A statistical framework decomposes Fisher's information matrix to help researchers calculate sample sizes that deliver precise and equitable risk estimates ...
Researchers developed an explainable gradient boosting model using routine liver function tests that achieved strong discrimination while uncovering widespread record ...
A retrospective study reveals that adding machine learning models to routine laboratory data fails to significantly improve Alvarado risk predictions ...
A retrospective pediatric leukemia study found pre-treatment protein and neutrophil levels modestly predict pegaspargase-associated acute pancreatitis.
A multicenter study has developed clinical models using routine perinatal and anesthesia data to estimate neonatal resuscitation and intensive care ...
Incorporating neurological status at hospital discharge into admission models significantly improved 90-day functional outcome predictions for acute ischemic stroke patients ...
A retrospective study of 616 ischemic stroke patients demonstrated that an XGBoost machine-learning model achieved high probabilistic discrimination across five ...
A commentary in Diagnostic and Prognostic Research finds that while large language models show promise for multimodal health records, critical ...
In an external validation study of 37,604 Canadian primary care patients with diabetes, researchers found that while most evaluated kidney ...
A systematic review shows the Thakar prediction model reliably identifies cardiac surgery patients requiring dialysis, though models for milder kidney ...
A retrospective study of 3,887 critical care patients found that integrating pre-admission dynamic lactate trajectories into an LSTM model predicted ...
A retrospective cohort study developed and temporally validated a clinical risk tool to identify drug-resistant tuberculosis among patients with diabetes.
A retrospective cohort study found that an XGBoost machine learning model accurately predicted in-hospital mortality among postoperative neurosurgical intensive care ...
A study of UK Biobank data revealed that age, waist-hip ratio, and handgrip strength predict hundreds of diverse health outcomes ...
Researchers have derived a closed-form Fisher information decomposition to calculate sample sizes needed for precise, fair individual risk predictions in ...
Researchers found that artificial intelligence algorithms predicting deep vein thrombosis produced widely conflicting probability estimates for individual patients despite showing ...
A new protocol for a systematic review aims to compare methodological approaches for developing clinical prediction models for individuals with ...
A Monte Carlo simulation study reveals that violations of the proportional hazards assumption meaningfully alter the discriminatory capacity and predictive ...
Researchers have introduced a continuous net benefit metric that summarises the clinical utility of prediction models across a whole range ...
An empirical study reveals that clinical prediction model stability depends not only on sample size but also on algorithm selection ...
