Big data/machine learning is coming to healthcare. And it’s through the smart watch.

My life has been fairly isolated over the last week. I have made advances in my understanding of computer vision replicating the Harris corner detection algorithm, achieving the following result:

box
It detects corners but there is some error

If you’re interested in the mathematical logic and code used to achieve this result download the brief guide I wrote on this here: [harris_corner_detection_readme]. I’m now on term two and considering the speed that term one rushed by it will not be long before I’m sitting in an exam hall wondering why I ever thought my brain could retain and compute all this information. For now, regular trips to the library and applying for jobs are giving me some form of comfort. Whether it’s ineffective delusion or actual productiveness will be revealed to me when I finally sit these exams. Another sad thing to contemplate is that a lot of my classmates will be going back to their own countries after study. Teresa May has not made it easy for them to secure PhD funding or get a job. I will definitely miss them, there’s some sort of bond fused between people who work towards a common goal or endure the same hardship.

In terms of jobs there seems to be a growing demand for health analytics and artificial intelligence. Whilst machine learning has shown its ability to uncover hidden trends and make predictions I thought that it’s application to healthcare would be slow. Data in the NHS is patchy with different formats and systems. Legislation such as data protection is valid but yet another thorn in the side of the people who want to collect large amounts of data and centrally process it. My mind goes back to an academic I know who was analyzing hospital outcomes. When me met to discuss the practicalities of processing data she told me that she had to go through a virtual machine and process/analyse the data there. This would be a nightmare if you where trying to plug the data with other data sets into a machine learning model. Having personal experience with NHS systems and helping with the adoption of the new computer system in my hospital I was convinced that it would be a long time before we started seeing machine learning and big data widely implemented in healthcare settings.

However, the smartwatch may have some use after all. A recent study from Stanford took over 250,000 daily measurements from 45 patients. Here they were able to detect the onset of Lyme disease and physiological differences with insulin resistance. Now this doesn’t mean that these watches will be saving lives tomorrow, there is still work that needs to be done. The study managed to detect, not predict. But this is a stepping stone. In the following years if previous machine learning advances are anything to go by we should see some predictive models emerge. The exciting thing for me is that we can bypass the tech mess that is the hospital records system.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s