AI promises to transform healthcare by not only improving clinical workflows and patient outcomes but also equitably distributing the benefits to everyone, everywhere. To live up to these promises, AI implementations must be informed by the needs of clinicians, patients, healthcare providers as well as legal policies, and thoroughly validated with a clinical mindset. Only then, the state-of-the-art AI models can be conceived as peers (or co-pilots) that train differently and make algorithmic decisions based on sophisticated analyses of data.

Building upon the vision of algorithmic peers in medicine, I develop principled machine learning approaches that exploit diverse data committed to healthcare repositories while prioritizing the safety and well-being of their users and society. I work closely with medical experts in order to understand their needs and challenges, distill them into precise learning problems, solve them by efficient and interpretable methods and ultimately validate the solutions in downstream scenarios.