We implemented random forest classifier for the binary classification of in-hospital mortality. A total of 2,553 cases of ST-elevation AMI were assigned to 80% training subset for cross validation and 20% test subset for model performance evaluation. Here, we developed an online application tool using a machine learning model to predict in-hospital mortality in patients with AMI. Although several studies have used machine learning techniques for prognostic prediction of patients with AMI, its clinical application has not been achieved. “Despite performing well in predicting kidney end points in our own cohort, these risk scores were not superior to the TIMI Risk Score for Kidney Disease Progression in Type 2 Diabetes,” they stress.Commonly used prediction methods for acute myocardial infarction (AMI) were created before contemporary percutaneous coronary intervention was recognized as the primary therapy. However, the current model differs in that it was developed using people with type 2 diabetes rather than a general population, predictions were broader and not limited to an eGFR below 60/mL/min per 1.73 m 2 and focused on an earlier outcome rather than kidney failure. Sabatine et al acknowledge that other kidney disease risk scores exist, such as the kidney failure risk equation, the chronic kidney disease (CKD) Prognosis Consortium risk equation for incident CKD, and the CKD Prognosis Consortium risk model for decline in kidney function. However, the investigators note that the absolute risk reductions increased significantly in line with the risk categories (0.2%, 1.0%, and 3.5%, respectively), showing that patients “with a higher baseline risk of kidney disease progression experience a greater magnitude of benefit from inhibition.” Relative risk reductions for kidney disease progression with dapagliflozin were generally consistent across the higher risk categories (15.0%, 49.0%, and 47.0% respectively). The team also assessed reductions in the risk for kidney disease progression with dapagliflozin according to predicted 4-year risk (low <1%, intermediate 1–4%, and high ≥4%) among included patients from the DECLARE-TIMI 58 trial. Moreover, good calibration was demonstrated, with the 2-year cumulative incidence of kidney disease progression falling within the predefined 2-year predicted risk categories of low (<0.5%), intermediate (0.5 to <2.0%), and high (≥2.0%). Sabatine et al point out that the same level of discrimination (0.798) was also seen when the tool was used in the remaining 12,362 (30%) patients – the validation cohort – whose baseline characteristics were very similar to those of the derivation cohort.Įach of the individual components of the model showed similarly good discrimination and the tool was able to predict kidney disease progression whether the patients had baseline eGFR above or below 60 mL/min per 1.73 m 2, and in those with and without baseline albuminuria, the researchers report. In addition, the cohort had a median eGFR of 76 mL/min per 1.73 m 2, 77.6% of patients had preserved kidney function (eGFR ≥60 mL/min per 1.73m 2), and a median UACR of 13.3 mg/g (67.0% had normal or mildly increased UACR <30 mg/g). At baseline, the patients (34% women, 21% non-White) had a median age of 64 years, a median HbA1c of 7.5% (58 mmol/mol), a median diabetes duration of 9.2 years, and 36% were receiving insulin therapy. The final risk prediction tool – The TIMI Risk Score for Kidney Disease Progression in Type 2 Diabetes – showed good discrimination when used in 28,842 (70%) patients comprising the derivation cohort, with a Harrell c-index of 0.798. The patients were followed up for a median of 2.4 years, during which time kidney disease progression occurred in 481 patients. These were selected by multivariable Cox regression from a possible 23 variables assessed in 41,204 patients with type 2 diabetes from four Thrombolysis in Myocardial Infarction (TIMI) trials – DECLARE-TIMI 58, SAVOR TIMI 53, FOURIER (TIMI 59), and CAMELLIA-TIMI 61. The eight candidate risk factors are baseline eGFR, urine albumin-to-creatinine ratio (UACR), hemoglobin and glycated hemoglobin (HbA1c) levels, duration of type 2 diabetes, systolic blood pressure, atherosclerotic cardiovascular disease, and heart failure.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |