Supporting healthcare policymaking via machine learning – batteries included!

Guest post by Marc Claesen from STADIUS, KU Leuven

Due to aging populations and the ever increasing strain on healthcare budgets, defining effective healthcare policies is a major challenge. This induces a critical need for tools to support optimal decision making by unveiling important drivers and dynamics and accurately simulating the impact of potential policy changes on various aspects of our healthcare system. The Big Data revolution heralds a new era for data-driven, fact-based decision support and knowledge discovery. We are convinced that optimally exploiting the wealth of relevant information embedded in diverse databases presents an enormous opportunity to inform and instruct healthcare policymaking. Hence, the participating groups of KU Leuven are thrilled to be part of the ambitious MIDAS project which aims to leverage Big Data to offer data-driven decision support to healthcare policymakers.

Unfortunately, the mere availability of large amounts of relevant data is not actionable without a means to distill insights or knowledge to support pending decisions. Machine learning is a key enabler to derive actionable insights from data and is currently considered one of the top tech trends that will innovate a wide breadth of domains. The combination of advances in machine learning and distributed cloud computing enable us to construct decision support systems capable of processing large quantities of complex medical and other data, often stored in different formats and distributed across physical sites.

The STADIUS lab of KU Leuven is honored to lead the data analytics efforts within the MIDAS project. We are committed to engage with all project partners and the policy board to develop a comprehensive solution. From a data analysis perspective, we focus on identifying dynamics and drivers in different healthcare segments, using these to inform the decision process and enabling policymakers to run accurate forecasts and simulations to assess the effects of certain policy changes. Given the sensitive nature of medical data, a key challenge in MIDAS is providing privacy-preserving data analysis methods that enable policymaker decision support, while simultaneously guaranteeing confidentiality of patients’ medical information. Despite the apparent paradox, we are convinced that significant steps can be made towards data-driven decision support without sacrificing privacy.

We are privileged to tackle this challenge as part of the MIDAS consortium, which provides a unique and powerful blend of expertise, knowledge and perspectives to draw upon in developing the solution.