News

Nathan Vandemoortele and Bram Steenwinckel have been awarded for their best performing solution in the Table Metadata to Knowledge Graph competition within Semantic Web Challenge on Tabular Data to Knowledge Graph Matching challenge at the the 23st International Semantic Web Conference held in Baltimore, USA.

Within this competition, teams were asked to match table metadata, e.g., column names, to knowledge graphs without any access to table data and content. This is a challenging task due to the limited available context that could be used by annotation systems to perform the semantic linking.

Nathan and Bram proposed a solution that employs a Retrieval-Augmented Generation (RAG) framework to conduct an initial broad search for potential matches, ensuring scalability. Matches were further refined using a Large Language Model (LLM) incorporating advanced prompting techniques such as Chain-of-Thought (CoT) and Self-Consistency (SC). The results are combined using Reciprocal Rank Fusion (RRF) to generate a final ranked list of matches More information can be found in their paper

Despite challenges like computational cost and reliance on high-quality metadata, this solution demonstrated significant improvements over traditional methods and offers a scalable alternative to human annotation.

Our PreDiCT-DLab researchers recently achieved outstanding success at the HASCA workshop, part of the UbiComp/ISWC Conference, held in Melbourne, Australia. Jonas and Jeroen Van Der Donckt, took first place in both the WEAR and SHL time series machine learning challenges - outperforming the competition by an impressive 10% higher (absolute!) macro F1 score.

Our solution, built on our in-house tsflex toolkit for feature extraction and insights gathered from visualization using our plotly-resampler library, proved highly effective. In particular we relied on a traditional machine learning pipeline consisting of processing, feature extraction, and a gradient boosted trees model.

For the SHL challenge, we proposed a novel feature aggregation technique, with the aim of constructing more “rotation-invariant” features. Whereas, for the WEAR challenge, we investigated a data augmentation scheme by swapping left-right wearables, to train a more robust model.

For those interested in diving deeper, the research papers are available here:

• HASCA paper
• WEAR paper

Explore the tools we used:

• tsflex
• plotly-resampler

Stay tuned as our team continues to lead the way in time series and machine learning!

We are delighted to announce that Jef Jonkers has been awarded the Best Student Paper Award at the 13th Symposium on Conformal and Probabilistic Prediction with Applications (COPA2024) held in Milan, Italy.

Jef’s and by extension our teams paper, titled “Conformal Predictive Systems Under Covariate Shift,” impressed the conference committee with its innovative approach and significant contributions to the field. You can read the preprint of this groundbreaking work on arXiv.

In true Italian style, the award presented to Jef was a “La Cupola” coffee maker by Alessi. This elegant piece was designed by Aldo Rossi, a distinguished alumnus of Politecnico di Milano, adding a touch of local flavor to the recognition.

This year’s award holds special significance as it is dedicated to the memory of Professor Aleksej Červonenkis, honoring his invaluable contributions to the field of machine learning and statistical learning theory.

We extend our warmest congratulations to Jef Jonkers on this outstanding achievement!

Jarne Verhaeghe behaalde de Best Presentation Award op de internationale conferentie ICMHI, in Yokohama Japan, voor zijn onderzoek omtrent het optimaliseren van antibioticatherapie met behulp van kennis-gedreven machine learning. Dit onderzoek vloeit uit het Heroi2c project samen met het UZGent. Proficiat!

On April 2nd, 2024, the Make-Marathon, organized by Comon, took place. During this intense four-day event, students from diverse academic backgrounds came together to devise innovative ways to promote physical activity in our daily city lives. The participants were split into different teams of 3-5 people and spent the 4 days brainstorming for a feasible, desirable and viable solution and designing a high level prototype. Among the participants was Stephanie Chen, a doctoral researcher from our very own Predict team. Stephanie, alongside her teammates Eileen (doctoral researcher applied economics), Camille (nutrition and dietetics) and Suëda (student oral care) tackle the sedentary lifestyle epidemic amongst workers and ultimately won the jury prize with their Walk & Meet mobile app and plugin.

