FDA approval to market in the US
For the technique to succeed, CARLO needed to learn to see. And AOT faced a further challenge – they were preparing to seek FDA approval to market the device in the US. Zühlke supported AOT with experts in machine learning, data engineering, software development and regulatory requirements.
Together, Zühlke and AOT developed a medical machine learning process. Building on this, the team then implemented and validated a legally compliant data platform for data collection and developing medical machine learning models. Finally, the team implemented an initial image recognition application, for which the new platform proved invaluable. This initial proof of concept involved enabling the robot to use optical coherence tomography to orient itself and precisely locate the bone and cutting sites – it gave CARLO vision.
Data platform to form the foundation for future software applications
For AOT, this result was a major milestone on the path to commercialisation. And working with Zühlke has had other benefits – the legally compliant data platform accelerates AOT’s ability to develop further medical software applications for a wide range of indications. The data platform also facilitates the FDA approval process. Good news not just for AOT, but above all for the patients who will benefit from new and improved therapies enabled by the device.
Dr. Gabriel Krummenacher
Gabriel Krummenacher leads the Data Science Team at Zühlke and has several years of experience in conducting data analytics and machine learning projects. His main focus is on medical machine learning applications and bringing prototypes to production. He holds a PhD and M.Sc. from the Institute for Machine Learning at ETH, where he worked on scalable methods for large-scale and robust learning, wheel defect detection and sleep stage prediction with deep learning.