• Nearest neighbour propensity score matching and bootstrapping for estimating binary patient response in oncology: A Monte Carlo simulation

      Geldof, Tine; Dusan, Popovic; Van Damme, Nancy; Huys, Isabelle; Van Dyck, Walter (Scientific Reports: A Nature Research Journal, 2020)
      Nearest Neighbour (NN) propensity score (PS) matching methods are commonly used in pharmacoepidemiology to estimate treatment response using observational data. Unfortunately, there is limited evidence on the optimal approach for accurately estimating binary treatment response and, more so, to estimate its variance. Bootstrapping, although commonly used to accurately estimate variance, is rarely used together with PS matching. In this Monte Carlo simulation-based study, we examined the performance of bootstrapping used in conjunction with PS matching, as opposed to diferent NN matching techniques, on a simulated dataset exhibiting varying levels of real world complexity. Thus, an experimental design was set up that independently varied the proportion of patients treated, the proportion of outcomes censored and the amount of PS matches used. Simulation results were externally validated on a real observational dataset obtained from the Belgian Cancer Registry. We found all investigated PS methods to be stable and concordant, with k-NN matching to be optimally dealing with the censoring problem, typically present in chronic cancer-related datasets, whilst being the least computationally expensive. In contrast, bootstrapping used in conjunction with PS matching, being the most computationally expensive, only showed superior results in small patient populations with long-term largely unobserved treatment efects.
    • Patient-level effectiveness prediction modeling for glioblastoma using classification trees

      Geldof, Tine; Van Damme, Nancy; Huys, Isabelle; Van Dyck, Walter (Frontiers in Pharmacology, 2020)
      Little research has been done in pharmacoepidemiology on the use of machine learning for exploring medicinal treatment effectiveness in oncology. Therefore, the aim of this study was to explore the added value of machine learning methods to investigate individual treatment responses for glioblastoma patients treated with temozolomide.
    • Real-world evidence gathering in oncology: The need for a biomedical big data insight-providing federated network

      Geldof, Tine; Huys, Isabelle; Van Dyck, Walter (Frontiers in Medicine, 2019)
      Moving towards new adaptive pathways for the development and access to innovative medicines implies that real-world data (RWD) collected throughout the medicinal product life cycle is becoming increasingly important. Big data analytics on RWD can obtain new and powerful insights into medicines’ effectiveness. However, the healthcare ecosystem still faces many sector-specific challenges that hamper the use of big data analytics delivering real world evidence (RWE). We distinguish between exploratory (ExTE) and hypotheses-evaluating (HETE) studies testing treatment effectiveness in the real world. From our experience and in the context of the four V’s of data management, we show that to get meaningful results data Variety and Veracity are needed regardless of the type of study conducted. More so, for ExTE studies high data Volume is needed while for HETE studies high Velocity becomes essential. Next, we highlight what are needed within the biomedical big data ecosystem, being: (a) international data reusability; (b) real-time RWD processing information systems; and (c) longitudinal RWD. Finally, in an effort to manage the four V’s whilst respecting patient privacy laws we argue for the development of an underlying federated RWD infrastructure on a common data model, capable of bringing the centrally-conducted big data analysis to the de-centrally kept biomedical data.