Sramana Mitra: Were you able to get that approval?

Joga Gobburu: Yes. We got the consensus from the FDA, and they agreed to this preemptive marker. It’s underway already.

Sramana Mitra: My next question in the context of this case study is, what is it that you know from the adult trial that can be applied to the pediatric trial? What are some of the parameters that you could model?

Joga Gobburu: For any type of disease, a vast number of measures are collected both to monitor desired effects as well as undesired effects. What we do not know is the interplay between these measures. There’s a first step, and that leads to the second step. That mother nature is a bit complex and shy. It doesn’t give you all the information right away. Through these clinical trials, that information is inherently hidden in the data collected.

Typically, scientists don’t spend enough time unraveling this interplay that leads to the long-term effect because the data are very complex. The software tools are outdated as well. For example, a scientist needs to master three to four software packages. They don’t speak to each other. I have to run something on software one. Then I need to make it amenable to be read on software two. I may have to come back to software one and I may need to change the format. It’s a cumbersome process.

If you have to deal with thousands of patients and years of data, you’re talking about an enormous number of data points. That is the biggest hurdle. That is one of the impetus for coming up with Pumas. It makes the scientific analysis simple for scientists so they can think about things.

Sramana Mitra: In the technology that you’re using, is that a machine learning technology?

Joga Gobburu: Pumas is a suite of tools. You can perform a variety of data analysis within the same software. You can perform scientific modeling which has underpinning in pharmacology. Then you can also perform statistical modeling. You can also perform Machine Learning.

We push the limits. We just introduced a new proprietary technology which is called DPumas. A scientist can use a range of data analytics techniques on the same platform. They don’t have to step out to another software. It’s integrated. Scientists are forced to choose scientific modeling and machine learning. We said, “Why do you have to choose?” We are going to bring them together. Nobody has done this before.

Now if you have a complex problem. Interrelationships between these measures are very complex. We understand some parts, and some we don’t. For the parts that we understand, you could use the scientific data analytical approach. For the other part, you let the machine learning part figure it out. They are done in a hybrid manner. The beauty is the deep learning gives you back the biological model that now you can complete your analysis.

Sramana Mitra: So you have come up with insights using these machine learning models as well.

Joga Gobburu: Not in the case study I explained but for another one. For that one, data is heterogeneous. Everything was used for the analysis.

Sramana Mitra: How many jobs are going through your software right now?

Joga Gobburu: There are several pharmaceutical companies that are using Pumas. Each of them have multiple assets that they’re developing. I wouldn’t say it’s a very big number.

Sramana Mitra: What is your estimate per drug of the cost and time saving?

Joga Gobburu: About 50% to 60% time saving for the scientist.

Sramana Mitra: And cost?

Joga Gobburu: It’s a good question. Cost comes in multiple ways. For example, IT overhead. Pumas is deployed both on cloud and desktop. In the cloud, it’s deployed on Julia Hub. That is a fully compliant system. That is a 60% savings there in terms of IT overhead. You would recuperate the cost in 18 to 24 months by switching to Pumas.

There are other benefits. Now you don’t have to train your scientists on four software packages. You don’t have to maintain four packages. You can gain insights that scientists are unable to with the current tools. That is enormous. We don’t even know how to quantify that. Those are the vectors.