In de race om tegen 2050 aan de strenge normen voor energie-efficiëntie te voldoen, staan we voor de immense taak om maar liefst 80% van de bestaande gebouwen grondig te renoveren. Het hergebruik en de renovatie van deze gebouwen brengen onvermijdelijk aanpassingen, verbeteringen en upgrades met zich mee. In een tijd waarin de bouwsector geconfronteerd wordt met voortdurende veranderingen in marktvereisten, regelgevingen en de verwachtingen van eindgebruikers, is het duidelijk dat we voor een grote uitdaging staan. De transformatie van oude gebouwen tot slimme gebouwen die voldoen aan de dynamische verwachtingen en omstandigheden, lijkt de ideale volgende stap.

Maar hoe kunnen retrofit-oplossingen ons in deze stap helpen? Wat is de cruciale rol van “full life cycle service” en “predictive maintenance” in dit proces? Deze vragen en enkele use cases zullen door onze experts behandeld worden.

Eén van de drie gastsprekers is Sofie van Team PreDiCT die een talk geeft over het potentieel van AI/ML voor gebouwen en hun technieken.

In dit themanummer ontrafelen we hoe de reis van dromen naar maken eruitziet, waarbij we afstappen van het idee van innovatie als het werk van eenzame genieën zoals Edison of Tesla. De special is een viering van de menselijke creativiteit en technologische vooruitgang, en nodigt iedereen uit om deel te nemen aan het innovatieproces.

Team PreDICT draagde bij tot deze special issue.

Sofie Van Hoecke van team PreDiCT doet op een niet-technische manier uit de doeken wat Artificiële Intelligentie (AI) is, kan en niet kan. Ze geeft daarbij voorbeelden van toepassingen in o.a. de gezondheidszorg, maar wijst eveneens op de uitdagingen.

Meer info

In een wereld waar stress, schermtijd en hectische levensstijlen de norm zijn, treft hoofdpijn één op zes mensen in België. Er zijn verschillende soorten hoofdpijn, zoals spanningshoofdpijn en migraine, die een aanzienlijke impact hebben op de kwaliteit van leven van miljoenen mensen wereldwijd. In dit wetenschapscafé laten ze twee onderzoekers aan het woord, die vanuit hun expertise dieper ingaan op hoofdpijnaandoeningen, en deze zijn beide van Team PreDiCT, namelijk Sofie Van Hoecke en Nicolas Vandenbussche. Kim Verhaeghe, redacteur en nieuwscoördinator bij Eos Wetenschap, is de moderator van het wetenschapscafé.

Meer info

Podcast & YouTube

The ICA Commission on Location Based Services and Ghent University organise the 18th International Conference on Location Based Services (LBS 2023), which will take place in Ghent, Belgium on 20-22 November 2023. Prof. Krzysztof Janowicz from University of Vienna, Bart Rosseau from Digitaal Vlaanderen, Prof. Wim Hardyns from Ghent University (UGent), Prof. Sofie Van Hoecke from UGent-imec (Team PreDiCT), and Prof. Sidharta Gautama from UGent will be the keynote speakers at LBS 2023.

Sofie will give a talk about harnessing the power of hybrid AI for designing spatio-temporal behavioral markers for health and their dynamic visualisation.

With the advent of wearable technology and the ubiquity of smartphones, the way we monitor our health and lifestyle is ready to undergo a revolution. No longer are we limited to self-reporting of our behaviors and health, but we can now collect objective, spatio-temporal data through sensors that monitor our movements, sleep patterns, stress, wellbeing and more. These behavioral markers provide a wealth of information about our daily routines and can help to make informed decisions about our health and well-being. In this talk, we will explore how wearables and smartphones are transforming the way we monitor our lifestyle. We will discuss the potential of hybrid AI for designing these behavioral markers and the role they can play in improving our health and quality of life. Finally, we will also show how spatio-temporal visualization can be enabled to enhance visualization of the health data and behavioral markers to clinicians.

More info and conference program.

