A modular approach to the integration of signal processing algorithms.
Abstract
This research proposes a modular approach for designing and integrating complex digital signal processing (DSP) systems that combine multiple functionalities such as noise reduction, dereverberation, dynamic range compression, feedback cancellation, and active noise control. Traditional DSP development often treats these functionalities in isolation, leading to challenges when they interact within integrated systems. The project addresses this limitation by applying Normalized Systems Theory (NST) to create a scalable and evolvable software architecture that enables flexible aggregation of DSP modules. A modular software infrastructure will be developed to support the simulation, configuration, and evaluation of different combinations and sequences of DSP algorithms. The framework will initially focus on hearing instruments, where interactions between DSP functionalities significantly influence system performance. Artificial intelligence will be employed to analyze the large design space created by different module variants and processing sequences, enabling the prediction and optimization of system behavior. The research will also investigate the use of AI-generated software "micro doses" to automatically create glue code that connects DSP modules implemented in different technologies and programming environments. Four work packages structure the research: development of the NST-based platform, AI-driven optimization of module aggregations, AI-assisted generation of integration code, and validation across multiple DSP applications. Beyond hearing instruments, the framework will be evaluated for broader applications.Researcher(s)
- Promoter: Verelst Jan
Research team(s)
Funding
- VL. INST.
Project type(s)
- Research Project
iCAREdata, Improving Care and Research Electronic Data Trust Antwerp.
Abstract
iCAREdata is a state-of-the-art research database processing routine clinical data of out-of-hours (OOH) patient contacts at general practice cooperatives (GPCs) (huisartsenwachtposten), pharmacies and emergency departments (EDs). The database contains data starting from 2015 until to date. Dataflows are secured by encryption of the social security numbers of the patients by eHealth as a trusted third party (TTP), as well as the RIZIV numbers of the physicians. At no point in the flow, personal data can be matched to clinical data. In de final database, pseudonymised personal data enable the linkage of different patient contacts to the same person, offering the opportunity to describe patient trajectories in the OOH care services; e. g. Person X consults a general practitioner at the GPC on Saturday morning, goes to the pharmacie to collect the prescribed medicines in the afternoon of the same day and finally shows up during the night on Sunday morning at 2am at the emergency department. To realise this data flow and linkage, thorough knowledge about the use and implementation of eHealth services is crucial, besides an up-to-date experience in standardisation and classification systems currently used in medical research datasets. A large network of health care services and health care workers is being part of and has to be fully involved with this project. This is only realised by implementing full transparency about the content of the dataset, how the flow is set up, and for what kind of research the data are used and by who. Researchers as well as data-supplying stakeholders stress the importance of upscaling the datasets by expanding with all GPCs in Flanders (at this moment 38 out of 44 GPCs deliver data on a daily basis) as well as adding extra EDs and pharmacies. Researchers suggest to add routine data of daytime general practices. This is necessary in the long term to be able to follow the entire flow of patients. Also the link with patient contacts in telephone- or selftriage, callcentres for urgent medical care (HC112), and follow-up contacts in general practice or hospitals complete the patient trajectories at different sites. iCAREdata has strong links with newly developed tools in referral to and cooperation with welfare services. This is a unique link between medical care and wellbeing which is in scope to be developed during the upcoming years. Besides research opportunities, iCAREdata also offers the opportunity to add up to a data driven policy on the micro, meso and macro level of health care and welfare services organisation, which was a crucial issue during Covid-19 pandemic. In addition several scientific research topics make the database worthwhile to maintain: evaluation of drug prescriptions by physicians in several contexts (antibiotics, opioids, …), use of health care services, epidemiologic topics like surveillance of infectious diseases, gaps in the offer of health and welfare services, waiting lists…iCAREdata has an experienced team of computer scientists developing modular software based on Normalized Systems Theory (NST). These modular software structures enable the introduction of additional data providers in a fast and reliable way. Thanks to financial resources, iCAREdata will further develop, validate and finalize the link with the emergency services and pharmacies. This enables us to set up valuable use cases to acquire users and (paying) customers. The funding will also be used to further develop the public dashboard and the user dashboards.Researcher(s)
- Promoter: De Loof Hans
- Promoter: Philips Hilde
- Co-promoter: De Loof Hans
- Co-promoter: Mannaert Herwig
- Co-promoter: Philips Hilde
- Co-promoter: Verelst Jan
- Co-promoter: Verhoeven Veronique
Research team(s)
Funding
- IOF
Project website
Project type(s)
- Research Project
Modular and evolvable document management systems based on normalized systems theory: case study based research and pilot applications.
Abstract
This project centers around developing an alternative to the established but outdated paradigm of document management. In recent research, an innovative modular approach to document management was established that enables the design of content-agnostic and evolvable documents. This new approach imagines documents as multidimensional and ever-changing entities instead of mere static representation of their analogue predecessors. The approach is based on theoretical insights of modularity and Normalized Systems (NS) theory. Based on case study research, this project establishes a clear definition of the market for evolvable documents. Furthermore, pilot applications are developed in collaboration with organizations to determine the obstacles, requirements, advantages and economic incentives for evolvable document solutions. Based on these outcomes, commercial solutions will be built in follow-up project(s).Researcher(s)
- Promoter: Mannaert Herwig
- Co-promoter: Verelst Jan
Research team(s)
Funding
- IOF
Project type(s)
- Research Project
Abstract
Researcher(s)
- Promoter: Verelst Jan
Research team(s)
Funding
- WERKING
Project type(s)
- Research Project