In 2007, a group of healthcare leaders from around the world gathered at a workshop organised by the US-based Institute of Medicine to discuss how the digitisation of healthcare could best be used to improve patients' health. From that workshop, a new concept emerged - the Learning Health System (LHS).
The idea is simple at its core: use the data routinely generated by digital health technologies - particularly EHR systems - to inform decision-making at multiple levels (clinic, hospital, region, nation, worldwide) so that the healthcare system at each level can learn and improve. For example, data from an outpatient clinic might show that the patient mix had changed in recent years and that staffing should be adjusted to better meet need. At a higher level, the data from multiple hospitals' EHR systems could be used to identify the outcomes of patients enrolled on a clinical trial.
Relationship with evidence-based medicine
Learning health systems have the potential to accelerate the adoption and influence of evidence-based medicine (EBM) - the idea that clinical decision-making should be informed by the current best evidence rather than expert or personal opinion. EBM has traditionally relied on randomised controlled trials (RCTs) to generate research evidence, which is then synthesised through systematic review and meta-analysis before being translated into evidence-based clinical guidelines.
Several challenges have emerged with this approach. The biggest is the cost and inconvenience of running large-scale RCTs. Many treatments in medicine have quite small effects - perhaps successfully treating only a small percentage of patients who take them. Those small effects can add up to large numbers of real people if enough patients take the treatment, but detecting them requires thousands of trial participants randomised between intervention and placebo. Such trials take years and vast sums of money.
By running trials using EHR systems for recruitment, allocation and follow-up, an LHS can rapidly advance knowledge about which treatments work best and for whom. This latter point may be enabled by a further development: personalised medicine. By linking a patient's genome to the data in their EHR (a so-called "digital phenotype"), it becomes possible to draw conclusions about how different genetic combinations influence how effective different medications are. In future, the drugs we are prescribed - and their dosages - may be determined by a genetic blood test before treatment begins.