I couldn’t have asked for a better partner than Dr. Victoria Hall who co-led this work with me. Thanks to the Ajmera Transplant Centre and PSI Foundation for their funding support, to Dr. Lavoie & his team, our UHN colleagues, and to the participants whose generosity made this work possible.

Thanks to Dr. Ghazal Azarfar for her hard work on this study, and Dr. Mamatha Bhat, the UHN Transplant AI Initiative and fellow co-authors for their support.

Traditional and deep learning architectures were compared. "Routed LSTM", which combines capsule networks with LSTM to reduce the need for larger datasets, performed best (0.02 MSE and 0.79 PCC).

Key predictors included age, BMI, immunosuppression, breakthrough infection and transplant type.