De context van een patiënt doet ertoe. Waar en bij wie woont een patiënt? Hoe zijn de slaap, voeding en lichaamsbeweging geweest de voorbije weken? Het zijn factoren die helpen om een diagnose te stellen, en om te bepalen wat de meest zinvolle behandeling is. En daarom probeert een arts die zaken in het korte bestek van een consultatie in kaart te krijgen. Artificiële intelligentie kan die opdracht vereenvoudigen en inzichten aanreiken.

Lees hier het opiniestuk van Peter Peumans over het Protego onderzoek van team PreDiCT in de Artsenkrant.

This seminar aims to address the challenges faced by companies and organizations in the blue economy when it comes to managing and extracting value from vast amounts of data. With the emergence of big data, artificial intelligence (AI), and machine learning (ML), there is an opportunity to turn this data deluge into actionable insights, driving innovation and sustainable growth. This event will bring together industry experts, practitioners, and researchers to discuss strategies, best practices, and practical solutions for effectively leveraging big data, AI, and ML in the blue economy.

Team PreDiCT will present their research on wind power forecasting using machine learning.

More info

On the 16th of October Flanders AI Research Program will have its annual Research Day! An inspiring get-together where researchers and AI specialists can meet up, share ideas and learn from each other.

Do you want to get inspired by the latest findings in AI research? Or would you like to present the challenges your organization is facing to the researchers of our program? Are you interested in the state-of-the-art research of the Flanders AI Research Program?

We will present our research on Semantic Tree Aggregation (poster).

More info and registration

Would you like to find out more about the latest AI solutions in personalized healthcare? Get recent findings first hand from top-notch researchers at this upcoming seminar. Vlaams AI Onderzoeksprogramma / Flanders AI Research Program and VAIA - Vlaamse AI Academie joined forces to organize this inspirational day for medtech professionals.

We will present our research on lifestyle monitoring using wearables and smartphone data (presentation), as well as our research on Semantic Tree Aggregation (poster).

More info and registration

Migraine blijft een slecht begrepen medisch probleem. Artificiële intelligentie kan helpen oplossingen te vinden.

Lees hier de column van Peter Peumans over het migraine onderzoek van team PreDiCT in De Tijd.

AI componeert nieuwe nummers in de stijl van Nick Cave en tekent een lachende olifant op een fiets. Verbluffend is dat zeker, maar het is zeker niet het enige waarin AI écht excelleert: snel veel data verwerken, anomalieën opsporen en inzichten genereren. En daarin schuilen (minstens) zeven toepassingen die ons leven effectief beter zullen maken.

In dit artikel schetsen Steven Latré, Sofie Van Hoecke van team PreDiCT, Matthias Strobbe en Francis wyffels zeven toepassingen, die dankzij AI, ons leven effectief beter zullen maken.

Lees hier het artikel in EOS Wetenschap.

Join the RestartSMEs project for a session where we delve into the transformative potential of fusing data-driven and knowledge-driven AI approaches. During this workshop, Sofie Van Hoecke from team PreDiCT will unveil the secrets of AI solutions in manufacturing. Learn how combining the power of advanced algorithms with the wisdom of domain expert can revolutionize your business outcomes.

Meer info en inschrijven

Fempreneurs wil onderneemsters versterken in hun inzichten, capaciteiten en kennis om hun zaak uit te bouwen maar ook in hun weerbaarheid en zelfvertrouwen. Ze organiseren verschillende digitale workshops waar onderneemsters zich kunnen onderdompelen in verschillende onderwerpen die hun zaak vooruit kunnen helpen.

Op 24 mei geeft Sofie Van Hoecke een workshop over AI. Tijdens deze online workshop verkennen we het potentieel van artificiële intelligentie voor ondernemingen aan de hand van enkele illustratieve use cases. We zoomen in op machine learning en belichten de voordelen die jij als ondernemer kan halen uit artificiële intelligentie.

