I sat down with Robin Röhm, CEO of Apheris AI to learn about his background, entrepreneurial journey, and mission. His company empowers life science and healthcare organizations to safely collaborate on data without needing to share the raw data.
“The core idea is to offer solutions such that companies can collectively analyze the data. We help pharma companies train drug discovery models and work with healthcare companies to draw out the value of their data in combinations with other companies,” Robin stated.
Robin has a background in medicine, philosophy, and mathematics. He is a serial entrepreneur and helped start Janus Genomics that builds AI tools to enable biomedical data sharing while preserving data privacy, and LoNova, which is an automated speech recognition company with a focus on children with developmental language disorders.
How does Apheris’ product work, and what can it bring to a biotech company?
“Imagine the biotech company that has already invested in data science. They've analyzed their own data and realized that they have a very biased view on data because their research was biased too. And so in order to get insights out of what data they already have, they need to train their models or run the analytics on data of other companies such as ones they may be partnered with. To assist them, we collaborate with the company to understand what type of collaboration partners they are thinking of.
Typically these are partnering companies, suppliers, or companies that already have business relationships with each other, but both of those companies own intellectual property in their data and they cannot share it. Based on this information, the data scientists of the biotech company can start to construct an analytics request that they are interested in executing. This is an analytical request that they want to perform on data that is not directly accessible,” Robin explains.
“Then they submit this analytics request to the Apheris platform. We test the request to ensure it is privacy-preserving by nature. In most cases, we protect it with cryptographic and data privacy technologies. We then share that protected analytics request to different data owners and executes it in their local environment. So we never moved the data. When the computation is finished and then the result of the computation is sent back to the data scientists of the biotech customer. We also ensure that these data scientists cannot reconstruct underlying data,” Robin stated.
New revenue sources for biotech startups and CROs
“We see a world where lots of companies own data relevant for pharma companies or biotech companies, but they're not using it yet. And for them, it might not be that valuable, but it's valuable for other third-party companies. Nevertheless, it contains IP. And so they are not willing to give it out because that would just harm their own business. And so what we offer them is pretty much a solution where a data owner can start to think about commercialization models on top of their data,” Robin said.
As many early-stage biotech companies with their own drug candidates have a hybrid business model. They offer contract research services to generate immediate revenue. Services like Apheris can open a new door for revenue generation for these companies if they can monetize their data without losing their intellectual property.
On-demand access to data will lead to R&D breakthroughs
“I believe access to data is the big revolution when we think about it. And I think what we see is that currently, access to high-quality data is the biggest impediment. Access to the right data will drive R&D breakthroughs across industries.”
Data as a commodity
Apheris AI aims to become the marketplace where data owners can monetize their data without losing any of their “secret recipes”and consumers can train their data model with others’ data. With this model, data will become a commodity. Now the question is how will you price the data? Will it be based on the value of the data, the cost of data generation, or something else?
“It's an incredibly big challenge. One of the questions is how much value do I generate in such a system? And that's a difficult question. We believe we have the solution, but we don’t yet have such a pricing scheme. I think we have to create more data points to start doing that. And as well, it's not just about the type of data, but obviously, it's all about the quality. And then it's about the ability to combine it. If you have genomics data in combination with patient data, both data sets alone should be priced very differently than in a collaborative setup. Value from connected data is often way more expensive," Robin explains.
Advice for entrepreneurs
In the early stages of the business, according to Robin, it is important to focus on the right problem and making sure there is enough willingness to pay. As the company already achieved some product-market fit, now his priorities are more on enabling knowledge sharing and collaboration among team members. He also encourages entrepreneurs to think big and align their communication so that it’s clear and simple.
“Entrepreneurship is about continuously willing to learn. As you build your company, new challenges pop up and grow in complexity.”
Want to be a speaker in the next Talk is Biotech episode? Apply here.