Research efforts to develop digital twins in medicine are rapidly increasing, with promising emerging applications in oncology, diabetes management and cardiovascular medicine. While medical digital twins hold great promise for personalized healthcare, their implementation is no easy feat. The field faces diverse challenges including collecting data, choosing computational model designs, ensuring safety and efficacy, and preventing biases. What can be learned from the successes achieved? Which are the most promising upcoming applications? We asked experts in the field for their thoughts.