It is now straightforward to collect large quantities of physiological data, such as blood pressure and ECG. The key challenge is to unlock the information in these complex data sets. In both heart conditions and infection, early detection allows for critical interventions that will save lives and reduce chronic morbidities. The attractor reconstruction (AR) method developed by Philip Aston and colleagues at KCL crucially analyses all of the data, in contrast to many other approaches that only analyse a reduced dataset. Philip has been awarded funding of £30,000 from the EPSRC Impact Acceleration Account to support a further 6 months for Esther Bonet Luz (a PhD graduate from Surrey, now PDRA) to continue her work on using machine learning for classifying physiological data based on features extracted from the attractor. Two applications will be considered. The first will use the AR method to provide early detection of the onset of fever which would allow for early treatment, which almost invariably corresponds to better outcomes. The second is to detect atrial fibrillation (AF), which is the most common sustained cardiac rhythm disorder and can be associated with serious complications, including an increased risk of stroke.