PolyModels Hub

🚀 Big news, friends. BIG NEWS. 🎉 We're proud to announce our successful seed funding round of GBP 1.1M by Marathon Venture Capital. We couldn't have done it without our earliest adopters, followers, supporters, and friends. At PolyModels Hub, we are revolutionizing pharma process development by empowering scientists with advanced digital tools and insights to enhance every step of drug development and manufacturing. If you're part of our community - please consider us as your partner in 2024 to dramatically reduce development time and optimize results for your drug development teams. Check out the article and the full press release for more details: https://lnkd.in/dPgpRj8d Cheers!

We have been a bit quiet lately, but it is for a good reason - this year-end brought some exciting developments for PolyModels Hub! 🚀 Our team is growing, our network of partners is expanding, and we have been busier than ever. But do not worry, we will be back in the new year, fully recharged and ready to share more updates and valuable content with our modeling community, while continuing to build on our mission to transform how drug development and manufacturing are optimized. 💊 Stay tuned, and happy holidays from all of us at PolyModels Hub! 🎄✨

🔍⚙️ Optimizing Kinetic Model Identification with Artificial Neural Networks 🤖🧪 Mathematical models for reaction kinetics are pivotal for process design, control, and optimization. However, identifying the correct kinetic model structure from competing candidates is a significant challenge. This study by Enrico Sangoi, (our own) Marco Quaglio, Fabrizio Bezzo, and Federico Galvanin presents an innovative framework combining artificial neural networks (ANNs) with optimal experimental design to streamline the identification process. 🧩 Key Insights: 🤖 ANN-Powered Model Identification: Leverages ANNs to distinguish between kinetic model structures, enhancing accuracy and robustness. 🧪 Optimal Experimental Design: Introduces a methodology to minimize experimental requirements while improving the ANN's capability to discern the correct kinetic equations. ⚙️ Case Study Validation: Applied to a batch reaction system, the approach demonstrates significant improvements in model identification under varying experimental conditions and noise levels. 📊 Efficiency Gains: Reduces the experimental burden without compromising precision, enabling faster kinetic model development. 🌟 Broader Implications: Supports more efficient process design and control strategies, fostering advancements in chemical engineering and related fields. 📚 Link to Publication: https://lnkd.in/dmEpgupV #KineticModeling #ExperimentalDesign #ArtificialNeuralNetworks #ProcessOptimization #ChemicalEngineering #PolyModelsHub #ModelIdentification

🔍🧬 Multiscale Hybrid Modeling for CHO Cell Cultures: Bridging Scales with Enzyme-Constrained Flux Analysis ⚙️📈 Mammalian cell cultures remain vital to the pharmaceutical industry, producing numerous FDA-approved biopharmaceuticals annually. Modeling these complex systems is challenging due to the interplay of dynamics across multiple scales. This study by Oliver Pennington, Sebastián Espinel Ríos, Mauro Torres , Alan Dickson, and DONGDA ZHANG introduces a multiscale hybrid modeling framework, integrating macro-kinetic approaches with enzyme-constrained Dynamic Metabolic Flux Analysis (ecDMFA) to address this complexity. 🧩 Key Insights: ⚙️ Multiscale Hybrid Framework: Combines macro-kinetic modeling with machine learning to bridge macroscale bioprocess dynamics and microscale metabolic flux distributions. 🧪 Enzyme-Constrained Analysis: Leverages ecDMFA for detailed flux predictions, enabling insights into metabolic changes during Trastuzumab production in CHO cells. 📊 Accuracy and Trust: Achieves an 8% modeling error for macroscale dynamics—representing a 70% reduction from purely mechanistic models—while maintaining low uncertainty through bootstrapping. 🔍 Metabolic Insights: Highlights significant metabolic shifts during cell culture, with predictions aligning closely with observed literature data. 🌐 Digital Twin Potential: Demonstrates a pathway for advancing digital twins in biopharmaceutical manufacturing, enhancing process understanding and control. 📚 Link to Publication: https://lnkd.in/dBamnKW5 #MultiscaleModeling #HybridModeling #CellCulture #Biopharmaceuticals #DigitalTwins #MetabolicFluxAnalysis #CHOCells #ProcessOptimization

