The Story of Gradient Health: Building the Future of Medical AI Data

The Story of Gradient Health: Building the Future of Medical AI Data

Sometimes the best startups begin with someone falling down. A lot.

In a recent Category Visionaries episode, Joshua Miller shared how Gradient Health’s journey started with his co-founder repeatedly falling and recording it: “He was just full sending it, man, it was a champ.” The goal? Generate training data for an AI-powered fall detection system for elderly care.

But this crash-test approach to data collection revealed a bigger problem: the scarcity of medical data for AI development. “If you’re an AI developer,” Joshua explains, “and you’re building an AI to detect, let’s say, liver cancer, you can’t just call up a hospital and be like, hey, can I have all of your scans of people’s livers? That’s a weird and kind of creepy question to ask.”

The solution? Work with hospitals worldwide to strip patient information from medical images and make them available for research. It’s a critical bridge between healthcare institutions sitting on valuable data and AI companies needing that data to build potentially life-saving technologies.

Interestingly, this wasn’t entirely new territory for Joshua. His previous company, Farm Shots, used satellite imagery to detect crop diseases. “The problem of finding a disease in a satellite image is, from a computer’s perspective, the same as finding a tumor in a liver,” he notes. “You’re trying to find something that shouldn’t be there in a picture.”

Landing their first hospital partnership proved challenging, but an angel investor from Venture South connected them with Cone Health. This first partnership created the momentum they needed, leading to what Joshua describes as an unusual situation for a startup: “We have Fortune 500 companies sending us emails. Gradient is a 15 person company. It’s kind of unheard of in the startup world.”

Their recently launched self-service platform gained 130 users in just 30 days, and the emergence of foundation models in radiology has dramatically expanded their potential market. As Joshua observes, “If you asked me six months ago, could you build a company in a large company in imaging alone, I probably would have said no. But now that we start to see people build these ultra large scale models imaging, I would make the argument that now you can.”

Looking ahead, Gradient Health’s vision is remarkably straightforward: “For anyone who is trying to do something good for humanity with medical AI gets the data they need within 24 hours for reasonable.” It’s a vision that could fundamentally change how medical AI is developed, making it faster and more accessible for companies to build potentially life-saving technologies.

This transformation of medical data access might also benefit patients directly. Joshua envisions a future where patients could get discounts on medical procedures by allowing their anonymized data to be used in research: “If I can get $100 off that ultrasound by saying, look, you can use my data and research, I think that’s a perfect and beautiful world.”

From recording falls to revolutionizing medical data access, Gradient Health’s journey shows how sometimes the biggest opportunities come from recognizing that your initial problem is just a symptom of a much larger one.