Meer info en inschrijven

Several technological breakthroughs will create ground-breaking innovations in various societal and economic sectors. During ITF Flanders, imec and several key partners look into the role of technology in the future: which challenges can we tackle leveraging an ingenious combination of hardware and software? In specific domains, this combination is clearly very promising, notably in preventive health care, sustainable industries, safe mobility, and a resilient energy system. How can different stakeholders in Flanders today start preparing for tomorrow’s technology and its vast possibilities?

Sofie Van Hoecke will give a talk on lifestyle monitoring through wearables and smartphone sensors to shift from self-reporting to behavioral markers at ITF Flanders during the Health thematic session.

More info on ITF Flanders and the thematic session on Health can be found here.

In het kader van haar dienst aan de gemeenschap ijvert Rotary club Gent, sinds haar oprichting in 1927, voor de verbetering van de levensomstandigheden van de zwakkere groepen in onze maatschappij. Het is duidelijk dat wetenschappelijk onderzoek hier een essentiële bijdrage kan leveren om tot duurzame oplossingen te komen. Sinds 2019 organiseert Rotary club Gent hiervoor dan ook een wedstrijd, onder de benaming Rotary Science Award (ROSA), die er in bestaat ieder jaar een andere Faculteit van de Universiteit Gent aan te spreken om deel te nemen, en waaraan lopende of net uitgevoerde wetenschappelijk projecten kunnen deelnemen waarmee de strijd tegen armoede bevorderd wordt, kunnen meedingen voor een prijs van 10.000€ die dan kan gebruikt worden om hun project praktisch te valoriseren.

In 2023 gaat de Rotary Science Award naar Team PreDiCT en meer bepaald naar Peter Ward’s PhD onderzoek naar het ontwerp en de ontwikkeling van een AI-driven digitaal pathologietoestel voor in-the-field parasietdetectie. De bekendmaking van de prijs gebeurt op 8 mei tijdens een gala-diner met de Rector van de Gentse Universiteit, de Burgemeester van de stad Gent, de decaan van de betrokken Faculteit en vertegenwoordigers van het Gentse Rotary District.

Tikt je collega hard op z’n toetsenbord? Dat ligt aan stress!

De manier waarop je op de computer werkt, zou een betere manier zijn om stress vast te stellen dan je hartslag. Dat blijkt uit een Zwitsers onderzoek.

“Als je gestresseerd bent zal je actiever met je muis bewegen, iets minder gecontroleerde bewegingen maken, meer fouten tikken en dus meer de backspace gebruiken” zegt Sofie Van Hoecke, professor aan de UGent en onderzoeker aan IMEC. Zij doet gelijkaardig onderzoek naar artificiële intelligentie.

Het onderzoek kan werkgevers op termijn helpen om voortijdig burn-outs te spotten bij hun werknemers.

Het interview kan je herbeluisteren op Radio 1.

Jarne Verhaeghe presented his work and his vision about causal AI for healthcare applications as a keynote speaker at the European AI week 29 March. He discussed how causality is essential in AI and how you should approach these problems from a causality point of view.

Voor de column ‘topingenieur’ in het blad Engineeringnet Magazine had Luc De Smet een gesprek met Sofie van Team PreDiCT over haar studie en loopbaan, de stenen in de rivier, en de professionele omgeving.

Het interview kan je vinden in het Engineeringnet Magazine van Maart 2023 alsook in NL of FR op de website van Engineeringnet.

In het leerwerkbedrijf Open Atelier in Pelt werden vandaag de eerste resultaten voorgesteld van het Living Lab project “Technologie Werkbaar Werk” van Vlaams minister van Werk en Innovatie Jo Brouns (CD&V). Het project wil het werk van werknemers op zowel fysiek als mentaal vlak lichter en werkbaarder maken door innovatieve technologie te gebruiken. Daarmee wil het project ook de werkzaamheidsgraad verhogen. Er wordt 2,4 miljoen euro in geïnvesteerd, verspreid over 3 jaar.

Ons PreDiCT team werkt hiervoor samen met andere collega’s van imec-IDLab-UGent, Flanders Make en zijn labo’s aan UHasselt en VUB, imec-SMIT-VUB, Sirris en het HIVA-Onderzoeksinstituut voor Arbeid en Samenleving (KU Leuven). PreDiCT kijkt vooral naar het monitoren van stress bij werknemers enerzijds en het monitoren van de werkomgeving anderzijds.

Persberichten:
VRT NWS - Robot Magazine - Engineeringnet

The Research Impact award at the Faculty of Engineering & Architecture Research Symposium (FEARS) 2022 at UGent, was awarded to Ganjour Mazaev for his work on leak localization in drinking water networks.

The 4th Industrial revolution, IoT and predictive analytics are bringing unseen possibilities in maintenance, reliability and condition monitoring. The Asset Performance 4.0 Virtual Conference offers a unique opportunity to learn how new 4.0 technologies and fundamentals in operations, maintenance and asset management reinforce each other in order to achieve higher equipment reliability and cost performance in asset intensive industries.

On Tuesday, October 25th, Sofie Van Hoecke gives a talk on “Hybrid AI for PHM: embedding expert knowledge into your AI” within the Artifical Intelligence Systems & Algorithms track of this virtual conference.

More info can be found here.

Machine learning opened up new ways of solving technical challenges by training models on data instead of directly implementing rules & logic. This offers a lot of new opportunities for solving difficult problems.

However, sometimes it can also be useful to combine these machine learning models with (expert) rules, to get the best possible outcome and leverage the benefits of both expert knowledge as well as machine learning models.

Hybrid AI is the name of this field, and focuses on combining non-symbolic AI (eg. machine learning), with symbolic AI (eg. expert rules). Sofie Van Hoecke (PreDiCT / Ghent University-imec) and Matthias Feys (ML6) will give an overview of this field by tackling the following topics:

• Why and when Hybrid AI is relevant for your situation
‍
• An overview of different ways to combine rules with machine learning models
• Concrete examples where hybrid AI was implemented

More info and registration can be found here.

Jarne Verhaeghe and Bram Steenwinckel both got the opportunity to present their work at ECML PKDD 2022.

Jarne Verhaeghe presented the novel Shapley-based wrapper feature selection method, Powershap, at ECML PKDD 2022 in both a presentation and poster format. A shorter format of the presentation can be found on YouTube while more information about Powershap can be found on GitHub.

Bram Steenwinckel presented INK, a novel interpretable Graph structure and embedding method. More information can be found on the GitHub.

InspiringFifty Belgium, the non-profit organization aiming to increase the number of women in leadership positions in the technology sector, today announced the 50 Belgian women leaders in the field. The jury team chose from no less than 173 nominations. ​Sofie Van Hoecke from team PreDiCT is one of the winners.

Ganjour Mazaev presented the hybrid AI solution designed in the SmartWaterGrid project on Trefdag Digitaal Vlaanderen, a govtech event bringing together digital experts from both the public and private sector.

Van het HealthTech Ondernemerscafé georganiseerd door BlueHealth Innovation Center waar Sofie van Team PreDiCT en Stephane Willaert van Robovision met elkaar in gesprek gingen over de toekomst van AI in Healthcare, maakte Tom Braekeleirs een HealthNerd Podcast getiteld: “Die over AI”.

De podcast kan je hier beluisteren.

During TechniShow 2022, the industry will come together to hear, taste, smell and see the future on developments in the manufacturing industry and learn how to ensure that you are ready for the future and how to prepare for the ever-changing challenges and keep all professional knowledge up-to-date. On Friday, September 2nd, Sofie will give a talk on hybrid AI and how to embed expert knowledge into machine learning for prognostics health management.

More info on the event can be found here.

Op woensdag 24 augustus gaat BlueHealth Innovation Center in gesprek met Sofie van Team PreDiCT en Stephane Willaert van Robovision om hun ervaringen te delen en het gesprek aan te gaan over de toekomst van AI in Healthcare.

More info on the event can be found here.

