AI for Allograft Diseases Diagnosis and Prognosis After Kidney Transplantation
Overview
- Phase
- Not Applicable
- Intervention
- Not specified
- Conditions
- Kidney Transplantation
- Sponsor
- Assistance Publique - Hôpitaux de Paris
- Enrollment
- 1500
- Locations
- 1
- Primary Endpoint
- Diagnostic model accuracy
- Status
- Withdrawn
- Last Updated
- 7 months ago
Overview
Brief Summary
Kidney transplantation is the treatment of choice for patients with end stage renal disease. One of the major challenges is to better diagnose the attacks undergone by the kidney transplant in order to increase its longevity. Multiple attacks are caused by non-immune and immune mechanisms, first and foremost the acute rejection of the transplant.
Biopsy, an invasive method, remains the "Gold Standard" for diagnosing rejection and other pathologies affecting the kidney transplant.
The invasive nature of these biopsies limits their use and alternative biomarkers have been evaluated in order to diagnose kidney transplant pathologies in a non-invasive manner. It is in this context that the nephrology and renal transplantation department of the Necker hospital and Inserm U1151 have carried out several studies leading to the identification of the diagnostic and prognostic potential of acute rejection, by the determination of urinary concentrations of two chemokines, CXCL9 and CXCL10.
The most recent study conducted within these teams demonstrated that the diagnostic potential of urinary chemokines could be improved by taking into account standard clinicobiological parameters in multiparametric models.
The main objective of the study is to develop, train and validate artificial intelligence models including urinary chemokines, efficient, robust, explainable and interpretable for the diagnosis and non-invasive prognosis of acute renal transplant rejection, trained on a data set made up of clinical and biological parameters.
Detailed Description
Kidney transplantation is the treatment of choice for patients with end stage renal disease. One of the major challenges is to better diagnose the attacks undergone by the kidney transplant in order to increase its longevity. Multiple attacks are caused by non-immune and immune mechanisms, first and foremost the acute rejection of the transplant. Biopsy, an invasive method, remains the "Gold Standard" for diagnosing rejection and other pathologies affecting the kidney transplant. The invasive nature of these biopsies limits their use and alternative biomarkers have been evaluated in order to diagnose kidney transplant pathologies in a non-invasive manner. It is in this context that the nephrology and renal transplantation department of the Necker hospital and Inserm U1151 have carried out several studies leading to the identification of the diagnostic and prognostic potential of acute rejection, by the determination of urinary concentrations of two chemokines, CXCL9 and CXCL10. The most recent study conducted within these teams demonstrated that the diagnostic potential of urinary chemokines could be improved by taking into account standard clinicobiological parameters in multiparametric models. The main objective of the study is to develop, train and validate artificial intelligence models including urinary chemokines, efficient, robust, explainable and interpretable for the diagnosis and non-invasive prognosis of acute renal transplant rejection, trained on a data set made up of clinical and biological parameters. For this, all the clinical parameters (demographic, medical history, characteristics of donors, immunosuppressive treatments, etc.) and biological (follow up of the usual biological parameters obtained as part of the routine care of transplant patients, urinary chemokines) of transplant patients followed in the nephrology and renal transplantation department of Necker hospital between 2004 and 2020, will be treated without a priori and by artificial intelligence methods.
Investigators
Eligibility Criteria
Inclusion Criteria
- •All renal transplant patients whose medical follow-up is provided by the nephrology and adult renal transplantation department of the Necker hospital between 2004 and 12/31/2020;
- •Patient having signed a consent form for the storage, use and transfer of samples taken during treatment, for scientific research purposes;
- •Patient not objecting to the processing of his personal data as part of the study.
Exclusion Criteria
- •\- A deceased patient who, during his lifetime, objected in writing to the processing of his data for research purposes.
Outcomes
Primary Outcomes
Diagnostic model accuracy
Time Frame: 3 years
ROC (receiver operating characteristic) curves AUC (Area under the Curve)
Prognostic model accuracy
Time Frame: 3 years
C-statistics
Secondary Outcomes
- Strenght of the models(3 years)