Research Project
Project Title: |
Searching unsuspected inherited kidney diseases: a needle in a haystack (NLP) |
Project Type: |
Translational research |
Disease group(s): |
Hereditary glomerulopathies, Tubulopathies, Metabolic & stone disorders, Thrombotic microangiopathies, AD structural kidney disorders |
Project Summary: |
Objectives: Primary: Identify patients with CKD (chronic kidney disease) and potential HKD (hereditary kidney disease) not previously diagnosed, through Natural Language Processing (NLP) methods from electronic medical records. Secondary: A) evaluate the clinical validity of the automated extraction of HPO (human phenotype ontology) terms (plus added terms) from narratives of electronic medical records for the identification of patients with HKD. B) analyze previous incorrect diagnoses of patients with HKD not previously diagnosed. C) Determine the percentage of patients with CKD and IKD from those on renal replacement therapy and diagnosed with "nephropathy of unknown origin". D) Determine the percentage of patients with IKD out of those with a diagnosis of hypertensive nephropathy, unspecified immune glomerulopathy, diabetic nephropathy, and chronic interstitial nephropathy. Methodology: Artificial Intelligence technology will be used to identify patients with undiagnosed IKD and CKD. Initially, a curation of the HPO terms associated with each HKD will be carried out. With this list of “HKD-terms”, the medical records of all patients with CKD controlled in the last year in each of the participating centers will be analyzed using NLP methods. Information captured in patients' electronic health records (EHRs) will be exported to a standardized database. The list of patients with potential IKD will be analyzed at a clinical level and a genetic study will be carried out in those cases in which it is considered indicated. |
Lead principal investigator(s): |
Roser Torra, Barcelona |
Project Period: |
01/2023 - 12/2025 |
Sponsors: |
National funding agency |
« Back to research page