Opportunities for Artificial Intelligence with real world patient data

31 Oct 2023, 10:30 → 2 Nov 2023, 17:15 Europe/Berlin

Helmholtz Information & Data Science Academy (HIDA)

Machine learning and artificial intelligence offer the possibility to find new ways to better diagnose and treat cancer. One of the most promising paths towards this goal is using real world patient data.

EHR records, observational data, images or genomic data are increasingly made available in large scale data repositories.These data repositories usually aim for federated data access and analysis capabilities.

In this workshop we plan to address three major questions that need to be answered to lift the potential that lies in real world patient data with AI:

1. Where can AI surpass traditional data analytics and even human capabilities?
2. How can we provide the most comprehensive and complete view on real world patient data?
3. What software infrastructure is needed to enable federated data access and analysis?

Our three day lunch to lunch workshop aims to bring together data and computer scientists, clinicians, biostatisticians, software and hardware infrastructure providers, guideline experts, stakeholders from science and industry and patient representatives to discuss and answer these questions.

The goal of the workshop is to draw connections on the European level to exchange information on the status of AI for understanding real world patient data.

We plan to assemble information on existing European and  national initiatives and solutions. With this information we want to identify and foster synergies and formulate a strategy supporting AI with real world patient data.

The outcome of this workshop will be a document detailing the opportunities of AI with real world patient data and what is needed to realize them.

Tuesday 31 October

14:00 → 14:50 Arrival & Registartion

14:50 → 15:00 Welcome

15:00 → 15:30 Introduction round and workshop overview

15:30 → 16:00 Coffee Break

16:00 → 17:10 Session Opportunities of AI with real world data Part 1

Session Chair: Michael Bussmann

16:00 'From single- to multi-modal learning, how far from medical precision?’

Speaker: Giorgos Papanastasiou (Pfizer Inc)

16:30 'Tumor detection, classification and segmentation with Mask R-CNN'

Speaker: Albert Saporta (EISBM)

16:50 'Using machine learning for the analysis of synchrotron radiation-based micro computed tomography data of bone implants'

Speaker: Berit Zeller-Plumhoff (Helmholtz-Zentrum Hereon)

17:10 → 18:00 Transfer to conference dinner

18:00 → 19:00 Conference dinner

Wednesday 1 November

09:00 → 10:00 Session Opportunities of AI with real world data Part 2

09:00 'Challenges of using electronic health records to build machine learning models for outcome predictions'

Speaker: Sebastian Boie (Pfizer Pharma GmbH)

09:20 'How AI is Reshaping Life Sciences Research: Examples from the Frontlines'

Speaker: Vedran Franke (BIMSB)

09:40 'The Helmholtz Foundation Model Initiative'

Speaker: Dagmar Kainmueller (MDC)

10:00 → 11:00 Discussion Opportunities of AI with real world data

Moderator: Michael Bussmann

11:00 → 11:30 Coffee Break

11:30 → 12:00 Wrap up Opportunities of AI with real world data

12:00 → 13:00 Lunch

13:00 → 13:40 Session Comprehensive and complete real world patient data Part 1

13:00 'Optimal treatment for patients with solid tumours in Europe through Artificial Intelligence'

Speaker: Frederic Kube (Pfizer Pharma GmbH)

13:20 'OPTIMA: Data Discovery and Outreach to Data Partners'

Speaker: Cheryl Tan (University of Oxford)

13:40 → 14:10 Coffee Break

14:10 → 15:10 Session Comprehensive and complete real world patient data Part 2

14:10 'How to unlock the potential of Multi-Omics for application in clinical routine?'

Speaker: Markus Kreuz (Fraunhofer IZI)

14:30 'Label-efficient deep learning for biomedical microscopy'

Speaker: Artur Yakimovich (CASUS - Center for Advanced Systems Understanding, Helmholtz-Zentrum Dresden-Rossendorf e.V. (HZDR), Untermarkt 20, 02826 Görlitz, Germany)

14:50 'Research with Real-World Data: Strategies and Structures at Charité and BIH'

Speaker: Fabian Prasser (BIH @ Charité)

15:10 → 16:10 Discussion Comprehensive and complete real world patient data

Moderator: Michael Bussmann

16:10 → 16:40 Coffee Break

16:40 → 17:10 Wrap up Comprehensive and complete real world patient data

17:10 → 17:50 Session Software infrastructures for federated data access and analysis Part 1

17:10 'Federated analytics and federated learning in big data healthcare projects'

Speaker: Bertrand De Meulder (EISBM)

17:30 'Federated Learning in Healthcare: Opportunities, Challenges, and Future Directions'

Speaker: Anshu Ankolekar (Maastricht University)

Thursday 2 November

09:00 → 10:00 Session Software infrastructures for federated data access and analysis Part 2

09:00 'HIFIS: Helmholtz Digital Services for Science — Collaboration made easy'

Speaker: Uwe Jandt (DESY, HIFIS Coordinator)

20231102_Jandt_CASUS.pdf

09:20 'Federated Scientific AI Services in the Helmholtz Cloud'

Speaker: Guido Juckeland (Helmholtz-Zentrum Dresden-Rossendorf)

09:40 'Developing a Federated Data Analysis Platform in the context of IMI OPTIMA'

Speaker: Andreas Kremer

10:00 → 11:00 Discussion Software infrastructures for federated data access and analysis

Moderator: Michael Bussmann

11:00 → 11:30 Coffee Break

11:30 → 12:00 Wrap up Software infrastructures for federated data access and analysis

12:00 → 13:00 Round table + open discussion; What is there? What is missing? What do we need?

13:00 → 14:00 Lunch

14:00 → 14:30 Action plan: How to find synergies and formulate an AI strategy

14:30 → 15:00 Summary of workshop: Status of the field. What have we learned? What can be next steps?