While our respected medical colleagues will stand firm that their focus is on the patient in front of them, and that they have therefore always practiced personalised medicine, the overall reliance on evidence-based medicine (EBM) means that when it comes to treatment selection; this is not always the case. EBM, under the guise of what we may cautiously describe as population health, identifies and prescribes the treatment most likely to be beneficial to the largest percentage of the entire cohort of patients with a given ailment. This approach focuses on the disease, and in many ways when government budgets become involved, on the business model of cost/benefit analysis.
Learning Health Systems (LHS) are considered as the next evolution of EBM. LHS are capable of supporting approaches that focuses, instead, on the patient. This approach is described as individualised or personalised medicine.
Earlier this year a group led by Scott McLachlan, and consisting of academics and clinicians from Queen Mary University of London (UK), Massey University (NZ), the University of Montana (USA), the University of Leeds (UK), and University College London (UK) wrote to the Editor of the BCS Journal of Innovation in Health Informatics (JIHI) highlighting an awareness challenge for LHS in the health informatics research community. In their letter (found here) the team identified that while LHS had been defined and established in the community for some ten years, many authors lacked awareness that a vast number of their efforts in designing health information systems (HIS) actually represented LHS. We believe this lack of awareness has inhibited ongoing LHS research, and prevented the formation of a critical mass within the domain. We closed by identifying that research into a framework and classification structure for LHS were urgently needed, as this would enable confident identification of works within the domain.
The group went on to consider the form that this framework and classification structure might take, and today we can announce that JIHI have published the culmination of our research in this regard (found here). Just as the Norse God Heimdall was said to be the son of nine mothers, our efforts identified a taxonomy of nine types of LHS. These ranged from simple systems that identify classes of patients known as Cohort Identification, through to systems that can conduct Comparative Effectiveness Research and Predictive Patient Risk Modelling using large datasets of prior patients. We also demonstrated where these LHS types are situated within the wider technological and learning environments of the healthcare provider, and how they help clinicians focus patient care from population, to individual.
Projects like Pambayesian seek to produce LHS that can use a combination of several LHS types, along with clinical expertise, to improve patient diagnosis, prognosis and overall outcomes. Pambayesian does not seek to replace the clinician. Rather, we seek to enable the patient to be better informed and participate in their own day-to-day care, and reduce those repetitive and time-consuming parts of the clinicians workload that reduce their ability to have the greatest positive effect on the lives of the largest possible number of patients.
Watch this space for further research and development in the area of Learning Health Systems, Medical Bayes Networks, and our efforts to support the work of our esteemed medical colleagues.
Recent Comments