyond traditional biostatistics, such as Bayesian networks. Bayesian approaches have an extreme importance in clinical problems, since they provide both qualitative and quantitative perspectives. They consider prior knowledge, making data analysis an update processing of prior knowledge with observed evidence. To potentiate the use of the statistical methods within the daily practice, we created simple web forms. These web forms do not require complex interactions, receiving clinical inputs and transmitting them to a Bayesian network inference engine. The information is processed by the engine and the output data is sent to the end-user through the same web form. This approach makes the derived models usable at bedside by both the clinicians and the patients themselves.

keywords: web-based forms; Bayesian models; clinical use

Poster: Bringing Bayesian networks to bedside: a web-based framework

Architecture proposal for a holistic lifelong health record

Duarte Gonçalves-Ferreira1, Pedro Vieira-Marques1, Priscila Maranhão1, Gustavo Marísio Bacelar-Silva1, Ricardo Correia1


Introduction: The lack of communication between different healthcare providers can slow down the process of caring, produce duplicate exams which increase the cost of care. Personal health data, lifestyle, nutritional and genetic data which are currently stored in a variety of repositories, could help improve the level of care by allowing faster diagnosis and a more personalized treatment. These improvements could also reduce the overall cost of care if it is accessible in a timely manner and with a high level of quality. In this article, we analyze the requirements for a lifelong health record and propose an architecture design for a lifelong health record, scalable, federated and distributed, capable of supplying a simple access to data while still answering the security and data access concerns from each of the stakeholders delegating to them the control on how, when and who accesses their data. Aim: To propose an architecture for the development of a holistic lifelong health record capable of storing data from heterogeneous data sources in a distributed environment. Methods: We will analyze the requirements for the storage of clinical, personal, lifestyle, nutritional, and genetic data, storing symbolic, image, video and vital signs; create use cases based on these requirements and design an architecture proposal. We will also take into consideration security, scalability and authorization requirements mandatory for distributed holistic lifelong health record. Current results: We currently have a draft of an architecture design that we are analysing for possible flaws based on the gathered use cases. We are also researching possible solution for the communication layer of the holistic lifelong health record based on openEHR.

keywords: Electronic health records, Federated health record, openEHR

Poster: Architecture proposal for a holistic lifelong health record

Anomaly detection through temporal abstractions on intensive care data

Giovana Gelatti1, Pedro Pereira Rodrigues1,2


A large amount of data is continuously generated in intensive health care. An analysis of these streaming data can provide important information to improve the monitoring of the health conditions of patients. The volume, velocity and complexity of these data, which come unlabeled, make their analysis a challenging task. Machine learning techniques have been successfully used for data stream mining. The application of these techniques to intensive health care stream data can extract useful knowledge for health care monitoring, which includes the detection of changes in the behaviour of sensors, failures on machines or systems, and data anomalies, which might represent abnormal activities. Anomaly (or outlier) detection aims to nd exceptions or abnormalities in a dataset. Data exception, in medical context, can represent new disease pattern, an event to be further investigated, behaviour changes or possible health complications. Nonetheless, its analysis in data streams is a challenging task. Machine learning-based temporal abstractions deal with the management and abstraction of time based data, providing a high level of visualization of each data object in its context. Their use for anomaly detection may provide the medical expert a focus on relevant data and warnings. Other applications of machine learning techniques to data streams can support medical diagnosis, giving new, useful and relevant knowledge about the patients being monitored. The purpose of this paper is to review recent research in anomaly detection and temporal abstraction and how they have been applied to intensive care data streams.

keywords: Anomaly detection, outlier detection, temporal abstraction, intensive care

Poster: Anomaly detection through temporal abstractions on intensive care data

SoTRAACE – Socio-Technical Risk-Adaptable Access Control Model

Pedro Moura1,2, Ana Ferreira1, Paulo Fazendeiro2

1Cintesis, 2University of Beira Interior

Smartphones are the most ubiquitous device that people hold nowadays.
In the healthcare domain, professionals can use smartphones to access Health Information Systems (HIS), access Electronic Health Records (EHR) to set and view exam results, share data and prescribe medications . Patients can use smartphones to access their medical records, control access to their health related data, monitor health statistics and view their prescriptions .

However, HIS and EHR’s comprise highly sensitive data, which raises serious concerns regarding patients’ privacy and safety , and are therefore subjected to legal and regulatory restrictions .
Access control aims to provide adequate means to protect health data from unauthorized accesses . With the new mobile paradigm of anytime/everywhere, there is a need to find more innovative, flexible, dynamic, transparent and resilient access control models, that are adaptable to more heterogeneous requests, as traditional solutions are based on predefined access policies and roles.

This work presents a new access control model, SoTRAACE – Socio-Technical Risk-Adaptable Access Control Model, that takes the inherent differences and security requirements present in each access situation and aggregates attributes to help performing a risk assessment at the moment of request. Attributes include: context/location, type of device, user profiling and access history, institution or legal requirements , type and sensitivity level of the resource, unanticipated situations and performed delegations. SoTRAACE is expected to provide a more dynamic, adaptable, transparent and secure access to health data by mobile users.

keywords: Data privacy, Health information systems (HIS), Mobile access control, Risk adaptable access, Socio-technical systems, Ubiquitous access.

