This system facilitates matchings of donors and recipients in hematopoietic stem cells transplantation for malignant blood diseases.
When transplanting hematopoietic stem cells （HSC） for malignant blood diseases, serious side effects such as graft-versus-host disease (GVHD) may occur depending on the combination of donor and recipient.
However, it is difficult to select the best combination and predict prognosis because there is a vast amount of information on bone marrow donors, cord blood donors and peripheral blood donors and need years of clinical experience.It is difficult to select the best transplant source in many cases depending on pychisians’ clinical experience.
NextGeM developed a companion diagnostic system that enables the selection of the best donor candidate by simulating the prognosis for each individual donor candidate based on transplant donors’ information from cord blood banks and bone marrow banks and data and pretreatment information from patient or recipient.
We built a machine learning model that learns from approximately 2,500 patients at Kyoto University group hospitals over different time series. Compared with the existing COX proportional hazards model, it has achieved improved accuracy in predicting graft relapse-free survival（GRFS：GVHD-free, Relapse-Free Survival) after one year.
The system is deployed in a cloud server and is used as software within a medical institution. An internet connection is required.
・Citations and References
Published in Blood Advances “Establishment of a Predictive Model for GvHD-free, Relapse-free Survival after Allogeneic HSCT using Ensemble Learning「Establishment of a Predictive Model for GvHD-free, Relapse-free Survival after Allogeneic HSCT using Ensemble Learning」