Jonas Van Der Donckt was invited to give a talk to the ubicomp research group of Siegen University, Germany. In this talk he highlighted the two main parts he works upon. More specifically, Jonas zoomed in on

1. the open source packages we make to ease data wrangling for time series
2. the mBrain research study in which we aim to improve chronic migraine management and answer research questions of these patients with data collected in ambulatory environments.

Sparked your interest? If so, the video can be found here!

Emiel Deprost, Jeroen Van Der Donckt, Vic Degraeve and Gilles Vandewiele from team PreDiCT (together with Gertjan Demulder from team KNoWS) achieved the 11th position in the recent BirdCLEF 2022 competition on Kaggle, resulting in a gold medal.

The goal of the competition was to create a model that can predict whether or not certain birds occur in provided soundscape recordings. Such a model allows for bioacoustic monitoring, a cost-effective strategy for studying endangered bird populations.

An overview of their solution can be found on Kaggle, or in an interview given by Gilles.

On Future Summits 2022, imec’s flagship event on nanoelectronics advances & deeptech solutions, team PreDiCT showcased three demos:

• Michael and Ganjour gave a demo of the SmartWaterGrid project,
• Jonas, Marija, Bram and Mathias gave a demo of our mBrain project, and
• Matthias and Emiel presented our Smart Buildings project.

Sofie Van Hoecke also gave a talk within the Food for Data, Data for Food BIZ FFWD session on the potential of our predictive maintenance research for the food industry.

The aim of Sofie Van Hoecke’s keynote is to gain more insight into the context, profile and lifestyle of people in order to identify triggers and behaviors linked to stress, depression and/or migraines. Identifying these triggers allows the people to make adaptations to their lifestyle to improve their health. To do so, data-driven and knowledge-driven AI approaches are combined to design a health management solution that enables the shift from subjective self-reporting towards objective continuous monitoring, empowering both patients and clinicians. To increase adoption of such AI models in clinical practice, and eliminate the risk of using these models for clinical decision support, we need to shift towards explainable and trustworthy AI.

More info and agenda of the symposium can be found here.

AI4Growth and Flanders AI organize the 2nd edition of the AI4 Your Business Conference, taking place during the European’s AI Week on Tuesday March 15, 2022 in Ghent.

Sofie Van Hoecke will give a talk on hybrid AI and how expert knowledge can be embedded in your AI solution to improve predictive performance.

Jonas and Jeroen Van Der Donckt will present our in-house designed plotly-resampler and tsflex open source libraries.

In the afternoon, Sofie Van Hoecke will present the fusion of machine learning and semantic technologies for improved migraine management in the session on Trustworthy AI in healthcare.

After 2 years of brainstorms, prototypes, and lots of research, the imec.icon SmartWaterGrid comes to an end.

The SmartWaterGrid consortium designed a smart system that substantiates, facilitates and automates leak localisation up to street level in soft real-time.

SmartWaterGrid surpassed the state of the art in terms of data capture across the water grid as well as on hybrid AI modeling, which involves both data-driven and model-driven digital twin approaches applied to collected sensor data. The resulting demonstrator and strategy will be incorporated into the offerings of all industry partners to boost service levels, decrease costs and waste and meet increased pressure from regulators.

Join our closing event and discover the innovative project results. Besides the technical insights, also integrated demos will be given during this closing event. We will also give a sneak peek on the next steps.

When? November 25th, 2021

Where? De Watergroep, Broekstraat 150, 3001 Heverlee

More info and registration

Gilles from Team PreDiCT and his Kaggle companions came out victorious out of 2659 teams in the Google Brain - Ventilator Pressure Challenge. In this competition, they were given data of a lung ventilator and were asked to predict the pressure within the lung at any given timestep. A short overview of their solution can already be found here but a more extensive write-up and accompanying code will be released in the nearby future.

Hoe kan artificiële intelligentie migraine-patiënten helpen? Ontdek het tijdens deze digistatie in De Krook!

Heel wat mensen hebben regelmatig last van migraine. Waar die migraine vandaan komt, of wat die migraine uitlokt, is helaas vaak moeilijk te achterhalen. Wordt het beter met bewegen of net niet?