Presentation: SoTRAACE Socio-Technical Risk-Adaptable Access Control Model

Chromosomal anomalies hospitalizations: a 15 year nation-wide analysis

Manuel Gonçalves-Pinho1, Alberto Freitas1

1Center for Health Technology and Services Research (CINTESIS)

Aim: Chromosomal anomalies (CA) hospitalizations represent an overall health and prognosis indicator. We aim to describe CA related hospitalizations burden, characteristics and specific trends as possible indicators of genetic diseases epidemiology.

Methods: A retrospective observational study was performed using a national hospitalization database between 2000 and 2014. CA were selected based on codes 758.0x to 758.7x codified by the International Classification of Diseases – 9th Revision – Clinical Modification. Birth date, gender, charges, admission/discharge date, discharge status, primary/secondary diagnoses were analyzed.

Results: CA related hospitalizations accounted for 0.08% of all the hospitalizations. Down’s syndrome represented 75.9% of all CA-related hospitalizations and 80.2% of all the charges attributed to CA related hospitalizations. The median age of CA-related patients was 9.0 years old. The leading causes of hospitalization in different CA varied between pneumonia (3.6-18.6%) and livebirth related diagnoses (7.9-52.5%). Mean number of hospitalizations ranged from 1.0 to 2.1 per patient and mean charges per hospitalization varied from 2,339 to 4,520 €.

Discussion: CA represent a substantial burden on the health system with high mean charges per hospitalization, high length of stay and high in-hospital mortality. Down’s syndrome accounts for the majority of CA hospitalizations. Klinefelter’s syndrome hospitalizations occur at a younger age than the described mean age of diagnoses in all Klinefelter’s syndrome patients.

keywords: Chromosome Aberrations, Chromosome Disorders, Genetic Epidemiology, Hospitalization, Cost Analysis, Observational Study

Presentation: Chromosomal anomalies hospitalizations: a 15 year nation-wide analysis


Priscila Alves Maranhão1, Gustavo Marísio Bacelar-Silva1, Duarte Nuno Gonçalves-Ferreira1, Pedro Vieira-Marques1, Ricardo João Cruz-Correia1

1Center for Research in Health Technologies and Information Systems (CINTESIS)

Introduction: Noncommunicable diseases (NCDs), also known as chronic diseases, are not transmission disease. The five main types of NCDs are cardiovascular diseases, cancers, chronic respiratory diseases, diabetes and obesity. Have been describing that NCDs kill 38 million people each year and cardiovascular diseases is responsible for most NCD deaths, or 17.5 million people annually. Researchers have been showing that sedentary behavior, unhealthy diet, alcohol and tobacco use are risk factors to NCDs and nutrition care can curb the high prevalence of NCDs. However electronic health record (EHR) is an important tool to clinicians to improve treatment, prevention and update the health care.

Aim: To propose a set of archetypes for the integration of Clinical Practice Guideline (CPG) recommendations for treatment and prevention of chronic non-communicable diseases in an openEHR based EHR.

Methods: We will assess CPG in five NCDs guidelines:  cancer,  diabetes; cardiovascular disease, respiratory disease and obesity in order to create archetypes in Clinical Knowledge Manager (CKM) repository, and, finally, compose a template in openEHR.

Results: All variables were identified in guidelines. However, only 31 archetypes were necessary to represent all of them. Of these 31 archetypes, 8 are already available in CKM, while 23 clinical statements had no archetype representation and they had to be created.

Conclusion: Based on the created archetypes, an openEHR template can be created for the representation of clinical statements regarding the most important clinical guidelines in NCDs.

keywords: Electronic health records, Nutrition, Clinical practice guideline, openEHR, Chronic non-communicable diseases

Poster: NUTRITION INFORMATION IN SPECIFIC CONDITIONS Modeling Archetypes in openEHR Specification

Reputation Based Method for Improving Security and Fairness in Distributed Networks.

Francis N. Nwebonyi1,2, Rolando Martins1,2, Manuel Eduardo Correia1,2

1Faculty of Science, University of Porto, 2CRACS/INESC-TEC

Aim: this work aims at deriving a fair, inexpensive, and highly robust reputation based security system, for peers in distributed networks. Its importance follows the emergence and popularity of distributed platforms such as P2P, IOT, etc, used to share information. Sensitive information can be shared through these
platforms, making security and privacy very essential.
Methods: we began with a detailed study of the literature, revealing their strengths and weaknesses. Mathematical methods were derived to capture vital trust concepts, and then algorithms were designed based on the derived methods. BitTorrent which is an example of P2P, was used as a test bed. Peersim simulator
was used to implement the algorithm, the simulator already has a version of BitTorrent implementation, with which we compared our method.
Results: we tested the new method, comparing it with the tit-for-tat based BitTorrent, in the presence of the following maliciousness and attacks; free-rider, bandwidth attack, badchunk attack, and sybil attack. The result shows that the new method is clearly more effective. Non-malicious peers were able to download faster, and resources, such as bandwidth, was seen to be utilized more efficiently.
Discussion: the new bootstrapping method discourages whitewashing, and minimizes sybil attack. Our algorithm also defends seeders and leechers from attacks, and encourages fairness, which results in enhanced efficiency.
Conclusion: we have developed a novel distributed reputation bootstrapping method for P2P and similar networks. The new method also incorporates earlier missing concepts such as personal trust, and it effectively defends peers from attacks.

