Personalised medicine has, over recent years, made the transition from ambition to reality. It moves healthcare from taking a one-size-fits-all or blunt approach to something far more responsive and tailored.
Essentially, it involves stratifying patients according to a range of factors – from broad categories such as risk profile, disease subsets and responses to particular diagnostic tests, through to highly individualised profiling based on factors like molecular and behavioural markers – and applying treatments tailored precisely to those profiles.
The results can be incredibly powerful. Clearly, the most compelling effect of all is improved patient outcomes, with faster and more comprehensive recovery from illness and accident, and fewer lives lost. But personalised healthcare can also generate great operational efficiencies too, by ensuring that medicines, therapies and interventions are applied in the most effective way, and where they are most impactful. And it can be a great research tool, generating new insights which can inform the next generation of healthcare.
However, developing personalised medicine requires data – and lots of it. When people’s health and even their lives are at stake, there is no room for error. Personalised medicine must be based on truly robust data, and large volumes of it. That data must then be analysed with the help of machine learning algorithms which can learn from themselves, generating new insights and spotting new patterns over time – the core mechanism of big data analytics.
Here are some of the types of data which can feed into personalised medicine:
- Electronic health records (EHRs): As individuals’ medical records are digitised, so they can be anonymised and analysed en masse.
- Connected health device data: As the IoT for healthcare ecosystem expands, so the range of connected devices that individuals can wear – or even have implanted into their bodies – is expanding too. These devices collect an array of useful medical data, from heart rate and blood pressure to activity levels, which can provide valuable insights into likely pathways.
- Clinical data: Medical research is a vast and diverse field. Laboratories of all scopes and sizes are working on myriad different tests and trials, and generating data across myriad different formats.
- Environmental data: Information such as pollution levels and temperature can play a key role in predicting who is more at risk of particular conditions.
- Genetic data: Genetic data can pertain to an individual or groups of individuals, and covers an enormous array of different information types. Genetic data can also pertain to different strands of bacteria or virus, and profiling of conditions like cancer.
- Behavioural data: Information on how people are likely to behave in different situations can be hugely useful in determining patient pathways. For example, how can patients be best persuaded to adopt healthier diets, or take more exercise?
Such datasets are already being analysed and harnessed in some hugely innovative ways. In the US, for example, the Precision Medicine Initiative Cohort Program aims to sequence the genomes of more than one million American citizens, creating a huge and detailed cohort which can then be tracked over time. This means that when, say, a drug company wants to work with 100 individuals with a particular genetic trait, they will be easily identifiable.
On a rather more immediate basis, there is plenty of research taking place in relation to the ongoing COVID-19 pandemic, with researchers and clinicians seeking to better understand why particular populations seem to be affected more than others, and whether particular treatments might be more effective in particular groups of patients.
Such a huge range of data is available to healthcare researchers that the possibilities for personalised medicine truly are enormous. We are on the edge of an exciting new frontier – all powered by big data and machine learning.