A convolutional neural network-based model that predicts acute graft-versus-host disease after allogeneic hematopoietic stem cell transplantation

• Published in: Communications medicine Vol. 3; no. 1; pp. 67 - 11
• Main Authors: Jo, Tomoyasu, Arai, Yasuyuki, Kanda, Junya, Kondo, Tadakazu, Ikegame, Kazuhiro, Uchida, Naoyuki, Doki, Noriko, Fukuda, Takahiro, Ozawa, Yukiyasu, Tanaka, Masatsugu, Ara, Takahide, Kuriyama, Takuro, Katayama, Yuta, Kawakita, Toshiro, Kanda, Yoshinobu, Onizuka, Makoto, Ichinohe, Tatsuo, Atsuta, Yoshiko, Terakura, Seitaro
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 16.05.2023; Springer Nature B.V; Nature Portfolio
• Subjects: 631/114; 631/532/1542; 692/699/1541; 692/699/249/1529; Algorithms; Antigens; Computer centers; Graft versus host disease; Hematology; Machine learning; Medicine; Medicine & Public Health; Natural language; Neural networks; Patients; Stem cell transplantation
• ISSN: 2730-664X; 2730-664X
• DOI: 10.1038/s43856-023-00299-5

Background Forecasting acute graft-versus-host disease (aGVHD) after allogeneic hematopoietic stem cell transplantation (HSCT) is highly challenging with conventional statistical techniques due to complex parameters and their interactions. The primary object of this study was to establish a convolutional neural network (CNN)-based prediction model for aGVHD. Method We analyzed adult patients who underwent allogeneic HSCT between 2008 and 2018, using the Japanese nationwide registry database. The CNN algorithm, equipped with a natural language processing technique and an interpretable explanation algorithm, was applied to develop and validate prediction models. Results Here, we evaluate 18,763 patients between 16 and 80 years of age (median, 50 years). In total, grade II–IV and grade III–IV aGVHD is observed among 42.0% and 15.6%. The CNN-based model eventually allows us to calculate a prediction score of aGVHD for an individual case, which is validated to distinguish the high-risk group of aGVHD in the test cohort: cumulative incidence of grade III–IV aGVHD at Day 100 after HSCT is 28.8% for patients assigned to a high-risk group by the CNN model, compared to 8.4% among low-risk patients (hazard ratio, 4.02; 95% confidence interval, 2.70–5.97; p  < 0.01), suggesting high generalizability. Furthermore, our CNN-based model succeeds in visualizing the learning process. Moreover, contributions of pre-transplant parameters other than HLA information to the risk of aGVHD are determined. Conclusions Our results suggest that CNN-based prediction provides a faithful prediction model for aGVHD, and can serve as a valuable tool for decision-making in clinical practice. Plain language summary Hematopoietic stem cell transplantation (HSCT) is a procedure used in patients to reestablish blood cell production. It involves the transplant of cells from a donor to the patient. In some patients the transplanted cells damage cells within the patients. This is called graft-versus-host disease (GVHD). We developed a computational code that can predict the likelihood a person will develop GVHD soon after HSCT. Using this computer program will enable doctors to better identify those at risk of GVHD and initiate treatments when required. Jo et al. establish a convolutional neural network-based model to predict acute graft-versus-host disease (aGVHD) after allogeneic hematopoietic stem cell transplantation. The model both predicts aGVHD and identifies pre-transplant parameters that increase risk of aGVHD.