At Innovation Sprint we learn models for patients and we build trust for them by validating their performance both on the technical side, but also on the medical side, collaborating with healthcare professionals at the field. Our models are both predictive, aiming at discovering digital composite biomarkers, and generative, describing patient groups and synthesizing data.
Our learning pipelines follow good clinical practice and GDPR guidelines from data collection to learning and application. Patients are in control of their data and healthcare professionals dictate what is important to learn and are convinced on the models’ usefulness being in the validation loop and understanding models’ decisions via explainable AI techniques.
We strongly believe that the way we are following good clinical practice in our model learning pipelines will change how we build trust to our AI models in Life Sciences and MedTech.