Bayesian modeling of pre-transplant variables accurately predicts kidney graft survival
An embodiment of the invention provides a method for determining a patient-specific probability of renal transplant survival. The method collects clinical parameters from a plurality of renal transplant donor and patient to create a training database. A fully unsupervised Bayesian Belief Network model is created using data from the training database; and, the fully unsupervised Bayesian Belief Network is validated. Clinical parameters are collected from an individual patient/donor; and, such clinical parameters are input into the fully unsupervised Bayesian Belief Network model via a graphical user interface. The patient-specific probability of disease is output from the fully unsupervised Bayesian Belief Network model and sent to the graphical user interface for use by a clinician in pre-operative organ matching. The fully unsupervised Bayesian Belief Network model is updated using the clinical parameters from the individual patient and the patient-specific probability of transplant survival.
1. A method for determining a patient-specific probability of renal transplant survival, said method including:
a) collecting clinical parameters from a plurality of patients and donor to create a training database, the clinical parameters;
b) creating a fully unsupervised Bayesian Belief Network model using data from the training database;
c) validating the fully unsupervised Bayesian Belief Network model;
d) collecting the clinical parameters for an individual patient and an donor;
e) receiving the clinical parameters for the individual patient and an donor into the fully unsupervised Bayesian Belief Network model;
f) outputting the patient-specific probability of transplant survival from the fully unsupervised Bayesian Belief Network model to a graphical user interface for use by a clinician; and
g) updating the fully unsupervised Bayesian Belief Network model using the clinical parameters for the individual patient and for the donor, and the patient-specific probability of transplant survival.
2. The method according to claim 1 , wherein the patient clinical parameters include a plurality of the following: age of diabetes onset; age at time of transplant; Body Mass Index (BMI); indication of creatinine decline by 25% or more in the first 24 hours following transplant; the primary diagnosis; type of diabetes; dialysis type at time of listing; number of days on dialysis pre-transplant; an indication of return to dialysis with the first week following transplantation; length of graft survival; graft loss due to thrombosis; induction medication administered; type of transplantation procedure applied; use of maintenance or anti-rejection medications post-transplant; result of Human Leukocyte Antigen (HLA) testing; most recent serum creatinine prior to discharge following transplantation procedure; most recent absolute creatinine at time of listing; most recent Panel Reactive Antibody (PRA) in the USRDS; requirement of pretransplant dialysis; blood type; recurrent disease in graft; race; gender; serum creatinine at time of transplant; and drug treated systemic hypertension at time of listing.
3. The method according to claim 1 , wherein the donor clinical parameters include: cardiac arrest since neurological event that led to declaration of brain death; total cold ischemia time; blood type; age; BMI; cause of death; serum creatinine; history of diabetes; history of hypertension; any sue of cigarettes; indication of donation after cardiac death (DCD); drug use; race; gender; organ warm ischemic time for DCD; cocaine use specifically; use of perfusion pump; and the total warm ischemia and anastomotic time.
4. The method according to claim 1 , wherein said creating of the fully unsupervised Bayesian Belief Network model includes creating the fully unsupervised Bayesian Belief Network model without human-developed decision support rules.
5. The method according to claim 1 , further including estimating an accuracy of the patient-specific probability of transplant survival, the accuracy including at least one of model sensitivity, model specificity, positive and negative predictive values, and overall accuracy.