BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:ENBIS Spring Meeting 2025
DTSTART:20250529T070000Z
DTEND:20250530T164000Z
DTSTAMP:20260714T101300Z
UID:indico-event-66@conferences.enbis.org
CONTACT:office@enbis.org
DESCRIPTION:Speakers: Raffaele Vitale (Université de Lille)\, Marco P. Se
 abra dos Reis (Department of Chemical Engineering\, University of Coimbra)
 \n\nWelcome to the ENBIS Spring Meeting \nQuality by Design (QbD) & Pro
 cess Analytical Technology (PAT): Statistical\, AI\, and “Grey” Approa
 ches\nCoimbra\, Portugal\, May 29-30\, 2025\n\nAim of the Spring Meeting\n
 Quality by Design (QbD) plays an important role in the modern industry. Of
 ten\, it goes hand in hand with technological solutions that secure the su
 pervision and stability of the processes in the design space (DS)\, such a
 s spectroscopy\, namely near-infrared\, Raman\, fluorescence\, or UV\, col
 lectively called Process Analytical Technology (PAT). QbD and PAT initiati
 ves have benefited from concepts\, methodologies and tools arising from di
 fferent corners of the data-driven sciences\, such as statistics\, chemome
 trics\, and machine learning. The integration of first principles and exis
 ting knowledge with empirical modelling\, called grey modelling\, is anoth
 er area of current active research that may bring important contributions 
 to QbD/PAT. Inevitably\, Artificial Intelligence will also play a role in 
 the future of these fields\, but which exactly is an open question. Theref
 ore\, it is both important and opportune to assess how these different app
 roaches can further contribute to improving or reinventing QbD and PAT ini
 tiatives\, either isolated or cooperatively. The main goal of the ENBIS Sp
 ring Meeting 2025 is to foster high-level discussions on these and other r
 elated topics.\nAll stakeholders\, from graduate students to professionals
 \, from researchers to industrial managers\, are invited to participate ac
 tively. We welcome contributions in the following areas (the list is not e
 xhaustive)\, with a focus on their application in industry in the scope of
  QbD/PAT:\n\nDesign of experiments (physical systems or in silico)\nActive
  learning\, Bayesian optimization and derivative-free optimization\nDesign
  spaces under uncertainty (Bayesian and frequentist approaches)\nExplainab
 le and Generative AI\nTransfer learning\nNew grey (hybrid) modelling archi
 tectures\nDigital Twins\nPhysics-informed neural networks (PINNs)\nChemome
 trics: classical and new AI methods\nIntegrated approaches of the above\n\
 n\nContact information\nFor any question about the meeting venue and scien
 tific programme\, registration and paper submission\, feel free to conta
 ct the ENBIS Permanent Office : office@enbis.org.\n \nLocal Organizing 
 Committee: \n\nMarco S. Reis (Chair)\nTiago J. Rato\nJoão Coutinho\nRúbe
 n Gariso\n\n \nProgramme Committee:\n\nMarco S. Reis (Chair)\, University
  of Coimbra\, Portugal\nRaffaele Vitale (Co-Chair)\, Université de Lille\
 , France\nJacqueline Asscher\, Kinneret College\, Israel\nAlberto Ferrer\,
  Universidad Politecnica de Valencia\, Spain\nPierantonio Facco\, Universi
 ty of Padova\, Italy\nSonja Kuhnt\, Dortmund University of Applied Science
 s and Arts\, Germany\n\n \nENBIS Spring Meeting 2025 Highlights\nPlenary 
 speakers\n\nAntonio Del Rio Chanona (Imperial College\, UK)\nLLM and human
 -in-the-loop Bayesian optimization for chemical experiments\nBayesian opti
 mization has proven effective for optimizing expensive-to-evaluate functio
 ns in Chemical Engineering. However\, valuable physical insights from doma
 in experts are often overlooked. This article introduces a collaborative B
 ayesian optimization approach that integrates both human expertise and lar
 ge language models (LLMs) into the data-driven decision-making process. By
  combining high-throughput Bayesian optimization with discrete decision th
 eory\, experts and LLMs collaboratively influence the selection of experim
 ents via a human-LLM-in-the-loop discrete choice mechanism. We propose a m
 ulti-objective approach to generate a diverse set of high-utility and dist
 inct solutions\, from which the expert\, supported by an LLM\, selects the
  preferred solution for evaluation at each iteration. The LLM assists in i
 nterpreting complex model outputs\, suggesting experimental strategies\, a
 nd mitigating cognitive biases\, thereby augmenting human decision-making 
 while maintaining interpretability and accountability. Our methodology ret
 ains the advantages of Bayesian optimization while incorporating expert kn
 owledge and AI-driven guidance. The approach is demonstrated across variou
 s case studies\, including bioprocess optimization and reactor geometry de
 sign\, showing that even with an uninformed practitioner\, the algorithm r
 ecovers the regret of standard Bayesian optimization. By including continu
 ous expert-LLM interaction\, the proposed method enables faster convergenc
 e\, improved decision-making\, and enhanced accountability for Bayesian op
 timization in engineering systems.\n\n\nCarl Duchesne (Université Laval\,
  Canada)\n \n\n\nEstablishing Multivariate Specification Regions for Inco
 ming Raw Materials – a QbD approach\n\n\nEstablishing multivariate speci
 fication regions for selecting raw material lots entering a customer’s p
 lant is crucial for ensuring smooth operations and consistently achieving 
 final product quality targets. Moreover\, these regions guide the selectio
 n of suppliers. By meeting these specifications\, suppliers contribute to 
 customer satisfaction\, which can\, in turn\, enhance market share. Latent
  Variable Methods\, such as Partial Least Squares regression (PLS) and the
  more recent Sequential Multi-block PLS (SMB-PLS)\, have proven to be effe
 ctive data-driven approaches for defining multivariate specification regio
 ns. These methods model the relationships between raw material properties 
 (Critical Material Attributes – CMA)\, process variables (Critical Proce
 ss Parameters – CPP)\, and final product quality (Critical Quality Attri
 butes – CQA)\, enabling the identification of a lower-dimensional latent
  variable subspace that captures quality-relevant variations introduced by
  raw materials and process conditions. This subspace is central to the met
 hodology.\nWithin this latent variable space\, several statistical limits 
 are defined to ensure final product quality and guarantee data compliance 
 with the latent variable model\, collectively forming the multivariate spe
 cification region. After providing historical context on the early develop
 ment of these methods\, this presentation will cover several key topics. T
 hese include the data requirements for constructing latent variable models
  and how to organize them into distinct blocks. The techniques used to def
 ine the limits (in terms of shape and size) within the latent variable sub
 space—via direct mapping and latent variable model inversion—will be d
 iscussed\, along with methods for addressing uncertainties. Particular att
 ention will be given to how process variations generate different scenario
 s and approaches for establishing meaningful specification regions. Case s
 tudies using both simulated and industrial data will illustrate these meth
 ods. It will be demonstrated that the proposed framework aligns with the p
 rinciples of Quality by Design (QbD)\, with notable similarities to the De
 sign Space (DS) concept. The presentation will conclude by exploring poten
 tial future developments\, including the establishment of multivariate pro
 cess capability indices based on specification regions.\n\n\nhttps://confe
 rences.enbis.org/event/66/
IMAGE;VALUE=URI:https://conferences.enbis.org/event/66/logo-2590257072.png
URL:https://conferences.enbis.org/event/66/
END:VEVENT
END:VCALENDAR