🔍⚗️ Unraveling Off-Cycle Mechanisms in Pd-Catalyzed Amination of Five-Membered Heteroaryl Halides 🔬🔗 Understanding complex catalytic processes is essential for optimizing challenging reactions in small-molecule synthesis. This study by Elaine Raguram, Jakob Dahl , Klavs Jensen, and Stephen Buchwald delves into the intricate mechanism of Pd-catalyzed amination of five-membered heteroaryl halides, revealing atypical pathways and off-cycle events. 🧩 Key Insights: 🔄 Atypical Pathways: Identifies an uncommon mechanism where NaOTMS, rather than the amine, binds first to the oxidative addition complex, forming an OTMS-bound Pd species as the resting state. ⚡ Rate-Limiting Steps: Reveals that turnover is limited by the formation of the Pd-amido complex, deviating from typical systems where reductive elimination is rate-limiting. 🔍 Off-Cycle Dynamics: Demonstrates that the amine-bound Pd complex, typically an on-cycle intermediate, functions as a reversible off-cycle species in this system. 🛠 Catalyst Deactivation: Highlights base-mediated decomposition of 4-bromothiazole as a key irreversible deactivation pathway, significantly impacting yields. 📈 Predictive Modeling: Employs kinetic modeling and predictive testing to uncover minor mechanistic pathways, paving the way for better reaction optimization strategies. 📚 Link to Publication: https://lnkd.in/dzp5fnZC #Catalysis #KineticModeling #PalladiumChemistry #OrganicSynthesis #ReactionMechanisms #SmallMoleculeResearch #Innovation

🔍🧪 Bioprocess Feeding Optimization through In Silico Dynamic Experiments and Hybrid Digital Models 🤖 🧫 Optimizing cell culture feeding strategies is essential for accelerating monoclonal antibody production while reducing the experimental and financial burden on biopharmaceutical process development. This proof-of-concept study by Gianmarco Barberi, Christian Giacopuzzi, and Pierantonio Facco introduces a novel hybrid digital model that combines Design of Dynamic Experiments (DoDE) with a semi-parametric modeling approach to optimize glucose and glutamine feeding schedules. 🧩 Key Insights: 🌐 Virtualized Experiments: By leveraging a hybrid digital model, this methodology minimizes experimental requirements to just nine runs, significantly reducing time and cost. 📈 Enhanced Productivity: Demonstrates a 34.9% increase in antibody titer compared to training data, outperforming traditional DoDE-based campaigns. 🛠 Time-Varying Optimization: Identifies dynamic feeding profiles that maximize antibody titer, offering precise control over process variables. 🚀 Accelerated Development: Enables in silico experimentation to complement traditional laboratory methods, streamlining the path to process optimization. 📚 Link to Publication: https://lnkd.in/dqFi-tHz #HybridModeling #DoDE #BioprocessOptimization #InSilicoExperimentation #MonoclonalAntibodies #Biopharmaceuticals #DigitalInnovation

🔍🧬 Model Predictive Control for Biopharmaceutical Production: A Strategic Framework ⚙️📈 The complexity of biopharmaceutical production demands innovative solutions to ensure product consistency and process efficiency. Model Predictive Control (MPC) has emerged as a transformative approach, offering real-time optimization and enhanced control over production variables. This review by Touraj Eslami and Alois Jungbauer underscores MPC’s pivotal role in enhancing operational reliability, ensuring robust control, and unlocking process innovation for next-generation biopharmaceuticals. 🧩 Key Insights: ⚙️ Process Optimization: MPC enhances operational efficiency by enabling real-time adjustments to production processes. 📡 PAT Integration: Seamlessly aligns with Process Analytical Technology (PAT) to support real-time monitoring and control. 🛡️ Robust Control: Addresses deterministic and stochastic challenges, ensuring consistent product quality. 🔄 Adaptive Flexibility: Enables swift process adjustments to mitigate deviations and maintain process integrity. 🧬 Biopharmaceutical Applications: Demonstrates broad utility across various production scenarios, equipping process engineers with advanced tools. 📚 Link to Publication: https://lnkd.in/dzN84KTP #ModelPredictiveControl #BiopharmaceuticalProduction #ProcessOptimization #PAT #AdvancedControlSystems #BiotechInnovation

