Welcome & Opening Remarks [5 minutes]
Organizers introduces the theme and agenda
Medical-Academia-Industry Insights (6 talks) [140 minutes]
From the Medical Fraternity
Speaker 1: Prof. Dr. Anurag Agrawal
Title: Clinical trials for Health AI : Designing for impact
Speaker 2: Prof. Dr. Ramona Woitek
Title: Qualitative evaluation of human and artificial intelligence generated data - where are we and what is needed?
From the Academia
Speaker 1: Prof. Xiaohong Wang Gao
Title: State-of-the-art Machine Intelligence for Generating Super Resolution Images for HP Viruses Detection
Speaker 2: Prof. Sridhar Krishnan
Title: Translation of Digital Health Technologies
From the Industry
Speaker 1: Dr. Bipul Das
Title: Synthetic Data for AI in Imaging Systems
Speaker 2: Dr. Rakesh Mullick
Title: AI Translation in HealthCare - Robustness and Reliability
Stretch & Refresh Break [10 minutes]
Quick coffee/tea & networking
Panel Discussion [45 minutes]
Title: From Bias to Balance: Navigating Uncertainty, Trust, and Sustainability in Biomedical AI
Moderator(s): Prof. Sridhar Krishnan and Dr. Palak Handa
Panelists and audience: 5 Experts + Authors + Attendees
Questions on 4 different themes:
1. Uncertainty in Ground Truths:
How should AI systems and human experts manage uncertainty and variability in ground truth annotations to ensure robust model evaluation?
2. Bias, Variability, and Fairness:
What strategies -- including the use of synthetic or augmented data - can help mitigate bias and inter and intra variability without introducing artifacts?
3. Reliability and Adaptability of AI:
How can we ensure that biomedical AI systems remain reliable under distribution shifts and that the rise of generative AI does not compromise data quality standards?
4. Sustainability and Efficiency:
How can we design AI models for biomedical data assessment that are both high-performing and environmentally sustainable?
Paper Presentations [4 papers (8 mins. presentation + Q&A each), 50 minutes]
Session Chair: Dr. Bipul Das and Dr. Rakesh Mullick
Selected paper(s) title (in order of their presentation):
Reform-Net: Respiratory Analysis using PPG Signal and CNN-Transformer-Based Hybrid Model
Spatial ON: Occlusion Network for Diagnostics, Image Augmentation Targeting Specific Diseases
Preprocessing Techniques Evaluation for Capsule Endoscopy Frames using Quality and Texture-Based Metrics
An Intelligent, End-to-End Framework for Multimodal Biomedical Sensor Fusion: From Signal Integrity to Clinically Actionable Insights
Closing & Next Steps [20 minutes]
Summary, acknowledgements, and group photo