Block Four: Module 21 - Healthcare and Life Sciences

Introduction

Presenters: Three longtime collaborators of the Executive Master in AI
- Mr. Joanni Bugti
  - Chair of AI in Digital Medicine at the University of M
  - Lecturer at Digital Health at University of Le
  - Consultant on health innovation in AI
  - Completed a postdoctoral fellowship at Harvard University (2020-2021)
- Mr. Roman Aldo
  - Senior lawyer at CMS De Backer
  - Specialized in EU competition law and mergers
  - Focus on electronic communication, media, and pharmaceutical sectors
- Mr. Thibaut Hut
  - Founder and CEO at DNA Analytics
  - Holds a PhD in Machine Learning
  - Company focuses on data science technology in healthcare

Initial Thoughts
- Challenges of AI in Clinical Settings
- Example: Scandals around romantic chatbots
- Case: Death by chatbot in Belgium due to eco-anxiety

The Necessity of AI
- Limited learning time for doctors
- AI can exploit large data sets
- Definition by High-Level Expert Group on AI

AI Does Not Replicate Human Intelligence Completely
- Ladder of Causation by Judea Pearl: AI at the first level of Association
- Press often exaggerates AI's capabilities
- Challenges in Diagnostic Fields (Radiology, Pathology, Lab Medicine)
Need for Clinical Validation
- Replication issues in Radiology research
- Systems often adopted without proper clinical validation

Domains in Clinical Medicine
- Monitoring, Diagnosis, Prevention, Administration
- Existing applications in Diabetes, Epilepsy, etc.
Future Prospects for AI in Healthcare
- Specialized Health AI, Augmented Patients, AI in Research
- Concerns about disparity and access to healthcare

Strengths
- Unique and strong AI ecosystem with limited financial resources
- Leading in AI educational efforts for healthcare professionals
- Belgium was the first to propose a federal adoption plan for AI
- Leading in clinical research and number of clinical trials

Introduction to Genomics
- Human Genome consists of over 20,000 genes
- Importance of understanding gene mutation and its pathogenicity
Technological Advances
- Transition from genetic approach to genomic approach
- Sequencing machines translating DNA into digital data
- Importance of sequencing entire genomes
Examples of Genomic Discoveries
- CCR5 Δ32 mutation related to immunity against AIDS
- Role of epistatic genes and protective genes

Significance in Research
- AI needed to analyze vast genomic data
- Tools like AlphaFold for visualizing protein structures
Applications and Projects
- Genome for Brussels initiative and tools like GVAP Live
- AlphaMissens for analyzing protein variations
- Personalized Medicine and the need for massive data for understanding and treating diseases

AI-Driven Drug Discovery
- Fast-tracking drug discovery processes
- Using tools like AlphaFold to visualize protein modifications
- Understanding the limits and capabilities of large language models
Clinical Research
- Importance of data interoperability for clinical trials
- Initiatives for better data sharing and usage (e.g., Ina project)
- Tools aiding clinical research and reducing time and costs

Biomanufacturing
- Complexities in producing biotech products
- Role of AI in process design, qualification, and improvement
- Examples of AI improving manufacturing efficiency

Existing and Upcoming Regulations
- Impact of GDPR, MH, IVDR, MDR, and AI Act on healthcare AI
- Need for good data practice and standards (GMLP, GAMP5)

Emphasis on keeping human oversight in AI applications.
Recognition of Belgium's strong ecosystem and innovative efforts in AI for healthcare.