A need for data
Big data is transforming healthcare and predicting future trends. New technologies are already available for patients living with asthma and chronic obstructive pulmonary disease (COPD), alerting users that their condition is worsening up to 10 days before they are at risk of hospitalisation.1 Despite its growing sophistication, can this technology ever truly replace humans in healthcare?
Predicting a trend
Predictive analytics, in all its forms, rely on data.2 Data ‘teaches’ mathematical models how to predict events that are yet to happen based on evidence from the past. The bigger the amount of data available, the more accurate our predictions can be.3
Large data sets can allow a computer algorithm to find patterns in intricate relationships between data points that humans alone would not be able to see.4 The greater the dataset, the greater the wealth of information a machine has to find these connections.
Where limited data sets may have historically been a barrier to effective predictions, we now generate more data than ever before. It is estimated that we generate around 2.5 quintillion bytes of data every day.5 This has seen another advance in our predictive capabilities: the ability to finetune predictions is theoretically greater than ever.
Quality as well as quantity
Understanding how best to use vast quantities of data can be a challenge. Quality as well as quantity is important. Good computational models can minimise data bias and variance with large data sets, however, training computational models with incomplete or poor-quality data sets can have serious implications. Algorithms based on data that does not reflect the diversity of a patient population can result in biases. In some instances, this has resulted in substantial bias when identifying patients for prioritised treatment6 or when developing tools to explore genomic data from a limited population sample.7
We cannot expect predictive algorithms to be balanced if we do not expose them to balanced data. When developing predictive algorithms in healthcare, data bias is an important hurdle to overcome.
Predictive analytics in healthcare
With advances in predictive analytics across so many different sectors, it is no surprise that the healthcare industry is also exploring ways it can be used to optimise patient care. The potential is huge. Predictive analytics could be used to improve differential diagnoses, to alert users to potential health risks or to provide treatment recommendations.8
Studies have shown promising results in predicting disease exacerbations, proactively identifying when patients may be at risk of an asthma attack, septic shock, organ failure or migraine.1,9-13 Artificial Intelligence (AI) is used by the NHS with HeartFlow Analysis to analyse the impact of blockages on blood flow .14
This technology may also improve patient care by streamlining administrative tasks and offering much needed cost-saving initiatives.14
A growing role
Whether this technology can ever replace humans in healthcare is the subject of much debate. It’s true, predictive analytics can already manage many of those tasks traditionally assigned to humans. However, as each advance brings something different to the treatment pathway, technology can never fully replace the complementary balance of human-led empathy and data-driven insight that patient care requires.
While much is still to be done, there is a growing role for analytics and predictive analytics in all our futures. How this will transform healthcare, no one can fully predict.