Smart analytics can play a significant role in healthcare of the future – so how do we make them work, asks John Loder
We all know the next few years in healthcare will be very difficult. But looking beyond that, I think there’s good reason to be extremely optimistic.
In fact, I think we could be on the verge of the most exciting period since the 1950s in terms of medical advance, effective care, and increasing productivity.
Data science may be as important to 21st century medicine as organic chemistry was to that of the 20th century
Patients are becoming increasingly connected. Data is pouring in via smartphones and wearables such as fit bits. We already have pills that broadcast a signal when they have been swallowed and cheap home ECGs that can pick up intermittent heart arrhythmia.
Looking ahead, it won’t be long until we see new products such as contact lenses that measure blood glucose levels or even toilets that can track levels of neurotransmitters.
This connection is two way. Millions of people use digital devices to connect to peers and experts, and to get helpful visualisations and analysis of the data they have previously collected, as well as behavioural prompts and triggers that help them manage their condition.
Patients and doctors are already using data together. A recent Nesta survey showed that 15 per cent of patients with a longer term condition are already using digital technology – whether an app or other IP-enabled wearable device – to track their symptoms.
More surprisingly, 89 per cent of GPs in the UK said they found patient-generated health data like this useful in consultations.
In coming years, we believe that medical practice will increasingly take advantage of this. “Prescribing” a sensor that a patient can wear will be the normal outcome of a GP visit, determining if referral is necessary, and making sure any follow-up is well informed.
Surgical follow-up will be determined by home measurement of individual recovery, while real-time monitoring of chronic conditions will trigger rapid interventions, avoiding deteriorations and unnecessary hospital admissions.
Patients who share their data could boost research progress, which has slowed. Medical science has built perhaps humanity’s largest collection of knowledge
Personalised medication and therapy will be based on details about an individual, rather than standard approaches for a given diagnosis.
What we lack today are analytics that can extract a useful signal from complex data. Data is useless without analysis. However, there is good reason to think this will come.
The rapidly increasing ability of machines to make complex judgements is in evidence all around us – self-driving cars, facial recognition, voice recognition, natural language processing etc.
The efforts of Apple, via HealthKit and ResearchKit, and Google via DeepMind Health, to explore this area, show the serious resource already at work here.
Indeed, data science may be as important to 21st century medicine as organic chemistry was to that of the 20th century.
So, will this trend make a real difference to key outcomes? There are three reasons connected patients and smart analytics should be seen as highly significant, clinically and financially, rather than all too common techno-utopianism.
Firstly, knowing more about patients allows us to apply medical knowledge in a more timely and accurate way, with fewer errors. Clinical medicine has become amazingly good at deducing diagnoses from a fairly small amount of information.
However, connected patients can make diagnosis and monitoring much easier and more reliable. Cambridge Cognition (for Dementia) or Parkinson’s Voice (for Parkinson’s) show how already commonplace technology – tablets and phones – can provide reliable remote diagnostics tests.
What we lack today are analytics that can extract a useful signal from complex data. Data is useless without analysis
Secondly, connected patients who share their data could boost medical research progress, which has slowed in recent decades. Medical science has built perhaps humanity’s largest collection of highly validated knowledge.
Yet nearly all of it concerns the average response to a treatment of a patient with a single diagnosis. It tells us little about what causes variation between patients, and what to do about co-morbid patients.
Rich data from connected patients, combined with advanced analytical techniques, is likely to be far more effective at unearthing these complex relationships than successive randomised controlled trials. Given the right relationship, they could also be much cheaper and much faster.
Thirdly, this precision medicine has another huge advantage – it can reduce costs. In medicine, unlike almost any other industry, new inventions and discoveries have tended to increase costs.
Each new treatment or intervention requires an accompanying dose of clinical labour, and sometimes a whole new specialism. Precision medicine, as the name implies, diagnoses more quickly and targets treatments more accurately: its influence reduces labour intensity and so cost.
How we make it happen
To realise this potential we need to develop the analytics. To do this we need to gather data sets of the size and quality necessary for which data scientists can work. And doing this requires that we:
- Build simple and clear digital consent procedures. This new data is produced, owned and controlled by patients. Care.data showed that the public can be unforgiving to ideas that seem not to respect this ownership. So, as is right, it will be accessed on their terms, and, largely, with their explicit consent.
- Produce well defined interfaces with compelling functionality. NHS patients tend to be older, and of a generation that is not well served by digital developers and designers. Digital interfaces remain hard to use, for those who did not grow up with them.
Rich data from connected patients, combined with advanced analytical techniques, is likely to be far more effective at unearthing these complex relationships than successive randomised controlled trials
Finally, and most importantly, we need to do all the above with people. We need to build active communities of engaged patients that are working on these problems together with researchers and developers.
Patients are clearly motivated, both to track their own health, and contribute to research. Patient associations are already extremely active research collaborators – maintaining registries, assembling biobanks, raising funds and lobbying government.
If we engage people as partners in research and development, and grasp the potential presented by the connected patient, we could see rapid clinical progress and begin to build a sustainable health system.
John Loder is head of strategy at Health Lab, Nesta - www.nesta.org.uk
The rise of the connected patient
- Currently reading
The rise of the connected patient