AnalyticsMD (YC W15) Applies AI to Optimize the ER | Hacker News Reader

AnalyticsMD (YC W15) Applies AI to Optimize the ER

techcrunch.com

26 pointsmuditgarg14 comments

I know a lot of ER doctors. They all hate the software that they have to work with; especially the "resource control" software that stands between them and the drugs and tools they need to treat people. This seems like another tool in that direction. It seems designed to take decision making away from doctors and put it in the hands of management.

What do you expect that this tool will do for the working environment of ER doctors and nurses? Or the relationship between them and their patients?

Good question. ER doctors hate when software gets in the way of their clinical decision making. We deliver recommendations to the front line managers but they decide what to act upon. One of the most frustrating things for the doctors often is when they are waiting to treat the patients but can't because there are no beds, or not enough staff or labs are not coming back. Our software anticipates those problems and recommends taking out those bottlenecks in advance.

Our hope is that we will make the day of the staff in the ER much less stressful and allow them to focus on spending more time with the patients.

Just launched our product on TechCrunch. Helping hospitals run smoothly a challenging problem but one we care a lot about.

If any of you want to talk about this or have stories from waiting in the ER or a doctors office please share :)

Congrats!

Thanks @mendelc

How has this performed in the real world? Has it been tried in high traffic areas? (i.e. chicago/nyc)

Does it get reinforced for doing an obviously bad or good job?

Yes. Our data feeds include the outcome variables we are optimizing for which helps the algorithms retrain as new patterns emerge.

Great question. Generally our system performs better in higher volume areas given the higher "n" in the data but we don't have a live deployment in a high traffic community ER yet. Hopefully we will have a more conclusive answer for you soon!

Can you tell more about DecisionOS? Is this something you've built inhouse or is it based on MLLib etc?

How does this data enter your system?

Unfortunately, there aren't any great APIs in the health IT world but we integrate with various hospital systems. We do not ask staff to enter any additional data than what is already in these systems

Is there any lash back from integrating with Epic within the Hospitals?

What is your tech stack?

Reply on news.ycombinator.com