Establishing Medical Intelligence - Leveraging FHIR to Improve Clinical Management: a retrospective cohort and clinical implementation study
Alexander Brehmer,Christopher Martin Sauer,Jayson Salazar Rodríguez,Kelsey Herrmann,Moon Kim,Julius Keyl,Fin Hendrik Bahnsen,Benedikt Frank,Martin Koehrmann,Tienush Rassaf,Amir-Abbas Mahabadi,Boris Hadaschik,Christopher Darr,Ken Herrmann,Susanne Tan,Jan Buer,Thorsten Brenner,Hans Christian Reinhardt,Felix Nensa,Michael Gertz,Jan Egger,Jens Kleesiek
DOI: https://doi.org/10.2196/55148
2024-08-20
Abstract:Background: FHIR (Fast Healthcare Interoperability Resources) has been proposed to enable health data interoperability. So far, its applicability has been demonstrated for selected research projects with limited data. Objective: Here, we designed and implemented a conceptual medical intelligence framework to leverage real-world care data for clinical decision-making. Methods: A Python package for the utilization of multimodal FHIR data (FHIRPACK) was developed and pioneered in five real-world clinical use cases, i.e., myocardial infarction (MI), stroke, diabetes, sepsis, and prostate cancer (PC). Patients were identified based on ICD-10 codes, and outcomes were derived from laboratory tests, prescriptions, procedures, and diagnostic reports. Results were provided as browser-based dashboards. Results: For 2022, 1,302,988 patient encounters were analyzed. MI: In 72.7% of cases (N=261) medication regimens fulfilled guideline recommendations. Stroke: Out of 1,277 patients, 165 patients received thrombolysis and 108 thrombectomy. Diabetes: In 443,866 serum glucose and 16,180 HbA1c measurements from 35,494 unique patients, the prevalence of dysglycemic findings was 39% (N=13,887). Among those with dysglycemia, diagnosis was coded in 44.2% (N=6,138) of the patients. Sepsis: In 1,803 patients, Staphylococcus epidermidis was the primarily isolated pathogen (N=773, 28.9%) and piperacillin/tazobactam was the primarily prescribed antibiotic (N=593, 37.2%). PC: Three out of 54 patients who received radical prostatectomy were identified as cases with PSA persistence or biochemical recurrence. Conclusions: Leveraging FHIR data through large-scale analytics can enhance healthcare quality and improve patient outcomes across five clinical specialties. We identified i) sepsis patients requiring less broad antibiotic therapy, ii) patients with myocardial infarction who could benefit from statin and antiplatelet therapy, iii) stroke patients with longer than recommended times to intervention, iv) patients with hyperglycemia who could benefit from specialist referral and v) PC patients with early increases in cancer markers.