Preventing the NHS winter crisis through predictive analytics

Despite plans put in place in September for 3,000 more beds to be made available this winter, between Christmas Day and New Year more than 20 NHS Trusts in England reported that they were at 100% operating capacity. As a result of demand for hospital care exceeding supply, non-urgent operations have been cancelled this month to try to ease pressure on the system. There appears to be a clear need for better hospital demand forecasting to aid with capacity management in the NHS.

Since approximately 40% of beds in the NHS are occupied by planned routine care, better modelling of demand for emergency care could allow capacity to be proactively freed-up in anticipation of peak winter demand. The flu season in Australia and New Zealand was particularly heavy during their most recent winter, and so it could have been recognised that 3,000 more beds would not be adequate in addressing the anticipated surge in demand for emergency NHS care.

Indeed, ancient Greek physician Hippocrates recognised the importance of health forecasting. For Hippocrates, forecasting disease outcomes was an important part of medical treatment. He sought to gain better insight into the environmental causes of disease to help provide more accurate disease prognosis, as well as better preventative medical care.

Better NHS forecasting of the link between winter weather and condition-specific disease could help to provide more reliable models for hospital demand. For example, based on research linking COPD symptom severity to cold weather conditions, the Met Office provided a Healthy Outlook service in the winter of 2012/13 to help inform those with COPD about any potential adverse cold weather periods that might worsen their condition so that they could take preventative action. A better understanding of the impact of extreme winter weather conditions on the risk of developing a range of different diseases could be used to help patients act through self-management or primary care. In doing so, this should help ensure less strain is placed on urgent emergency care.

There is also the opportunity for the NHS to leverage big data to help improve sickness forecasting. In recent times, technological advances have enabled health service indicators to be more easily and cheaply measured. However, population-level health indicators are less accurate and well-validated. The NHS could work with tech companies to develop dynamic crowd-sourced statistics on population health and risk of disease that are more responsive than those presently available. For example, the app Sickweather scans social media networks and other 3rd party data sources for indicators of illness and provides users with their location-based probability of sickness. Based on this crowdsourced approach, the app claims to be able to predict the rate of illness up to 15 weeks in advance with over 90% accuracy. Coupling location-based condition-specific sickness data with its own statistical data on disease incidence could help the NHS to create more flexible patient flow forecasts that are more responsive changes in the external environment.

It is evident that hospital capacity in the NHS needs to be better managed during winter to prevent demand outstripping supply. Predictive analytics models that leverage crowd-sourced sickness data could provide more dynamic forecasting of location-based demand for emergency care and help to more accurately project potential patient overflows. It could also enable the NHS to be more proactive in preventing capacity problems. Routine care could be scheduled at times of year when and in locations where the probability of demand surge is less severe. Moreover, the NHS could help those with a high probability of contracting an illness that could require urgent hospital care to proactively self-manage or seek primary care before their condition develops into an emergency.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: