A systematic review of multi-variate time series approaches to extract predictive asthma biomarkers from routinely collected diary data

Objectives: Longitudinal data is commonly acquired in asthma studies, to help assess asthma progression in patients, and to determine predictors of future outcomes, including asthma exacerbations and asthma control. Different methods exist for quantifying temporal behaviour in routinely collected diary variables to obtain meaningful predictive biomarkers of asthma outcomes. The aims of this systematic review were to evaluate the methods for extracting biomarkers from longitudinally collected diary data in asthma and investigate associations between the extracted measures and asthma patient reported outcomes (PROs). Setting: A systematic review of MEDLINE, EMBASE, CINAHL and the Cochrane Library was conducted, using index terms relating to diary variables and asthma outcomes. Studies that focused on preschool children were excluded, to avoid confounding asthma with multi-factorial preschool wheeze. Study quality and risk of bias were assessed using the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) and the Prediction model Risk Of Bias ASessment Tool (PROBAST), respectively. Participants: Adults and/or children of school age (≥5 years old), with clinician-diagnosed asthma Primary outcomes: Asthma PROs, namely asthma exacerbations, asthma control, asthma-related quality of life and asthma severity Results: 24 full-text articles met the inclusion criteria and were included in the review. Generally, higher levels of variability in the diary variables were associated with poorer outcomes, especially increased asthma exacerbation risk, and poor asthma control. There was increasing interest in nonparametric methods to quantify complex behaviour of diary variables (6/24). TRIPOD and PROBAST highlighted a lack of consistent reporting of model performance measures and potential for model bias. Discussion: Routinely collected diary variables aid in generating asthma assessment tools, including surrogate endpoints, for clinical trials, and predictive biomarkers of adverse outcomes, warranting monitoring through remote sensors. Studies consistently lacked robust reporting of model performance. Future research should utilise diary variable-derived biomarkers.