An ontological framework for computer interpretable clinical guidelines and associated tools

The problem 

Clinical practice guidelines (CPGs) are used by health professionals to guide best practice, evidence-based care for specific health conditions, for example diabetes, hypertension, chronic heart failure and obesity. Since the early 2000s, guideline-based computerised platforms, have been developed to help health professionals deliver the best possible care by enhancing clinical decision support systems (CDSSs), which make it easier to implement the guidelines in practice. However, there is a lack of standardization in formatting the computer-interpretable guidelines (CIGs) that underpin such systems. This can create conflicts and contradictions between conditions or treatment approaches and is a particular issue when dealing with patients with multiple conditions or comorbidities. The management of multimorbid patients is complex and a key health challenge, accounting for 78% of all GP patient visits and being closely associated with mortality and severe disability. Patients with multimorbidity receive multiple treatment regimens and if these are not coordinated it can result in an increased risk of adverse drug interactions and poor adherence to treatment and medication. 

Professor Theodoros Arvanitis, Associate Director HDR UK Midlands Site, explains: “When these guidelines are produced, they usually tend to review the evidence separately for each condition. In some cases, say with diabetes and cardiovascular disease, which are very much linked, the guidelines and treatment paths tend to be more harmonised. But when you then throw in say arthritis and COPD [chronic obstructive pulmonary disease] into that mix, it starts to get very complex, indeed.”  

The solution 

By developing a CDSS which applies CIG in practice, this project aims to code guidelines in a way that is standardised, consistent and unambiguous. The aim of the Better Care project, led by Professor Arvanitis, is to develop a standard language that can execute multiple CIGs for different conditions. It is based on an ontological approach, which is essentially a way of defining various diseases, their attributes or characteristics as well as their relationships with each other.  

“The problem with guidelines is that they can become very rigid if they follow a very specific representation – so we are working towards developing dynamic modelling for CIGs. Here you don’t have static knowledge that simply shows basic relationships, but you have dynamic knowledge and temporal relationships, such as start time, end time and duration.” 

“Ultimately this is about how we support clinicians and patients to make decisions and help them reconcile best practice guidelines which may conflict. That doesn’t necessarily mean convergence to a single complete solution for everyone, but rather a tailored approach to minimise the risk for each patient. For example, we might accept a degree of risk in terms of diabetes management in order to treat a very high risk of renal failure.” 

These dynamic models will incorporate both drug-disease, and disease-disease interactions alongside allergies, drug intolerances, genetics and past treatment history. Initially, the team will work with groups of patients with multiple different conditions to develop specific use cases.  

Impact and outcomes 

This project will develop flexible CDSSs and a supporting standard language.  

The systems will help clinicians and patients to devise long-term care plans with treatment choices and goals tailored to individual patients, involving them in a shared decision-making process that accommodates their preferences. Overall, this will help to maximise treatment adherence and satisfaction with outcomes. 

Professor Arvanitis comments: “Our aim is to translate this work into a practical decision support system and work with NICE [The National Institute for Health and Care Excellence] to address care planning of complex comorbidities. I believe this is an area where we can make a real difference going forward. Through the pandemic we’ve come to appreciate the complexity of COVID-19 and the many complications of infection which will need to be managed alongside existing comorbidities. This represents a key opportunity to apply these frameworks to help provide the best possible long-term care to the many millions of people affected by the pandemic.”