1Mby1M Virtual Accelerator AI Investor Forum: With Mo Islam, Partner at Threshold Ventures (Part 5)

Sramana Mitra: What have you been up to in the field of medical imaging? What are you learning? What insights can we convey here?

Mo Islam: We’ve backed a number of companies here. Imaging is one company that’s working well in medical imaging and diagnostics, with multiple FDA-approved devices now. VisAI is another company working on AI-powered stroke detection in the clinical workflow. These companies are more mature now, but in the early days, getting access to novel data sets was crucial.

For example, Imaging spun out of the Hospital for Special Surgery in New York, where they gained access to novel datasets to build their initial models. This access helped them secure FDA approval for their Class II medical devices, which was a significant milestone in their journey.

For entrepreneurs working in medical imaging and diagnostics, the models available now are quite amazing in terms of what’s publicly accessible. You can go to platforms like Hugging Face and download good general-purpose models, perhaps even imagery-focused, but not yet fine-tuned to medical imagery. With access to specific data, you can build something novel, whether it’s detecting cancer in X-rays or MRIs or stroke detection like VisAI.

There’s a lot of powerful open-source models available now, providing a significant advantage for those building in this area, but getting access to proprietary data is the big differentiator.

Sramana Mitra: But this kind of companies has to go through FDA approval, right?

Mo Islam: It depends on what you’re doing. In these cases, because it’s clinical decision support, they are actively providing radiologists or physicians with information for making a diagnosis. Therefore, it becomes critical to get FDA approval before you can start selling the product.

However, there are other companies doing different things where FDA approval isn’t necessarily required, such as in an R&D setting. We’ve seen companies go down that path. For these companies, we believe there is significant long-term value and defensibility if they can become FDA-approved devices.

Sramana Mitra: What is the timeline of the FDA approval?

Mo Islam: It depends on how mature the team is and how well they know the process. For a 510(k) submission, it usually takes about six to nine months from the initial submission, but there is often some back and forth in that process. It depends on what your device is, whether there are predicate devices that exist already, or if it’s a de novo submission.

It largely depends on what exactly you’re doing and how knowledgeable the team is about the process. Many of our companies have used FDA advisors to help them through the process, and it has been successful for them.

Sramana Mitra: So there is more knowledge of how to go through that process developing in the venture capital ecosystem. Now, the biotech world has always known how to do this, but this is coming out of the software investment world, right?

Mo Islam: A lot of us invest in healthcare companies, which are technically software companies since the product they’re building is largely software. However, it’s regulated like a medical device, so you need to have that expertise or at least a network that can provide it. This is something we’ve built up over time, along with having co-investors and partners who help the company with that.

In many of these companies, it’s been helpful to have a mix of dedicated healthcare and tech investors. I encourage any founder, especially in life sciences and other areas of healthcare, to have a mix of investors who can bring different expertise and networks to the table.

Sramana Mitra: Well, it’s been fascinating and we should continue the conversation in the future. It was a pleasure meeting. Thank you for coming.

This segment is part 5 in the series : 1Mby1M Virtual Accelerator AI Investor Forum: With Mo Islam, Partner at Threshold Ventures
1 2 3 4 5