Telepaxx today opens the first marketplace for the development, testing, quality assurance and commercial application of medical applications based on artificial intelligence in radiology. In line with the start of HIMSS 2019, StartUps and researchers will thus have access to high-quality training data and can offer certified solutions directly for use in the medical workflow.
Telepaxx has access to a network of more than 600 decentralized servers in medical facilities in German-speaking countries and a data pool of more than 13 billion image and findings data.
“The development of AI-based medical applications is growing rapidly, especially in the field of pattern recognition through machine learning with artificial neural networks of radiological image data. Researchers have demonstrated impressive successes in this field, almost on a weekly basis. However, researchers and developers often lack access to the market and to validated training data sets for the respective application cases,” says Thomas Pettinger, project manager and responsible for business development at Telepaxx.
The new offer now provides a solution with which such applications can be trained and developed to market maturity in a quality-assured and data protection-compliant environment. Pettinger explains the concept of the Telepaxx AI MarketPlace as follows: “Afterwards they would be available for commercial use to a large circle of customers.“
“With our offer, we are targeting startups and researchers at an early stage who want to develop applications for use in radiology,” says Rainer Kasan, partner at Telepaxx and experienced mastermind behind the idea of the AI MarketPlace.
In return, they can ensure that they will later be able to offer high-quality solutions in the AI Marketplace, explains Kasan further.
“In addition to the highest possible quality, it is just as important to integrate smoothly into the medical workflow of the facilities. We support our candidates for the Telepaxx AI MarketPlace right from the start not only with training data, but also with know-how. We clarify whether, for example, supervised learning can be used with smaller training data sets in the highest possible quality, but we also advise on quality assurance, the avoidance of bias or product approval and on a product design that fits in with the work processes in medicine and radiology,” says Rainer Kasan.