Development and validation of clinical prediction models for personalised renal function monitoring in people with heart failure in primary care: The Renal-HF Study Protocol

Abstract Aim Heart failure is a growing problem in society with an ageing population and many patients with heart failure are affected by renal dysfunction. The RENAL-HF project aims to develop predictive risk models to support personalised renal function monitoring and treatment in patients with heart failure in primary care. Methods This study will use electronic health records from the Clinical Practice Research Datalink (CPRD) database for patients who were diagnosed with heart failure. We will develop 3 prediction models - Mixed-effects model, Growth mixture model, and recurrent neural network-long short-term memory (RNN-LSTM) model to predict future worsening renal function (WRF), including events that lead to hospitalisation, and death. Using an internal-external validation approach based on geographic region, we will choose the top-performing model using various metrics to evaluate the predictive performance. Conclusion This protocol provides a detailed description of the methods used for developing and validating prognostic models for personalised renal function monitoring in people with heart failure in primary care. Protocol registration The study and use of CPRD data were approved by the Independent Scientific Advisory Committee for Clinical Practice Research Datalink research (Protocol Number: 22_001794).