Biomaterial Database

Machine Learning Prediction Model for Neutrophil Recovery after Unrelated Cord Blood Transplantation. 2024

Yachiyo Kuwatsuka, and Rika Kasajima, and Rui Yamaguchi, and Naoyuki Uchida, and Takaaki Konuma, and Masatsugu Tanaka, and Naoki Shingai, and Shigesaburo Miyakoshi, and Yasuji Kozai, and Yasufumi Uehara, and Tetsuya Eto, and Masako Toyosaki, and Tetsuya Nishida, and Fumihiko Ishimaru, and Koji Kato, and Takahiro Fukuda, and Seiya Imoto, and Yoshiko Atsuta, and Satoshi Takahashi

Department of Advanced Medicine, Nagoya University Hospital, Nagoya, Japan.

BACKGROUND Delayed neutrophil recovery is an important limitation to the administration of cord blood transplantation (CBT) and leaves the recipient vulnerable to life-threatening infection and increases the risk of other complications. OBJECTIVE A predictive model for neutrophil recovery after single-unit CBT was developed by using a machine learning method, which can handle large and complex datasets, allowing for the analysis of massive amounts of information to uncover patterns and make accurate predictions. METHODS Japanese registry data, the largest real-world dataset of CBT, was selected as the data source. Ninety-eight variables with observed values for more than 80% of the subjects known at the time of CBT were selected. Model building was performed with a competing risk regression model with lasso penalty. Prediction accuracy of the models was evaluated by calculating area under curve (AUC) using a test dataset. The primary outcome was neutrophil recovery at day 28 (D28), and D14 and D42 were analyzed as secondary outcomes. RESULTS The final cord blood engraftment prediction (CBEP) models included 2,991 single-unit CBT recipients with acute leukemia. Median AUC of a D28-CBEP lasso regression model run 100 times was 0.74, and those of D14 and D42 were 0.88 and 0.68, respectively. The D28-CBEP model predictivity was higher than four different legacy models that were separately constructed. CONCLUSIONS A highly predictive model for neutrophil recovery by 28 days after CBT was constructed using machine learning techniques. However, identification of significant risk factors was insufficient for outcome prediction for an individual patient, which is necessary for improving therapeutic outcomes. Notably, the prediction accuracy for days 14, 28, and 42 post-transplant decreased, and the model became more complex with more associated factors with increased time after transplantation.