A fully interpretable machine learning model for increasing the effectiveness of urine screening.

OBJECTIVES: This article addresses the need for effective screening methods to identify negative urine samples before urine culture, reducing the workload, cost, and release time of results in the microbiology laboratory. We try to overcome the limitations of current solutions, which are either too simple, limiting effectiveness (1 or 2 parameters), or too complex, limiting interpretation, trust, and real-world implementation ("black box" machine learning models). METHODS: The study analyzed 15,312 samples from 10,534 patients with clinical features and the Sysmex Uf-1000i automated analyzer data. Decision tree (DT) models with or without lookahead strategy were used, as they offer a transparent set of logical rules that can be easily understood by medical professionals and implemented into automated analyzers. RESULTS: The best model achieved a sensitivity of 94.5% and classified negative samples based on age, bacteria, mucus, and 2 scattering parameters. The model reduced the workload by an additional 16% compared to the current procedure in the laboratory, with an estimated financial impact of €40,000/y considering 15,000 samples/y. Identified logical rules have a scientific rationale matched to existing knowledge in the literature. CONCLUSIONS: Overall, this study provides an effective and interpretable screening method for urine culture in microbiology laboratories, using data from the Sysmex UF-1000i automated analyzer. Unlike other machine learning models, our model is interpretable, generating trust and enabling real-world implementation.

Comparison of C-reactive protein and albumin excretion as prognostic markers for 10-year mortality after myocardial infarction.

BACKGROUND: C-reactive protein (CRP) is an established prognostic marker in the setting of acute coronary syndromes. Recently, albumin excretion rate also has been found to be associated with adverse outcomes in this clinical setting. Our aim was to compare the prognostic power of CRP and albumin excretion rate for long-term mortality following acute myocardial infarction (AMI). HYPOTHESIS: To determine whether albumin excretion rate is a better predictor of long-term outcome than CRP in post-AMI patients. METHODS: We prospectively studied 220 unselected patients with definite AMI (median [interquartile] age 67 [60-74] y, female 26%, heart failure 39%). CRP and albumin-to-creatinine ratio (ACR) were measured on day 1, day 3, and day 7 after admission in 24-hour urine samples. Follow-up duration was 10 years for all patients. RESULTS: At survival analysis, both CRP and ACR were associated with increased risk of 10-year all-cause mortality, also after adjusting for age, hypertension, diabetes mellitus, prehospital time delay, creatine kinase-MB isoenzyme peak, heart failure, and creatinine clearance. CRP and ACR were associated with nonsudden cardiovascular (non-SCV) mortality but not with sudden death (SD) or noncardiovascular (non-CV) death. CRP was not associated with long-term mortality, while ACR was independently associated with outcome both in short- and long-term analyses. At C-statistic analysis, CRP did not improve the baseline prediction model for all-cause mortality, while it did for short-term non-SCV mortality. ACR improved all-cause and non-SCV mortality prediction, both in the short and long term. CONCLUSIONS: ACR was a better predictor of long-term mortality after AMI than CRP.