UZGent, UGent en imec onderzoeken hoe A.I. kan helpen bij het nauwkeuriger identificeren van je activiteiten en de mogelijke migraine triggers, om migraine-patiënten sneller vooruit te helpen.

Wil je graag te weten komen hoe dit precies werkt? Wil je zelf ook graag uitproberen hoe dit onderzoek je migraine kan helpen bestrijden? We leggen je tijdens deze digistatie graag uit hoe!

Tech showcase on using hybrid AI to detect migraine triggers and support patients

Many people regularly suffer from migraines. Unfortunately, it is often difficult to determine where the migraine comes from, or what triggers the migraine. Is it getting better with exercise or on the contrary? This demo shows how AI can help to accurately identify your user behavior and detect your migraine triggers. To do so, we designed a mobile health application that, accompanied with a wearable, machine learning and AI models, automatically learns the user behavior, type of migraine and potential migraine triggers. Patients and their doctors are currently evaluating this app and accompanying dashboards within a longitudinal clinical trial at the Ghent University Hospital to get a better view on the context in which they get headaches and optimize migraine management.

More info and free registration

Attending ESICM LIVES2021 and interested to see how our team wants to leverage data science to enhance patient care and clinical research? Join us at 15:30 CEST on Channel 8 via https://lives2021.e-lives.org for our thought-provoking talk on the prediction and causality analysis of Atrial Fibrillation in the ICU using machine learning.

Maandag 12 juli was de officiële kick off van het ADAM-project in ons leer- en innovatiecentrum RADar. Vorige maand kreeg RADar hiervoor een subsidie toegekend van Vlaams minister voor Innovatie Hilde Crevits. Het project wil de fundamenten leggen om via Artificiële intelligentie nieuwe kennis uit data te halen.

ADAM staat voor Advanced Data-Aided Medicine. Het doel van dit driejarig onderzoeksproject is dan ook artificiële intelligentie-modellen opzetten die ondersteuning bieden bij medische beslissingen. Via data kan op die manier snel de juiste behandeling gevonden worden voor patiënten.

Meer info

The combination of innovations in the field of AIML (artificial intelligence & machine learning), technology and real-world data drive improvements in patient care and outcomes. This webinar will showcase concrete examples of how AIML is used to optimize patient care and outcomes in multi-stakeholder projects. The aim of the webinar is to share these examples and highlight how innovative medicine companies are contributing to these initiatives. More info can be found here.

Comon is een collectief van wetenschappers, technologen en creatievelingen van Stad Gent, imec, Universiteit Gent en Bibliotheek De Krook. Het doel? Van Gent een betere, aangenamere en wijzere stad maken. Met wetenschap en technologie als deel van de oplossing.

Eén van de makers van Comon is Sofie Van Hoecke, professor Artificiële Intelligentie aan de UGent. Sofie werkt met haar team aan een app die migraine-aanvallen voorspelt.

Lees hier het interview.

Team IC/DC, with team PreDiCT’s doctoral students Jarne Verhaeghe and Thomas De Corte as the lead of team IC/DC, received third place on 28/05/21 in the 2021 European Society of Intensive Care Medicine Datathon with the prediction of Atrial Fibrilation using uncertainty quantifying machine learning. The winning team members received free access to the conference and may present their results at the ESICM tech forum. Jarne Verhaeghe and Thomas De Corte are promoted by Sofie Van Hoecke (PreDiCT), Femke Ongenae (IDLab), and Jan De Waele (UZGhent) and together they work on Hybrid machinE leaRning for Improved Infection management in Critically ill patients (FWO HEROI²C project).

How to find synergies in data-driven and expert-based approaches for smart maintenance in industrial processes? Can we combine expert knowledge and machine learning for predictive maintenance? What about root cause analysis and anomaly detection from an AI perspective? How do the expert feedback and physics-based models fit in all this?

Join us at our free Webinar-event ‘Hybrid Machine Learning for Smart Maintenance’ on 25th March 2021 from 9 till 11 am to find out the answers to the questions above and to interact with us!

