When:
January 30th, 2025
Location:
ilionx
Laan Corpus Den Hoorn 100
9728 JR Groningen
Dietary restrictions:
Please reach out as soon as possible if you have any dietary restrictions.
Agenda:
17:30 - Walk in and food
18:20 - Welcome by GroningenML and ilionx
18:30 - Presentation 1: Using NLP for Better Autism Recognition in Women by Olav Gruppen (ilionx)
19:00 - Presentation 2: All deep learning is convolution by Hylke Donker (UMCG)
19:30 - Break
20:00 - Presentation 3: Predicting Accurate Care Indications with Text Analytics at Noorderbreedte & Van Boeijen Groep by Rik Opdam, Thirza Osinga en Bart Joosten (ilionx)
20:30 - Closing statements, drinks & networking
21:00 - End of meetup
Presentation 1: Using NLP for Better Autism Recognition in Women
In girls and women, autism is often not recognized or diagnosed too late. In the Elucidating Female Autism Study (EmFASiS), we use Named Entity Recognition (NER) to analyze electronic health records of clients at Lentis. This helps us understand the differences in symptoms between females and males with autism. By categorizing and filtering symptoms, we aim to improve early diagnosis and tailored treatment for women.
Keywords: nlp; autism; gender; EHR analysis
Presentation 2: All deep learning is convolution
The breakthrough of AI started with a simple inductive bias: translation invariance. And thus, AlexNet was born. The next big discovery came from Google, inventing the transformer, which still forms the foundation of large language models today. Recent work on state space models provides a fresh perspective on these different architectures. It turns out that all these models, recursive and convolutional neural nets, transformers and state space models are just different ways of looking at the same thing. In this talk, I'll try to give a birds eye view how all these concepts are connected.
Keywords: Neural architectures; State space models; Transformers
Presentation 3: Predicting Accurate Care Indications with Text Analytics at Noorderbreedte & Van Boeijen Groep
In the healthcare sector, providing the best possible care for clients is paramount. At Noorderbreedte, a leading elderly care institution, care workers often delivered more care than indicated in the care packages (ZZP) or care profiles of their clients, resulting in significant costs. To address this, ilionx developed a machine learning model using text analytics to predict the accurate care indication for each client. This innovative approach ensures that the care packages for approximately 1,200 clients are up-to-date, reducing unnecessary costs and administrative burdens. By leveraging AI and big data, Noorderbreedte can now provide more personalized and efficient care, ultimately enhancing the quality of life for their client
Keywords: nlp; elderly care, EHR analysis
