[Phailadephia, August 10, 2023] –
Philadelphia – Enkrisi is proud to announce the development of EnkrisiGPT, a large-scale generative AI model purpose-built for the pharmaceutical and biotech industry, designed and fine-tuned specifically for Regulatory Drug Development. This model is trained on various drug regulations, including International Council for Harmonisation technical requirements (ICH) and Food and Drug Administration (FDA) guidance data, to support a diverse set of NLP tasks within the industry.
While AI and large language models have found applications across various domains, drug development’s intricate and specialized language necessitates a domain-specific model. EnkrisiGPT is our pioneering step in harnessing this technology for the pharmaceutical industry. This model will enhance Enkrisi’s existing drug development NLP tasks, such as regulatory strategy, compliance and submission authoring, product development and validation, and quality assurance and risk management. Moreover, EnkrisiGPT will leverage the vast data available on the Enkrisi Terminal to serve our clients better, bringing the full potential of AI to the life sciences domain.
Enkrisi has been at the forefront of applying AI, Machine Learning, and NLP in regulatory drug development for over a decade. Our researchers have developed a mixed approach combining guidance data with general-purpose strategic datasets to train a model that excels in regulatory compliance benchmarks while maintaining competitive performance on general language model benchmarks.
To reach this milestone, Enkrisi’s ML Regulaotry Product and Research group collaborated with our AI team to construct the largest domain-specific datasets yet. Our data scientists have collected and maintained drug development language documents over twenty years, creating a comprehensive 36 million token dataset consisting of English development documents. This data was supplemented with an 11 million token public dataset to create a larger compliance knowledge base with over 50 million tokens. The resulting model was validated on existing development-specific NLP benchmarks, a suite of Enkrisi proprietary benchmarks, and assorted categories of technical-purpose NLP tasks from popular benchmarks. Notably, EnkrisiGPT outperforms existing open models of a similar size on drug development tasks by large margins while still achieving on par or better on general NLP benchmarks.
Edward Narke, Enkrisi’s Chief Innovation Officer, said, “For all the reasons generative LLMs are fascinating – intelligence learning, text generation, conversational coordination, etc. – we see tremendous value in creating the first LLM focused on the drug development domain. EnkrisiGPT will enable us to undertake many new regulatory application types while delivering much higher performance out-of-the-box than general models for each application.”
The Head of Enkrisi’s ML Product and Research team added, “Thanks to the collection of development documents Enkrisi has curated over two decades, we were able to cautiously create a large and clean, domain-specific dataset to train a language model that is best suited for drug development use cases. The superiority of machine learning and NLP models comes down to the data they are trained on. We’re excited to use EnkrisiGPT to improve existing NLP workflows while imagining further ways to put this model to work for the lifesciences industry.”
Enkrisi, which means Approval in Greek, was founded on the premise that biotechnology, medical devices, and pharmaceuticals promise a better future for everyone. Making good on that promise requires innovative approaches, world-class science, regulatory know-how and a team dedicated to bringing all of it together.
Introducing ENKRISI – the leading consulting group helping organizations comply with drug regulations through a comprehensive drug development model incorporating augmented Intelligence (A.I.) and data analytics. Our team of industry experts is equipped to help both emerging biotechs and established pharmaceuticals understand how to use A.I. and machine learning to stay on top of rapidly changing regulatory requirements. We’re here to help navigate the complex world of compliance, making it more manageable than ever!
We’re more than a team of experts, we’re your Partners.