The Digital Apothecary

View Original

Amazon's push to analyze the medical datasphere with their Amazon Comprehend Medical service

In a recent blog post on Amazon AWS website on Tuesday, Amazon announced the creation of their Amazon Comprehend Medical service to allow developers to start analyzing medical databases. Now, the cool thing here is the machine learning program can parse unstructured medical text and identify key ares such as diagnosis, treatments, medical doses, signs and symptoms of diseases, etc. This is obvious competition for other companies such as IBM Watson who has been in the field for the past few years, but perhaps a well known new-comer is needed to shake things up due in part to recent disappointments in the field.

Amazon is positioning this system to help patients/consumers (I love the different language you see companies in tech vs healthcare use these days) help with their medical care, such as medication management, scheduling appointments, or make decisions in their own care. The interesting item here is the Comprehend Medical system is seemed to be build as an ‘idiot’ proof way of mining medical data as you do not need to create custom codes or rules. Now I have never used AWS and have no experience in AI/Data analytics personally, so this is an outsiders view looking at what Amazon is up to, but hey, sounds cool right?

But one thing I can appreciate is that this is significant competition for other start-ups that have been trying to get this to work for the past few years. Pharma is an obvious target for this, along with its use in clinical trials. I can envision this being used to a large extent to analyze electronic patient diaries or reports. Now, let’s say we have a patient with an Amazon Alexa in their home to track certain data collection for a clinical trial and give oral reports, this platform could be effective in data analyzation.

One significant quote provided in the blog post from Anish Kejariwal, Director of Software Engineering for Roche Diagnostics Information Solutions who said that the platform had the opportunity to help analyze ‘petabytes of unstructured data being generated’ to find useful insights to use for creating better patient assessments.

Overall, this is really cool, but the blog post is very lean overall in what this may ultimately mean.

See this content in the original post


Some of my personal thoughts

If you have been reading my blog for the past few years, you’ll know I drafted up at one point a diagram of where I thought Amazon would enter the health field. One area was the pharmacy, though I thought that Amazon would make it’s own platform, and not acquire Pillpack. However, another area I had thought about was what Amazon could do if it could analyze medical data. I feel that is worth revisiting at this point.

Yeah, I added some solid CAPS here.

This isnt pretty, I know, and I’ve been meaning to update it since I made it almost two years ago. In any event, most of the things I wondered if Amazon might do seem to be relevant to an extent, but this Amazon Comprehend Medical system really seems to fit the bill for where I thought they could also sell services at some point in the far future, but now I’m rethinking its role beyond just consumer product delivery but also implications for other companies and start-ups.

As always, I welcome your thoughts!