Hugo Peixoto

Intelligence in interoperability with AIDA

Healthcare systems have to be addressed in terms of a wide variety of heterogeneous, distributed and ubiquitous systems speaking different languages, integrating medical equipments and customized by different entities, which in turn were set by different people aiming at different goals. Demands of information within the healthcare sector range from clinically valuable patient-specific information to a variety of aggregation levels for follow-up and statistical and/or quantifiable reporting. The main goal is to gathering this information and present it in a readable way to physicians. In this work we show how to achieve interoperability in healthcare institutions using AIDA, an interoperability platform developed by researchers from the University of Minho and being used in some major Portuguese hospitals.

Modelling Intelligent Behaviours in Multi-agent Based HL7 Services

With the dissemination of Health Information Systems and the greater relevance of interoperability towards the quality of the information available to the clinical personnel, distinct architectures and methodologies have been devised in order to improve the existing platforms in the healthcare environment. However, most of them are based on HL7, an international standard for healthcare interoperability, which depending on the implementation as any technology has its advantages and limitations. This paper details the architecture and methodologies of a multi-agent based HL7 interoperation service. The mentioned system is incorporated in an integration platform, which is implemented in several healthcare institutions and uses Multi-Agent Systems to control and enable the flow of data and information within them. The log registry and extracted statistics of several years of interoperation in one institution are used to analyse the development of prediction models to imbue intelligent behaviour in the existing platform. The resulting models are studied and embedded into a validation HL7 server agent.

Semantic Interoperability and Health Records

Systems Interoperability and Electronic Health Records are responsible for an exponential number of visits in electronic repository, either in terms of medical professionals or related staff. This is paramount for a better and sustainable quality-of-care in clinical assistance and of great potential to medical research. Following these lines of though, we present an agency for the diffusion, integration and archiving of medical information, and show how semantic web can enforce the use of electronic documents in order to envisage free-paper hospitals.

Understanding Stroke in Dialysis and Chronic Kidney Disease

Abstract Patients with severe kidney failure need to be carefully monitored. One of the many treatments is called Continuous Ambulatory Peritoneal Dialysis (CAPD). This kind of treatment intends to maintain the blood tests as normal as possible. Data Mining and Machine Learning can take a simple and meaningless blood’s test data set and build it into a Decision Support System. Through this article, Machine Learning algorithms will be explored with different Data Mining Models in order to extract knowledge and classify a patient with a stroke risk or not, according to their blood analysis.

Computer-Aided Diagnosis in Brain Computed Tomography Screening

Currently, interpretation of medical images is almost exclusively made by specialized physicians. Although, the next decades will most certainly be of change and computer-aided diagnosis systems will play an important role in the reading process. Assisted interpretation of medical images has become one of the major research subjects in medical imaging and diagnostic radiology. From a methodological point of view, the main attraction for the resolution of this kind of problem arises from the combination of the image reading made by the radiologists, with the results obtained from using Artificial Intelligence based applications that will contribute to the reduction and eventually the elimination of perception errors. This article describes how machine learning algorithms can help distinguish normal readings in brain Computed Tomography from all its variations. The goal is to have a system that is able to detect normal appearing structures, thus identifying normal studies, making the reading by the radiologist unnecessary for a large proportion of the brain Computed Tomography scans.

Iron Value Classification in Patients Undergoing Continuous Ambulatory Peritoneal Dialysis using Data Mining.

This work has been supported by Compete: POCI -01-0145-FEDER-007043 and FCT within the Project Scope UID/CEC/00319/2013. This work is also supported by the Deus ex Machina (DEM): Symbiotic technology for societal efficiency gains - NORTE-01-0145-FEDER-000026.

A Hospital Service Kiosk in the Patient’s Pocket

Nowadays, with the use of upcoming technologies such as augmented reality, organizations can improve their services more intelligently. Customer service continues to be an important factor for improving quality in healthcare units and, for this reason they are providing self-service systems, such as kiosks, as an option. Therefore, and having also in consideration the current use of mobile devices in day-to-day life, an augmented reality solution that takes advantage of mobile technology was advised. The goal is to optimise the process of customer service in healthcare units. The main element of this solution is a mobile application that provides an augmented reality experience to the user due to its self-awareness about his location and about the tasks that the he is performing. This application complements the current kiosks system in which it is integrated, providing most of its functionalities.

Improving Maternity Care with Business Intelligence

The aim of this paper is to develop clinical indicators for obstetrics through the use of Business Intelligence (BI) tools, since valid and reliable clinical indicators can help measuring quality of healthcare services and support decision-making processes. This paper gives an overview of concepts related to Health Information Systems (HIS) and BI, along with some related work to highlight the advantages that BI solutions can bring when applied to healthcare. In this paper is also presented the data warehousing and the ETL process, that was necessary for the development of indicators and which is usually hidden from endusers, is described. The indicators were developed using Power BI and were analysed and compared with reference values from both national and international health reports. The discussion of the developed indicators made it possible to measure the quality of the obstetrics service, to identify the problematic areas and to decide whether improvement measures should be taken.

Step towards prediction of perineal tear

Abstract The aim of this study is to predict, through data mining tools, the incidence of perineal tear. This kind of laceration developed during child delivery might imply surgery and entails a set of several consequences. Clinical Decision Support Systems, with the information collected from patients’ electronic health records combined with the data mining techniques, may decrease the incidence of perineal tears during labour.

Predicting the need of Neonatal Resuscitation using Data Mining

Abstract It is estimated that approximately 10% of newborns require some kind of assistance for breathing at birth. Aiming to prevent neonatal mortality, the goal behind this paper is to predict the need for neonatal resuscitation given some health conditions of both the newborn and the mother, and also the characteristics of the pregnancy and the delivery using Data Mining (DM) models induced with classification techniques. During the DM process, the CRISP-DM Methodology was followed and the WEKA software tool was used to induce the DM models. For some models, it was possible to achieve sensitivity results higher than 90% and specificity and accuracy results superior to 98%, which were considered to be satisfactory.