Analyzing a Patent Thicket

Use Case: ORSERDU Patent Analysis

To demonstrate how Knowledge Finder can help to understand a patent thicket, I am going to investigate the recently approved breast cancer drug, Orserdu (elacestrant). To get started, I pulled up the drug’s patents using the FDA’s Orange Book database and downloaded the PDFs for each one using Google Patents.

Next, I uploaded each PDF to a collection in Knowledge Finder. To help give Knowledge Finder some additional context about intellectual property (if needed), I also uploaded a collection of literature reviews that talk about biopharma intellectual property. This gives me a nice collection of source and reference documents to quickly get me (and the chatbot) up to speed.

Composition of Matter

I’m interested in learning more about the patents that are protecting Orserdu's formulations and/or composition of matter. Knowledge Finder let's me use semantic search to query “composition of matter” in the patent collection (finding the most relevant sections of text in my collections) and then I ask the chatbot to summarize.

From the summary, it looks like the patent US10745343 is the most important in covering the formulations and/or composition of matter, so I decide to drill down on this document. With a click, I open this document in the individual document view and then ask Knowledge Finder to summarize:

As you can see, Knowledge Finder quickly identifies the name of the drug, its purported indication, and other important details.

I wanted to get a good idea of each claim within the patent, so next I asked Knowledge Finder to summarize these:

Knowledge Finder quickly parsed out each claim, and even grouped overlapping ones together with a common description.

Converting Chat to Tabular Output

Eventually, I know I will want to have these claims as cells in a spreadsheet. Thankfully, Knowledge Finder's chatbot can generate table-formatted outputs:

Next, I simply copy and paste the output into my Excel sheet:

Method of Use

Next I turn my attention to learning more about the data supporting the therapeutic potential of this drug in cancer. So I run another semantic search on “method of use” and find patent 10,071,066 titled “Method of Treating Cancer Using Selective Estrogen Receptor Modulators”. I ask Knowledge Finder to summarize the claims for me, but this time I drill deeper and ask it to tell me about the study data cited by the patent to back the therapeutic potential:

As you can see, Knowledge Finder quickly gives me the rundown on the experimental data provided in this patent suggesting potential efficacy in the target indication.

Closing Words

All of this work was completed in under an hour. Doing the same thing manually, without Knowledge Finder, would have taken days.

Of course, patents are just one of the many types of documents that you can use Knowledge Finder to distill large volumes of text into granular insights.