A5072021658- Privacy-Preserving Technologies in Data
- Data Quality and Management
- Cryptography and Data Security
- Bayesian Modeling and Causal Inference
- Data Stream Mining Techniques
- Mental Health Research Topics
- Mobile Crowdsensing and Crowdsourcing
- Advanced Bandit Algorithms Research
- Complex Network Analysis Techniques
- Privacy, Security, and Data Protection
- Control Systems and Identification
- Data-Driven Disease Surveillance
- Cardiac Health and Mental Health
- Ethics in Clinical Research
- Artificial Intelligence in Healthcare
- Advanced Adaptive Filtering Techniques
- Human Mobility and Location-Based Analysis
- Social and Cultural Dynamics
- Government, Law, and Information Management
- Vehicular Ad Hoc Networks (VANETs)
- Cultural Differences and Values
- Statistical Methods and Inference
- Advanced Statistical Modeling Techniques
- Mobile Health and mHealth Applications
- Behavioral Health and Interventions
Centraal Bureau voor de Statistiek
2021-2024
Maastricht University
2019-2021
Tilburg University
2013-2016
It is widely anticipated that the use and analysis of health-related big data will enable further understanding improvements in human health wellbeing. Here, we propose an innovative infrastructure, which supports secure privacy-preserving personal from multiple providers with different governance policies. Our objective to this infrastructure explore relation between Type 2 Diabetes Mellitus status healthcare costs. approach involves distributed machine learning analyze vertically...
Privacy-preserving machine learning enables the training of models on decentralized datasets without need to reveal information, both horizontally and vertically partitioned data. However, it requires specialized techniques algorithms perform necessary computations. The privacy preserving scalar product protocol, which dot vectors revealing them, is one popular example for its versatility. For can be used analyses that require counting number samples fulfill certain criteria defined across...
Abstract Federated learning makes it possible to train a machine model on decentralized data. Bayesian networks are widely used probabilistic graphical models. While some research has been published the federated of networks, publications in vertically partitioned data setting limited, with important omissions, such as handling missing We propose novel method called VertiBayes (structure and parameters) data, which can handle values well an arbitrary number parties. For structure we adapted...
The potential of big data in health research is dependent upon the ability to collect, reuse, link, and analyse those datasets. EU General Data Protection Regulation 2016/679 presents with several difficulties uncertainties. In this Article we analyze impact five elements GDPR on research: (1) distinction made between non-special versus special categories data; (2) informed consent its relation secondary processing (3) use national identification number; (4) principle minimization; and, (5)...
Digital health is a promising tool to support people with an elevated risk for atherosclerotic cardiovascular disease (ASCVD) and patients established improve outcomes. Many digital initiatives have been developed employed. However, barriers their large-scale implementation remained. This paper focuses on these presents solutions as proposed by the Dutch CARRIER (ie, Coronary ARtery disease: Risk estimations Interventions prevention EaRly detection) consortium. We will focus in 4 sections...
Abstract. Novel technological advances allow distributed and automatic measurement of human behavior. While these technologies provide exciting new research opportunities, they also challenges: datasets collected using grow increasingly large, in many applications the data are continuously augmented. These streams make standard computation well-known estimators inefficient as has to be repeated each time a point enters. In this tutorial paper, we detail online learning, an analysis method...
Social scientists are often faced with data that have a nested structure: pupils within schools, employees companies, or repeated measurements individuals. Nested typically analyzed using multilevel models. However, when sets extremely large new continuously augment the set, estimating models can be challenging: current algorithms used to fit repeatedly revisit all points and end up consuming much time computer memory. This is especially troublesome predictions needed in real observations...
We reply to the comments on our proposed privacy preserving n-party scalar product protocol made by Liu. In their comment Liu raised concerns regarding security and scalability of $n$-party protocol. this reply, we show that are unfounded is safe for its intended purposes. Their based a misunderstanding Additionally, while puts limitations use, still has numerous practical applications when applied in correct scenarios. Specifically within vertically partitioned scenarios, which often...
In dichotomous item response theory (IRT) framework, the asymptotic standard error (ASE) is most common statistic to evaluate precision of various ability estimators. Easy-to-use ASE formulas are readily available; however, accuracy some these was recently questioned and new were derived from a general framework. Furthermore, exact errors suggested better estimators, especially with short tests for which framework invalid. Unfortunately, assessed so far only in very limiting setting. The...
Combining data from varied sources has considerable potential for knowledge discovery: collaborating parties can mine in an expanded feature space, allowing them to explore a larger range of scientific questions. However, sharing among different is highly restricted by legal conditions, ethical concerns, and / or volume. Fueled these the fields cryptography distributed learning have made great progress towards privacy-preserving mining. practical implementations been hampered limited scope...
Federated learning makes it possible to train a machine model on decentralized data. Bayesian networks are probabilistic graphical models that have been widely used in artificial intelligence applications. Their popularity stems from the fact they can be built by combining existing expert knowledge with data and highly interpretable, which them useful for decision support, e.g. healthcare. While some research has published federated of networks, publications vertically partitioned or...
Privacy-preserving machine learning enables the training of models on decentralized datasets without need to reveal data, both horizontal and vertically partitioned data. However, it relies specialized techniques algorithms perform necessary computations. The privacy preserving scalar product protocol, which dot vectors revealing them, is one popular example for its versatility. Unfortunately, solutions currently proposed in literature focus mainly two-party scenarios, even though scenarios...
Federated learning allows us to run machine algorithms on decentralized data when sharing is not permitted due privacy concerns. Ensemble-based works by training multiple (weak) classifiers whose output aggregated. ensembles are applied a federated setting, where each classifier in the ensemble trained one location. In this article, we explore use of Bayesian networks (FBNE) range experiments and compare their performance with locally models VertiBayes, algorithm train from data. Our results...
It is widely anticipated that the use of health-related big data will enable further understanding and improvements in human health wellbeing. Our current project, funded through Dutch National Research Agenda, aims to explore relationship between development diabetes socio-economic factors such as lifestyle care utilization. The analysis involves combining from Maastricht Study (DMS), a prospective clinical study, collected by Statistics Netherlands (CBS) part its routine operations....