The study database contains the predicted outcome’s value and a series of variables selected as predictors for each patient.
The outcome to be predicted by our model is the presence or absence of postoperative complications in the first month after surgical intervention. Short-term postoperative complications were defined to be: sepsis; variceal hemorrhage; renal dysfunction; respiratory failure; disseminated intravascular coagulation; septic shock; multiple organ dysfunction syndromes; cardiac arrest; multiple systems organ failures; post-transplant lymphoproliferative disorder; biliary anastomosis stenosis—endoscopic stent; tumor recurrence, peritoneal carcinomatosis; HCV reinfection; graft infection with the hepatitis B virus; idiopathic transverse colon necrosis; bone and brain metastases; necrotizing pancreatitis; hepatic artery thrombosis; hemoperitoneum; primary non-functioning of the transplant graft; or common bile duct necrosis.
The following 14 clinical and laboratory pre-transplant parameters were collected and used as predictors: age, sex, blood type (ABO, RH), the diagnosis which prompted the need for liver transplantation (1—hepatitis C cirrhosis; 2—hepatitis C cirrhosis and HCC; 3—coinfection of HCV, hepatitis B virus and hepatitis D virus; 4—HCC associated with the coinfection of HCV, hepatitis B virus and hepatitis D virus), age at diagnosis, MELD-Na score, alpha-fetoprotein, pre-transplant antiviral treatment, liver re-transplantation, total bilirubin, platelet count, albumin, international normalized ratio, and the presence of ascites.
Zabara M.L., Popescu I., Burlacu A., Geman O., Dabija R.A., Popa I.V, & Lupascu C. (2023). Machine Learning Model Validated to Predict Outcomes of Liver Transplantation Recipients with Hepatitis C: The Romanian National Transplant Agency Cohort Experience. Sensors (Basel, Switzerland), 23(4), 2149.