🔍📊 CADET-Julia: Advancing Batch Chromatography Simulation with Open-Source Efficiency ⚙️💻 Batch chromatography simulations are pivotal in designing and optimizing separation processes. CADET-Julia, an open-source chromatography solver implemented in Julia, offers a robust and versatile platform for prototyping and refining chromatography models, including the General Rate Model (GRM). This publication by Jesper Frandsen, Jan Michael Breuer, Johannes Schmölder, Jakob Huusom, Krist Gernaey, Jens Abildskov and Eric von Lieres demonstrates the capabilities of CADET-Julia through benchmark comparisons, highlighting its computational efficiency and the advantages of employing advanced numerical methods. 🧩 Key Insights: 🛠 Advanced Numerical Techniques: Employs a discontinuous Galerkin spectral element method (DGSEM)for interstitial column mass balance and a generalized spatial Galerkin spectral method (GSM) for particle mass balance, ensuring precision and performance. 🚀 Performance Benchmarks: Julia implementations consistently outperformed traditional C++ implementations, showcasing the potential of Julia for computational efficiency in process simulations. ⚙ Comparison of Methods: Demonstrated the superiority of Galerkin methods over finite volume approaches for batch chromatography simulations. 🌐 Open-Source Accessibility: Offers a flexible, community-driven platform for researchers and engineers, accelerating innovation in chromatography modeling and simulation. 📂Link to Github: https://lnkd.in/dKXPkuYH 📚Link to Publication: https://lnkd.in/drv_P6Pk #CADETJulia #ChromatographyModeling #OpenSource #JuliaLang #ProcessSimulation #PharmaceuticalEngineering #SeparationScience

🔍🧬 Hybrid Modeling of Fed-Batch Cell Culture: Leveraging Physics-Informed Neural Networks ⚙️🧫 Fed-batch cell culture remains a cornerstone in biopharmaceutical manufacturing, producing essential therapeutics worldwide. The inherent variability of biological systems presents challenges in ensuring consistent product quality. This study by Shu Yang, Will Fahey, Brendha Truccollo, Ph.D., Jill Browning, PhD, Reza Kamyar, and Huiyi Cao introduces a hybrid modeling approach using Physics-Informed Neural Networks (PINNs), bridging first-principle biological insights with data-driven neural network capabilities. 🧩 Key Insights: ⚙️ Hybrid Model Innovation: Combines the rigor of first-principle models with the adaptability of neural networks for a robust representation of CHO cell culture processes. 🦠 Data-Driven Accuracy: Demonstrates better predictive power compared to pure mechanistic or data-driven models across various bioprocess conditions. 🎞  Versatile Scenarios: Performs effectively under diverse instrumentation setups and varying data availability, ensuring scalability and adaptability. 📈 Enhanced Predictions: Daily calibration using cell culture analyzer data significantly improves model accuracy, offering a pathway to real-time monitoring and advanced process control. 📚 Link to Publication: https://lnkd.in/dpx5YSVE #HybridModeling #PhysicsInformedAI #CellCulture #Biomanufacturing #ProcessControl #NeuralNetworks

🔍🔄 Characterising Continuous Blenders: Exploring Mass Holdup Behaviour 📏⚙️ Continuous blenders play a vital role in Continuous Direct Compaction (CDC), an emerging approach for manufacturing solid oral dosage forms. Among the critical factors influencing blender performance, mass holdup stands out as a key variable, directly impacted by material properties, equipment configurations, and process settings. This study by Hikaru Graeme Jolliffe, Maria Alejandra Velazco, Luis Martin de Juan, Martin Prostredny, Carlota Mendez Torrecillas, Gavin Reynolds, Deborah McElhone and John Robertson evaluates the Gericke GCM-450 blender under varying conditions, providing a deeper understanding of how factors such as weir geometry, material density, and flowability shape mass holdup behaviour. 🧩 Key Insights: 🛠 Impact of Weir Geometry: Examined angled versus horizontal weir configurations, revealing contrasting mass holdup behaviours based on material flow characteristics. 🌐 Material Properties: Density and flowability emerged as key drivers in determining whether material forms an inclined powder surface compatible with the weir geometry. 📈 Process Dynamics: Explored how throughput and impeller speed influence blender performance, emphasizing the interplay of multiple process dimensions. 🔍 Fundamental Process Understanding: Highlights the need for mechanistic insight to predict blender behaviour across diverse materials and equipment setups, supporting robust process design. 📚 Link to Publication: https://lnkd.in/dvkRY5Aa #ContinuousDirectCompaction #BlenderCharacterization #PharmaceuticalEngineering #ProcessOptimization #SolidDosageForms