Last year I presented our pioneering work on serverless Machine Learning for genomic health care to 1000+ innovators and developers at the #AWSPSSummit.

In this blog post, I am sharing my 3 key messages for the presentation.

1) Healthcare needs to shift from treatment to prevention

Healthcare is moving away from the reactive illness treatment towards proactive measures such as prevention to intervene early. This will provide better outcomes for individuals and also help curb the recent unsustainable increase in health care cost. This is discussed in detail in the CSIRO #FutureOfHealth report.

2) Once you go serverless you never go back

I strongly believe that serverless allows the setup of minimal viable products (MVP) and prototypes at a speed and minimal cost that is required to stay competitive. Serverless also allows those prototypes to mature and enter production in a fully scalable manner.

While distributed architecture comes at the cost of being difficult to monitor and require sub-partition for large tasks due to the resource limitation, we found that overall the benefits by far outweigh the difficulties. So most of our new development is serverless as default, such as the Gen-Phen-Insight tool, a clinical decision support tool, that we released at the summit. Read more about this in our AWS Global Public Sector Blog post.

3) Evolve architecture more effectively using hypothesis driven approaches.

With new cloud services or software libraries released everyday it has become impossible to know which approach will result in the best outcome. Hence framing the problem as a scientific experiment will help keep the pace of innovation high even in an environment of ever increasing choice.

In our blog post we argue that serverless lends itself particularly well to this approach as individual services can be swapped in and out especially when using Infrastructure as code where a change can be as simple as changing a single line of text. The performance of these competing architectures can then be tested empirically through a standard green-blue deployment approach, where the winning architecture gets evolved further.

Let's build a healthier Australia together.

CSIRO is Australia’s innovation catalyst - but you are its influencers, innovators and developers, so let's work together to build a healthier Australia. Therefore if you are interested in working together or contribute your time on machine learning, serverless or genomics problems please reach out to me on LinkedIN or Twitter.

Here is the video (I am on about 30 min in) and I also would like to share my slides. I am happy for you to use them to make the argument for increasing the engagement between research and industry.