Article Text

Download PDFPDF

902 A generative artificial intelligence engine for tertiary research and predictive modeling in immunotherapy
  1. Sanat Mohanty1,
  2. Vaijayanti Gupta2,
  3. Omkar Patil3 and
  4. Abhishek Mamdapure3
  1. 1Rockville, MD, USA
  2. 2G-KnowMe, Bangalore, India
  3. 3Wynum Automation, Pune, India
  • Journal for ImmunoTherapy of Cancer (JITC) preprint. The copyright holder for this preprint are the authors/funders, who have granted JITC permission to display the preprint. All rights reserved. No reuse allowed without permission.


Background We describe the building of Knolens as a tool for generative tertiary research and some examples of its use in analysis of drug development by therapeutic area and mechanism of action. The tools allows for quasi real time access to updated data across accessible document libraries as well as rapid analysis of trends and correlations or identification of new insights or outliers. Through interactive changes in assumptions, it allows for prediction and analysis of scenarios that impact decisions around target analysis, clinical study design, regulatory decisions or business development strategies to be assessed.

Methods Knolens uses a fine tuned Large Language Model to acquire data from unstructured documents including publications, conference proceedings, regulatory agency documents, patent filings, press releases among others. The acquired data assesses whether there occurs any inconsistencies in embeddings, to filter what is referred to as hallucinations. This dataset is used to generate a knowledge graph spanning drug discovery, translational research, clinical trials, regulatory data, post market studies, business and financial data as well as intellectual property. Modules specific to functional domains are built on the knowledge graph with interactive UIs to analyze the data for insights, such as ones analyzing progress in a specific Mechanism of Action (MoA).

Results As a case study, we compare all approved drugs and clinical studies for for JAK inhibitors, overall and by some of the more popular therapy areas where this MoA has been deployed. The tool assesses the progress of drug development by phase of trial and choice of comparators used in these studies. The analysis also scrutinizes endpoints used in these studies and their evolution by disease area and therapy. The results include performance assessment and validation of the tool, by the time needed for generation of content and accuracy. We show that the tool has an accuracy of >99% in the data it presents and missed data being less than 1%.

Conclusions Knolens as a tool provides researchers across functions a reliable AI based solution to rapidly acquire and analyze data across functional domains and databases.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.