The range of data clinicians need to engage with and synthesize is expanding exponentially. Support tools such as machine learning platforms, deep learning and novel AI techniques are urgently required.
Delivering effective healthcare boils down to three linked questions: Who to treat, what treatment to apply, and when in the patient’s personal health journey is the best moment to intervene for maximum benefit. This is true for everything from primary care through to hospice care.
Traditionally, clinicians faced with making decisions about the patient in front of them may draw on the recorded history of that patient, on their own experience and on the experience of other clinicians with other patients, often summarised in clinical guidelines. However, such guidelines are typically designed for the average patient, not the individual patient.
We are now in a new era. The range of data that clinicians need to engage with and synthesize is expanding exponentially. Support tools such as machine learning platforms, deep learning and novel AI techniques are urgently required to draw out the powerful, actionable intelligence hidden in this data avalanche that, in combination with clinical expertise, will transform how we address the questions of who, what and when.
CCAIM is exploring precision medicine in multiple streams, including applied precision medicine, pre-symptomatic care and AI-enabled hospitals.
Applied Precision Medicine (Individualised Therapeutics)
We are integrating omics data and a wide variety of other data sets, including electronic health records (EHRs), and applying novel machine-learning techniques to better characterise individual patients, improve diagnosis, account for co-morbidities and reliably predict patient trajectories. This enables the optimisation of targeted treatment interventions.
An important aspect of precision medicine is the use of machine learning to develop personalised “counterfactuals”. These reveal what is likely to happen if this particular patient decides to delay treatment, chooses a different treatment or opts for no treatment at all. Counterfactuals allow patients and their clinicians to explore the personalised potential consequences of big decisions before making them.
Machine learning and AI can also predict health outcomes for hospital patients, such as lengths of stay and complexities of stay – ward, high-dependency unit, intensive care? – responses to resuscitation efforts, the probability of readmission and the outcome of that readmission.
Delivering precision medicine requires access to the relevant data, the latest thinking in machine learning, and deep partnerships with healthcare providers such as the NHS. These are CCAIM’s strengths. Here’s a good example of pioneering work – led by CCAIM Director Professor Mihaela van der Schaar in collaboration with the NHS – on transforming cancer care through AI-enabled precision medicine.
Pre-symptomatic Care (‘Precision Prevention’)
An ounce of prevention is worth a pound of cure, so can we identify looming health issues in large numbers of people even before they develop symptoms and visit their GP, enabling low-cost, early and effective interventions? The answer is yes, absolutely, and the pay-off for developing such technology could be enormous, not only in terms of public health but also in savings to the health economy.
We aim to develop technology to identify cohorts of pre-symptomatic individuals who we can reliably predict will develop common, chronic conditions. We will achieve this using AI to interrogate primary care/community clinical data sets, including genomics, in addition to social demographics data sets – deprivation indices, ethnicities, fuel poverty, air pollution and more – and ultimately measures of patient engagement in their own healthcare management (also known as patient activation).
For such cohorts, early and personalised preventative measures or behavioural nudges, offered before their health would otherwise deteriorate, could result in markedly increased healthspan for a large number of people. See the NHS Diabetes Prevention Programme as an example of the idea in action.
Our work will lead naturally to an exploration of the cost-effectiveness of early, ML/AI-driven interventions. For every pound spent on earlier prevention, what is the saving to the overall health economy?
This subtheme overlaps with many aspects of applied precision medicine, of course. One of the key aims in unlocking the potential of precision medicine in the hospitals of tomorrow is to make data-driven insights visible – and ultimately indispensable – to busy healthcare professionals.
The development of AI-driven decision-support tools is pointless if these tools are not accessible and comprehensible to the clinicians and other healthcare professionals that they are design to support. This is why the CCAIM team includes senior NHS consultants such as intensive care consultant, Dr Ari Ercole, and Consultant Transplant Hepatologist, Dr Alexander Gimson, who in addition to our senior partners in the NHS will champion the translation of our tools into clinical practise.
Hospitals create an immense stream of data across multiple domains. CCAIM aims to transform these data into valuable clinical and administrative insights, to the great benefit of the patients of AI-enabled hospitals.