In this exclusive webinar outreach session, speakers from both the renowned scientific community @UGent (IDLab & EELab), imec, Flanders Make and industry will elaborate upon hybrid machine learning for smart maintenance. We show how to find synergies in data-driven and expert-based approaches. We detail how providing expert feedback can be used for model optimization of smart maintenance, what physics-based machine learning strategies can be deployed, and how expert knowledge can be incorporated into machine learning for root cause analysis and anomaly detection. Next to academic perspectives on smart maintenance, testimonials from a service and manufacturing company will be provided on how they perform smart maintenance using hybrid machine learning. We also warmly invite you to interact with us during the Q&A sessions.

Agenda and registration

Artificial intelligence (AI) plays an increasingly important role in various aspects of our lives. Also for companies, emerging AI and ML applications lead to a competitive advantage for the early adopters embracing these technologies. To accelerate the economic potential of these technologies, to link expertise and knowledge, AI4Growth was founded as a local network within AI & ML. With members from both research-driven AI companies, AI experts from industry and academia, and companies that want to focus more on the application of AI & ML, AI4Growth creates a unique platform. In this session AI4Growth explores what the AI-(r)evolution means for society and for companies. The potential challenges of the increasing impact of AI technology on society are discussed, but also the opportunities offered by this technological revolution. With AI experts such as Sofie Van Hoecke, and input from AI believers and adopters Benjamin Vandemarliere (Boltzmann) and Jonathan Berte (Robovision) this will be an AI-session not to be missed! Digital-first professional Clo Willaerts will moderate and give you insights into the benefits and challenges of AI.

Learn more about our research on machine learning and dynamic dashboard technologies for scalable REB health monitoring and Remaining Useful Life prediction during this webinar.

The development of Artificial Intelligence has gained momentum all over the world. The Flemish AI impulse program was set up to ensure that Flanders can continue to keep its leading position on AI. This program enables our researchers to find solutions for the challenges of tomorrow and today.

This special details AI research in Health and Society, and more specifically the PreDiCT research within the mBrain project on AI on wearables and context data for improved and personalized migraine management.

In a year in which public health is high on the agenda, science magazine Eos, in collaboration with research institute imec and Ghent University, is devoting a special issue to health and technology, with three contributions from PreDiCT.

The Eos special has three parts. The first is about prevention. Thanks to sensors, digital doubles that collect all our medical data and personalized medicines, we can detect and prevent more diseases than ever before. Part two focuses on therapy. New AI models allow doctors to design more efficient treatments, tailored to the patient. The final part focuses on care and how robots can lend a mechanical hand, and investigates the most pressing question of the moment: how can we overcome infectious diseases?

The contributions from our PreDiCT team focus on our research on migraine management using wearables, our research on personalized antibiotics therapy in the intensive care, and our vision on the future of AI for health. Read here this latter contribution (in Dutch).

Master thesis student Ward Haeck received the ArcelorMittal 2020 award for his masterthesis on ‘Transfer learning and one-shot learning for surface inspection of flat steel.’ Promotors were Sofie Van Hoecke (PreDiCT), Andy Van Yperen-De Deyne (ArcelorMittal), and Dieter De Paepe (PreDiCT).

DyVerSIFy facilitates operators in exploiting the value generated by sensor analytics to the fullest by delivering software components and methodologies that advance the state-of-the-art in dynamic visualization, adaptive anomaly detection (AD) & root cause analysis (RCA), and scalable services & processing.

Discover the innovative project results that fuse the power of machine learning and semantics to predictive maintenance, enable both user-driven and anomaly-driven dashboards, and process everything in a scalable way. Besides the technical insights, also three integrated business demos will be given during this closing event.

Leaflet

Zal een vrouw te vroeg bevallen of niet? Op die vraag kan een app, ontworpen door IDlab (Universiteit Gent), antwoord bieden. De software zou ook kunnen dienen voor mensen met migraine of dementie.

Persartikel