Data to find Digital Solutions in Health (or oil to power an car engine)

  • Data is not everything and although it has been compared to oil, there are big differences
  • Clive Humby, director of the Tesco Loyalty card program coined the very optimistic phrase:

“Data is the New oil”

  • However contrarily to oil…Data cannot be sold as it is on the stock market once extracted:
    • It has to be classified into its use an is often stored before its use is determined. It can then be sold.
    • Unlike an oil field, who owns the data “field” and privacy issues ?
    • However there are clear data ownership issues. Just because you submit the data on Facebook for example, does it mean that because you agree to terms and conditions, you don’t necessarily understand or take the time to read, that the company can do whatever it wants.
    • This also leads to the problematic of data privacy. There are numerous tricks on media platforms where you can hide things to other users. It prevent Facebook to bundle up and sell that information to third parties such as in the Cambridge Analytics Scandal in 2018.
      • Our interpretation of this privacy and data ownership problem is because of a relationship to the platform which has more in common to an addiction than to rational thought out behavior in any decisional process.
      • The questions that we can ask: is an addiction to gambling or drugs a reason to enter an agreement, and is anything other than a helpline and occasional bans the remedy chosen by society enough ?
    • Oil is in limited resources…Data on the contrary is infinite.
    • Oil and Data need to be refined to become useful.
      • Crude Oil needs to be broken down into its constituents (kerosene for planes….) through distillation columns in refineries.
      • Data needs to be broken down into solutions to questions through machine learning derived algorithms (for example) through computer processors:
        • the goal is to detect patterns and predict
        • let’s add that for health, explainability is paramount, so supervised learning with labelling of the data is required (see “Machine learning in AI crash course”)

Reference: Big Data: A very short introduction by Dawn E. Holmes. Oxford University Press, 2017


Scroll to Top