Krishna Harsha Adusumilli’s Post

🔬 Unlocking the Power of LLMs in Biopharma with Aganitha’s ARC™ Framework Large Language Models (LLMs) like ChatGPT will transform biomedical research, clinical trials, and medical affairs, but their integration requires strategic solutions. 🔍 Challenges in implementing LLMs for Biopharma 1. Common Pitfalls: Off-the-shelf models may produce hallucinations, struggle with math & stats, and fail to meet stringent Protected Health Information (PHI) requirements. 2. Private data integration: Cannot leverage valuable private data out of the box  3. Operational Hurdles: Training costs and deployment complexities pose significant barriers. 🚀 Aganitha’s Solution: The ARC™ Framework ARC™ combines LLM capabilities with biopharma expertise, offering customized tools and services. - Omics Data Insights: Utilize AI-powered tools to analyze genomic, transcriptomic, and proteomic data. - Hypothesis Generation: Analyze diverse datasets to generate testable hypotheses and discover novel drug targets and biomarkers. - Document workflows: Generate, update, compare and review documents such as regulatory submissions and contracts. - Reduced Hallucination: Minimize inaccurate outputs by incorporating authoritative datasets and ontologies. 👉 📽️ Watch ARC™ demos at https://lnkd.in/gWw5QKQj #Biopharma #AI #LLM #ChatGPT

Interested in how Large Language Models can be applied in biopharma? Check out this insightful playlist launched by Aganitha! This series explores practical applications of LLMs across various areas, including omics research, clinical trials, and medical affairs. I encourage all biopharma professionals and researchers to explore these demos and see how LLMs can enhance their work. #LLMs #ai #biopharma

Congratulations to the Aganitha team for this great leap in AI-based solutions! The ARC™ Framework is set to revolutionize biomedical research, clinical trials, and medical affairs by integrating advanced Large Language Models (LLMs) like ChatGPT with biopharma expertise. While off-the-shelf models like ChatGPT are powerful, they often face challenges such as producing hallucinations, struggling with math & statistics, and failing to meet stringent Protected Health Information (PHI) requirements. Aganitha’s ARC™ Framework addresses these issues, offering a tailored solution for the biopharma industry. For those in biopharma looking to overcome these hurdles and enhance their workflows, this could be the game-changer you've been waiting for. Explore how ARC™ can address common challenges, integrate valuable private data, and streamline your operations.

Out of the box capabilities of #LLMs are necessary but not sufficient to address some of the complex use cases in #biopharma R&D, clinical trials and medical affairs. Aganitha ARC™ framework brings together patterns and tools necessary to solve these complex use cases.

It's incredible to see how #ai is transforming efforts of University of Illinois Urbana-Champaign alumni in the field of biopharma, drug discovery, clinical trials and more. Prasad Chodavarapu is one such alum innovating in this realm. This underscores why U of I created the world's first engineering-focused college of medicine, Carle Illinois College of Medicine.

🚀 Discovering the Potential of Large Language Models in Drug Development 🧬✨ I recently came across a fascinating paper titled "Comparative Analysis of LLaMA and ChatGPT Embeddings for Molecule Embedding" from the University of Windsor, Canada. This research explores how Large Language Models (LLMs) like LLaMA and ChatGPT can be applied to cheminformatics, particularly focusing on molecular property prediction and drug-drug interaction prediction. 🔍 Key Highlights: 🏆 Performance: The study found that LLaMA-based SMILES embeddings outperform those generated by ChatGPT in predicting molecular properties and drug-drug interactions. 🎯 Applications: These LLMs have shown significant promise in advancing drug development by providing novel approaches to understanding chemical graphs. 💡 Conclusion: The research underscores the potential of LLMs in cheminformatics, highlighting their role in improving drug discovery processes. If you're interested in how AI and machine learning are transforming healthcare and drug development, this paper is definitely worth a read! 🔗Check out the full paper here: https://lnkd.in/euw5QgQf #ArtificialIntelligence #MachineLearning #DrugDevelopment #HealthcareInnovation #MolecularEmbedding #Cheminformatics #Research

AI has helped life sciences speed up the drug discovery process for years...way before ChatGPT was a shimmer in a developer's eye. AI and machine learning are saving lives with protein drug development and better antibody creation. But the document authoring and submission process that supports bringing new drugs to market is not maintaining the same pace. Join Regina Lynn Preciado and David M. Turner as they discuss how AI technology combined with structured content authoring can make innovation soar and produce outcomes that support patients as well as your bottom line. https://lnkd.in/eFiztUyN #structuredcontentauthoring #AIanddrugdevelopment #DCLLearningSeries

Everyone agrees there's high anticipation around AI's impact in drug development. However, in order for AI to reach its full impact in drug development, it's not as easy as tossing searches into ChatGPT. The successful application of AI to drug development requires expertise in knowing how best to incorporate AI into research as well as technology that appropriately integrates AI with other in silico approaches, such as QSP and PBPK. This article by Amin Rostami-Hodjegan and Piet Van Der Graaf demonstrates the various applications of AI to drug development currently implemented at Certara. View the full article at: https://ow.ly/Z8TM50SAqK8 #AI #DrugDiscovery #DrugDevelopment #QuantitativeSystemsPharmacology #PBPK

I'm excited to share some personal insights into our latest paper on FragLlama, our flagship model at the frontier of LLM and molecular design. Our work is driven by a series of "whys" that have guided our approach: --LLMs in Molecular Design: We commented why Large Language Models (LLMs) are so successful from a math perspective and why they can be adapted for molecular design tasks. --Fragment-Level Tokenization: Despite fragment-based modification being a straightforward chemical intuition, we delved into why using fragment-level tokens is efficient for molecular design. --Llama Model Modifications: We explain the rationale behind our specific modifications to the Llama model modules. --NLP-to-Molecule Mapping: Our paper details how key concepts from Natural Language Processing (NLP) translate to molecular structures, particularly what multi-head attention learns in this context. Since our primary focus is on computational degrader design, we included demonstration cases for the design of molecular glues and PROTACs. FragLlama demonstrates an impressive ability to reproduce expert-level designs while also exploring novel and promising chemical spaces. This highlights its potential to significantly augment the capabilities of medicinal chemists in drug design. PS: This is the first of this series—if you think we miss cite some work, please drop me a message. We will add in the future work. Extra thanks to the Oracle cloud team (amazingly stable infrastructure and strong tech support) Bryan Barker Ruzhu Chen and Prof. Jin Wang for all the demo cases discussions. #AI #MolecularDesign #DrugDiscovery #LLM #Innovation #PROTACs #MolecularGlues

Just out yesterday...FragLlama, a #LLM trained for #moleculardesign, featuring custom tokens that represent molecular fragments & functional groups. The model is for generating molecules given one or two fragments, for application scenarios like general hit-to-lead & lead optimization stage drug design, PROTAC linker design; mapping to commonly used drug design strategies like fragment growing and scaffold hopping. FragLlama demonstrates proficiency in reproducing expert-level designs while also exploring novel and promising #chemicalspaces, highlighting its potential to augment the capabilities of #medicinalchemists in #drugdesign. #NVIDIA GPUs on #OCI were used for training & represented a significant computational resource tailored for the demands of #largescale #languagemodel training in the domain of #chemistry. #Oracle #drugdiscovery #computationalchemistry