Abstract

Abstract
Missing data is a problem often found in real-world datasets and it can degrade the performance of most machine learning models. Several deep learning techniques have been used to address this issue, and one of them is the Autoencoder and its Denoising and Variational variants. These models are able to learn a representation of the data with missing values and generate plausible new ones to replace them. This study surveys the use of Autoencoders for the imputation of tabular data and considers 26 works published between 2014 and 2020. The analysis is mainly focused on discussing patterns and recommendations for the architecture, hyperparameters and training settings of the network, while providing a detailed discussion of the results obtained by Autoencoders when compared to other state-of-the-art methods, and of the data contexts where they have been applied. The conclusions include a set of recommendations for the technical settings of the network, and show that Denoising Autoencoders outperform their competitors, particularly the often used statistical methods.

Abstract
The preoperative localisation of non-palpable lesions guided by breast imaging is an important and required procedure for breast-conserving surgery. We conducted a systematic review and meta-analysis of the literature on the comparative impact of different techniques for guided surgical excision of non-palpable breast lesions from reports of clinical or patient-reported outcomes and costs. A literature search of PubMed, ISI, SCOPUS and Cochrane databases was conducted for relevant publications and their references, along with public documents, national and international guidelines, conference proceedings and presentations. From 5720 retrieved articles screened through title and abstract, 5346 were excluded and 374 assessed for full-text eligibility. For data extraction and quality assessment, 49 studies were included. Results of this review demonstrate that Radioactive Seed Localisation (RSL) and Radioactive Occult Lesion Localisation (ROLL) outperform Wire in terms of involved margins and reoperations. Between RSL and ROLL, there is a tendency to favour RSL. Similarly, Clip-guided localisation seems preferred when compared to ROLL, however further studies are needed. In summary, there seems to exist evidence that RSL and ROLL are better than Wire, representing potential alternatives, with a quick learning curve, better scheduling and management issues. Although, for recent techniques, more research is needed in order to achieve the same level of evidence.

Abstract
Current tools for self-management of chronic obstructive respiratory diseases (CORD) are difficult to use, not individualized and requiring laborious analysis by health professionals, discouraging their use in healthcare. There is an opportunity for cost-effective and easy-to-disseminate advanced technological solutions directed to patients and attractive to different stakeholders. The strategy of AIRDOC is to develop and integrate self-monitoring and self-managing tools, making use of the smartphone's presence in everyday life. AIRDOC intends to innovate on: i) technologies for remote monitoring of respiratory function and computerized lung auscultation; ii) coaching solutions, integrating psychoeducation, gamification and disease management support systems; and iii) management of personal health data, focusing on security, privacy and interoperability. It is expected that AIRDOC results will contribute for the innovation in CORD healthcare, with increased patient involvement and empowerment while providing quality prospective information for better clinical decisions, allowing more efficient and sustainable healthcare delivery.

Abstract
AbstractBackgroundHigh-quality data is crucial for guiding decision making and practicing evidence-based healthcare, especially if previous knowledge is lacking. Nevertheless, data quality frailties have been exposed worldwide during the current COVID-19 pandemic. Focusing on a major Portuguese surveillance dataset, our study aims to assess data quality issues and suggest possible solutions.MethodsOn April 27th 2020, the Portuguese Directorate-General of Health (DGS) made available a dataset (DGSApril) for researchers, upon request. On August 4th, an updated dataset (DGSAugust) was also obtained. The quality of data was assessed through analysis of data completeness and consistency between both datasets.ResultsDGSAugust has not followed the data format and variables as DGSApril and a significant number of missing data and inconsistencies were found (e.g. 4,075 cases from the DGSApril were apparently not included in DGSAugust). Several variables also showed a low degree of completeness and/or changed their values from one dataset to another (e.g. the variable ‘underlying conditions’ had more than half of cases showing different information between datasets). There were also significant inconsistencies between the number of cases and deaths due to COVID-19 shown in DGSAugust and by the DGS reports publicly provided daily.ConclusionsThe low quality of COVID-19 surveillance datasets limits its usability to inform good decisions and perform useful research. Major improvements in surveillance datasets are therefore urgently needed - e.g. simplification of data entry processes, constant monitoring of data, and increased training and awareness of health care providers - as low data quality may lead to a deficient pandemic control.

Abstract
INTRODUCTION The COVID-19 pandemic is an ongoing event disrupting lives, health systems, and economies worldwide. Clear data about the pandemic's impact is lacking, namely regarding mortality. This work aims to study the impact of COVID-19 through the analysis of all-cause mortality data made available by different European countries, and to critique their mortality surveillance data. METHODS European countries that had publicly available data about the number of deaths per day/week were selected (England and Wales, France, Italy, Netherlands and Portugal). Two different methods were selected to estimate the excess mortality due to COVID19: (DEV) deviation from the expected value from homologue periods, and (RSTS) remainder after seasonal time series decomposition. We estimate total, age- and gender-specific excess mortality. Furthermore, we compare different policy responses to COVID-19. RESULTS Excess mortality was found in all 5 countries, ranging from 10.6% in Portugal (DEV) to 98.5% in Italy (DEV). Furthermore, excess mortality is higher than COVID-attributed deaths in all 5 countries. DISCUSSION The impact of COVID-19 on mortality appears to be larger than officially attributed deaths, in varying degrees in different countries. Comparisons between countries would be useful, but large disparities in mortality surveillance data could not be overcome. Unreliable data, and even a lack of cause-specific mortality data undermine the understanding of the impact of policy choices on both direct and indirect deaths during COVID-19. European countries should invest more on mortality surveillance systems to improve the publicly available data.