RESUMO
Digitization in healthcare systems, with the wid adoption of Electronic Health Records, connected medical devices, software and systems providing efficient healthcare service delivery and management. On the other hand, the use of these systems has significantly increased cyber threats in the healthcare sector. Vulnerabilities in the existing and legacy systems are one of the key causes for the threats and related risks. Understanding and addressing the threats from the connected medical devices and other parts of the ICT health infrastructure are of paramount importance for ensuring security within the overall healthcare ecosystem. Threat and vulnerability analysis provides an effective way to lower the impact of risks relating to the existing vulnerabilities. However, this is a challenging task due to the availability of massive data which makes it difficult to identify potential patterns of security issues. This paper contributes towards an effective threats and vulnerabilities analysis by adopting Machine Learning models, such as the BERT neural language model and XGBoost, to extract updated information from the Natural Language documents largely available on the web, evaluating at the same time the level of the identified threats and vulnerabilities that can impact on the healthcare system, providing the required information for the most appropriate management of the risk. Experiments were performed based on CS news extracted from the Hacker News website and on Common Vulnerabilities and Exposures (CVE) vulnerability reports. The results demonstrate the effectiveness of the proposed approach, which provides a realistic manner to assess the threats and vulnerabilities from Natural Language texts, allowing adopting it in real-world Healthcare ecosystems.
Assuntos
Segurança Computacional , Ecossistema , Atenção à Saúde , Registros Eletrônicos de Saúde , Aprendizado de MáquinaRESUMO
Intensive care unit (ICU) patients with venous thromboembolism (VTE) and/or cancer suffer from high mortality rates. Mortality prediction in the ICU has been a major medical challenge for which several scoring systems exist but lack in specificity. This study focuses on two target groups, namely patients with thrombosis or cancer. The main goal is to develop and validate interpretable machine learning (ML) models to predict early and late mortality, while exploiting all available data stored in the medical record. To this end, retrospective data from two freely accessible databases, MIMIC-III and eICU, were used. Well-established ML algorithms were implemented utilizing automated and purposely built ML frameworks for addressing class imbalance. Prediction of early mortality showed excellent performance in both disease categories, in terms of the area under the receiver operating characteristic curve (AUC-ROC): VTE-MIMIC-III 0.93, eICU 0.87, cancer-MIMIC-III 0.94. On the other hand, late mortality prediction showed lower performance, i.e., AUC-ROC: VTE 0.82, cancer 0.74-0.88. The predictive model of early mortality developed from 1651 VTE patients (MIMIC-III) ended up with a signature of 35 features and was externally validated in 2659 patients from the eICU dataset. Our model outperformed traditional scoring systems in predicting early as well as late mortality. Novel biomarkers, such as red cell distribution width, were identified.