Human Errors in our Healthcare System

Since this is where I write random stuff, I just want to say a few things about human errors, especially in the field of pharmaceuticals.

While working at McKesson, I worked on the reporting capability of a product call RxSafety Advisor.

Read all about it here

While I can’t go into details about how it works, what it basically does is realtime rejection of prescriptions with drugs that either look alike or sound alike drugs the patient typically picks up, but is NOT.

This is only one of its many features.


Two things really bugged me about this project: the scale of human fallacy in our way of dispensing these life saving substances and the under utilization of our available data. Think about the amount of data McKesson has to have in order to do this. This data could be useful in so many different ways to derive better formulation of drug cocktails against different diseases.

After reading about Max Levchin and the work his team at Glow is doing, I think regulatory reform in medical data usage will be as ground breaking as our discovery of antibiotics. It is a field with a rediculous amount of un-organized data. As McKesson have demonstrated, just structuring a tiny chunk of this data pool has already led to millions of lives saved in the form of human error prevention.

However grim the current climate for medical data usage is, it does seem like there is a bright future ahead. Akido Labs, a recent YC start up, is creating an API layer for developers to interface with medical software in the form of a JSON API. A problem CommonWell Health Alliance is also pushing to address. If efforts like these continue, healthcare will soon enjoy the same change Wikipedia brought to education.

p.s. People complain about ObamaCare, but it encourges ancient behemoths like Humana, Cigna, and CVS Caremark to gather complete history of care. I think with some parsing and incorporation of machine learning it’s pretty easy to imagine the possibilities!

p.p.s. The infrastructure for healthcare transactions still uses Tandem Non-Stops and working with that is… a little rough.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.