Predicting Donor Selection and Multi-Organ Transplantation within Organ Procurement Organizations Using Machine...

Area of Science:

• Transplantation Science
• Medical Informatics
• Public Health

Background:

• Organ procurement organizations (OPOs) are vital for organ transplantation but face transparency challenges in organ allocation.
• Lack of transparency in OPOs can hinder effectiveness and raise ethical concerns regarding organ distribution.

Purpose of the Study:

• To analyze organ procurement and donor selection strategies within OPOs using machine learning.
• To predict donor selection and the likelihood of multi-organ suitability for transplantation.

Main Methods:

• Utilized the ORCHID dataset (133,101 organ donor referrals).
• Developed seven machine learning classification models.
• Evaluated model performance using recall and AUC metrics for donor selection and multi-organ transplant prediction.

Main Results:

• Machine learning models achieved high performance (AUC > 0.95) for donor selection and multi-organ transplant prediction (AUC > 0.97).
• Recall values for donor selection ranged from 0.62 to 0.80.
• Recall values for multi-organ transplant prediction were high, ranging from 0.88 to 0.98.
• Identified administrative milestones and specific organ transplants as key determinants in organ allocation.

Conclusions:

• Machine learning effectively predicts donor selection and multi-organ transplant potential.
• Optimizing administrative practices presents significant opportunities to improve organ allocation strategies.
• Findings highlight the substantial impact of administrative processes on transplantation success rates.