Data Science is emerging as a business-critical practice in a number of sectors, says Matt Lovell, CTO of Centiq, but making optimum use of the opportunity may require a different mindset
The challenge with data is not its collection but understanding its true meaning and value. IBM estimates 90% of all the data existing in the world today has been generated in the last 24 months. Indeed, for data intense organisations such as the NHS, with disparate trusts throughout the country generating vast amounts of patient and operational data, the ability to harness this information to create insight has never been more important.
Ironically, we, in the technology profession, may have promoted Big Data as a panacea without truly and fully understanding the problems it can solve. There is a big opportunity here – we just need to think differently. We have much more to gain and exploit from Data Science but how we apply and secure this is now more important than infrastructure.
Data Science is emerging in a number of sectors as a business-critical practice and the healthcare sector has some of the most interesting developments. For example, aggregation of data sets from multiple sources are helping us understand the dynamics of specific cancers in more detail so much faster than ever before.
Aggregation of data sets from multiple sources are helping us understand the dynamics of specific cancers in more detail so much faster than ever before
Take the work Google DeepMind is undertaking with University College London Hospital NHS Foundation Trust. By applying its DeepMind artificial intelligence to the CT and MRI scans of 700 former cancer patients, Google’s technology will quickly distinguish healthy from cancerous tissue. It is hoped a resulting algorithm will cut the time needed to design targeted radiotherapy treatments from four hours to one.
The fact is, Data Science in an environment like this has life-critical implications. It provides new angles to understand and learn more from historic, existing and new data and patient outcomes can be improved as a direct result of data science. This is where the addition of Deep Learning and Machine Learning (intrinsic parts of Data Science) comes into play. These allow us to create and apply structure to data through automated approaches which, until recently, would have taken years to process and analyse.
Data abstraction is a process to understand the essential data characteristics that are always present and additional ones that emerge from it. Just like a picture paints a thousand words, a data scientist can see beauty in a Big Data set.
Matt Lovell, CTO of Centiq
Removing all personal identifiers, for example, means anonymised patient records along with past and present data treatment assessments allow us to analyse and improve our understanding of future treatments. Machine learning automates these predictions and allows proactive analysis of many conditions; it can even trigger intervention. Where the trigger to intervention or assistance is critical, and the time to response is a matter of life or death, this is extremely powerful.
Data Science essentially enables clinicians to make better treatment decisions, faster. A good example of this is the work Sophia Genetics is carrying out to detect cancer in the lungs, skin, ovaries and breast, as well as congenital diseases. By sequencing the patient's tissue samples Sophia Genetics then uses machine learning to compare the results and suggest the most effective treatments.
Combine this with wearable technology and we start to envisage a world where, regardless of location and providing we are contactable and connected, technology can truly understand so much more.
Data Science essentially enables clinicians to make better treatment decisions, faster
A recent report published in PLOS ONE assessed the effectiveness of doctors to predict life expectancy accurately, an area which can be hugely complex to predict with so many variables involved for patients. If doctors and medical professionals had greater access to machine learning algorithms, which permitted data entry to compare circumstances and current status, it could provide them with an extremely powerful check and balance in providing crucial patient information and support.
There is untold value in the data we already have. For many businesses, there are significant repositories of data contained in warehouses and databases, which Deep Learning and Data Science – when connected together – can and do yield a much deeper understanding of our customers and lives.
Data Science allows us to ask questions with a different logic in different ways to understand our customers. Many larger enterprise ERP and business warehouse systems do not currently link these systems together, most often because this is complex in terms of technologies, processes and people. Unlocking this ability is key to businesses creating new value from existing and hugely powerful data.