Founded in 2021 by Simona Santamaria, Kathrin Khadra, and Jonas Ils, the startup Ryver.ai aims to solve the data shortage for medical AI without compromising patient privacy. The company develops generative AI to create artificial, yet highly realistic datasets of radiological images. According to Studies AI-assisted diagnostics in radiology produce significantly poorer results, including in ethnic minorities. The reason for this is a severe lack of diverse training and test data. Ryver.ai aims to change this: Generative models are designed to enable medical AI providers to generate diverse test and training data quickly and cost-effectively.
Fresh capital for technical development
Co-founder and CTO Kathrin Khadra explained:
"Our generative models understand the characteristics of radiological images as well as the subtle differences between patient groups, scanners, and clinical pictures. Based on this understanding, a multitude of entirely new images can be generated that mimic the real world. Since this synthetic data is essentially fictitious and not directly linked to real patients, the data protection risk is minimized. We thoroughly examine both data quality and data protection. To achieve this, we combine mathematical methods with the expert opinion of radiologists."
The current financing round was led by Nina Capital, a Spanish health-tech venture capital firm. Bayern Kapital and FundF are also participating as investors. Regarding the intended use of the newly raised capital, co-founder Simona Santamaria:
"The capital will primarily flow into technical development. We will continue to hire talented AI developers and invest in computing capacity to ensure the highest image quality."
“Innovative solution for a massive problem”
And Marta Gaia Zanchi, Managing Partner of Nina Capital, says:
"Most medical image datasets are skewed: they poorly represent underserved communities, new devices, and rare diseases. Using skewed datasets for AI development ultimately impacts the reliability of AI in clinical applications: algorithms that search for low-prevalence conditions have significantly lower positive predictive value than those with higher prevalence, and the AI can drift over time. Ryver.ai's synthetic data generation technology is an innovative solution to this massive problem. Ryver.ai's diverse founding team not only has the expertise and commitment to tackle such a technical challenge, but also a clear guiding light in the pursuit of reliable and affordable medical AI."