keywords: Reputation, P2P, Security, Fairness

Presentation: Reputation Based Method for Improving Security and Fairness in Distributed Networks

Secure Multiparty Computation from SGX

Bernardo Portela1,2, Manuel Barbosa1,2, Guillaume Scerri3,4, Bogdan Warinschi5,  Raad Bahmani6, Ferdinand Brasser6, Ahmad-Reza Sadeghi6

1HASLab – INESC TEC, 2DCC-FCUP, 3Laboratoire DAVID – Université de Versailles St-Quentin, 4INRIA, 5University of Bristol, 6Technische Universität Darmstadt

In this paper we show how Isolated Execution Environments (IEE) offered by novel commodity hardware such as Intel’s SGX provide a new path to constructing general secure multiparty computation (MPC) protocols. Our protocol is intuitive and elegant: it uses code within an IEE to play the role of a trusted third party (TTP), and the attestation guarantees of SGX to bootstrap secure communications between participants and the TTP. The load of communications and computations on participants only depends on the size of each party’s inputs and
outputs and is thus small and independent from the intricacies of the functionality to be computed. The remaining computational load– essentially that of computing the functionality – is moved to an untrusted party running an IEE-enabled machine, an attractive feature for Cloud-based scenarios.

Our rigorous modular security analysis relies on the novel notion of labeled attested computation which we put forth in this paper. This notion is a convenient abstraction of the kind of attestation guarantees one can obtain from trusted hardware in multi-user scenarios.

Finally, we present an extensive experimental evaluation of our solution on SGX enabled hardware. Our implementation is open-source and it is functionality agnostic: it can be used to securely outsource to the Cloud arbitrary off-the-shelf collaborative software, such as the one employed on financial data applications, enabling secure collaborative execution over private inputs provided by multiple parties.

keywords: Secure Computation, Provable Security, Trusted Hardware, Cryptography

Presentation: Secure Multiparty Computation from SGX

Health Kiosk

João Silva1, Pedro Brandão1, Rui Prior1

1Instituto de Telecomunicações, Faculdade de Ciências da Universidade do Porto

Aim: The Health Kiosk aims to be a self-assessment health tool to be used autonomously by its users. It is modular, easily configurable, adaptable to the circumstances.
It uses web technologies to provide a fully functional system composed of different components. Features such as text-to-speech instructions, results in a QR Code, printed out results, are integrated into the system. Moreover, the possibility of creating different combinations of medical devices can easily be done due to the developments made.

Method: To analyze the usability of the system we tested the system on a university setting with voluntary participation. The participants were students, teachers and faculty staff. We used questionnaires and usage observation to determine population characteristics and detect usability issues.

Results: Some provided instructions can be simplified to provide a better understanding of what is necessary to correctly manage the medical devices. The experiment showed that some minor modifications are needed. Nonetheless, all participants successfully used the system from start to finish.

Discussion: The results obtained show us that some provided instructions, can be simplified. Despite that fact that system has proved to be functional and prepared to handle different usage types.

Conclusion: We can conclude that this system will have an impact on locations with a difficult access to medical information. It can help to reduce the gap existing between rural and urban areas in terms of medical care, and can also be used with other intentions such as a pre-consultation tool, or in elderly communities.

keywords: Health Kiosk, Medical Devices, Data Collection, Autonomous Measurements, Health Monitoring

Presentation: Health Kiosk

Poster: Health Kiosk

Modelling the suppression of autoimmunity after pathogen infection

B.M.P.M. Oliveira1, R. Trinchet A.2, A. A. Pinto3, N. J. Burroughs4

1Faculdade de Ciências da Nutrição e Alimentação da Universidade do Porto, Portugal; LIAAD, INESC-TEC, Porto, Portugal, 2LIAAD, INESC-TEC, Porto, Portugal; Universidade de Santiago de Compostela, Spain, 3LIAAD, INESC-TEC, Porto, Portugal; Departamento de Matemática da Faculdade de Ciências da Universidade do Porto, Portugal, 4Systems Biology Centre, University of Warwick, UK

We study a mathematical model of immune response by T cells where the regulatory T cells (Treg) inhibit interleukine 2 (IL-2) secretion. We model the suppression of the autoimmune line of T cells after a different line of T cells was stimulated by a pathogen infection. In this paper we show that if the population of the pathogen responding line of T cells becomes large enough, the competition for IL-2 and the increase in the death rates may lead to a depletion in the concentration of autoimmune T cells. Provided this lasts for a suffciently long time period, the concentration of autoimmune T cells can be brought down to values inside the basin of attraction of the controlled state and autoimmunity can be suppressed.

keywords: Autoimmunity suppression, T cells, Tregs, cytokines, ODE model

Poster: Modelling the suppression of autoimmunity after pathogen infection