What is BioGPT, Microsoft’s language model for Life Sciences?

In the past few months, ChatGPT has been grabbing all the headlines and attention worldwide. But there are other language models that are undoubtedly revolutionizing all kinds of industries, such as Life Sciences. That is why many companies, like Narrativa, are investing in creating the best technology for the advancement of science. What is BioGPT, Microsoft’s language model for Life Sciences?

BioGPT, a model that exceeds human capacity

BioGPT is a language model transformer that has been developed by Microsoft researchers. It’s primary function  is to answer biomedical questions, and according to the American company, BioGPT even exceeds the knowledge level of human experts.

These types of models accelerate medical research for treatments and drugs, so its reliability is essential. It has two main uses:

• Question-Answering
• Entity extraction

As you know, Narrativa  develops solutions for the pharmaceutical sector, so that approvals and introductions to the marketplace  occur more quickly. We’ve tested this model and find the combination of Narrativa and Microsoft technology to be incredible!

How has BioGPT been trained?

Microsoft Research has created and trained this model solely and exclusively for the biomedical field specifically, biomedical papers—15 million PubMed abstracts—have been used.

The fine-tuning of the model of course also  surpasses all the other models created globally (e.g. GPT-3, FLAN) with 81% accuracy—the next  closest so far was 79%; impressive, right?

An amazing fact: of all versions of BioGPT, the most powerful has 1.5 Billion parameters!

When it comes to training models, more is not always better.

Why is BioGPT  a very good model at a technological level? Simply put, it’s because the model has been trained with very specific texts; it’s base of knowledge is highly precise. 

If we were going to train a model to know everything about cars, for instance, wouldn’t it be better to feed it millions of pieces of information  specifically about cars? Well the same applies here; the more tailored and discrete the source, the better the output will be. As the saying goes: garbage in, garbage out.  

Can I try it out myself?

Yes, Microsoft has released the model, though only for certain tasks. Check it out here!

Accelerating drug creation

Despite the utilization of AI in the healthcare industry, there are areas where the potential of this technology is underrepresented. Regulatory submission is one such area that could use AI and its subsets. Regulatory submissions, like clinical study reports, are based on information collected before, during, and after clinical trials. 

These data could come from many sources, including thousands of subjects across multiple regions and continents! But it doesn’t stop there;  someone needs to take these data, polish them, and then put them in a form for others to go through and generate the reports.

Regulatory Submissions Automation

Patient Safety Narratives Automation

Patient safety narratives are a key part of clinical study reporting. The automation of patient safety reports can help save time by automating the repetitive elements.

Automating Tables, Listings, and Figures

Narrativa’s TLF automation solution analyzes and summarizes data sets from a clinical study and transforms them into ready-to-use Tables, Listings, and Figures.

Automation of Clinical Study Reports (CSRs)

Narrativa Knowledge Graph® and deep learning language algorithms automate the generation of clinical study reports through data extraction and augmentation.

About Narrativa

Narrativa is an internationally recognized content services company that uses its proprietary artificial intelligence and machine learning platforms to build and deploy digital content solutions for enterprises. Its technology suite, consisting of data extraction, data analysis, natural language processing (NLP) and natural language generation (NLG) tools, all seamlessly work together to power a lineup of smart content creation, automated business intelligence reporting and process optimization products for a variety of industries.

Contact us to learn more about our solutions!