Exit133 is about Tacoma

How machine learning is saving lives while saving hospitals money

The Center for Data Science may be one of the less well-known departments at UW Tacoma, but it's doing some interesting and valuable work that is really putting the University on the map.

One project is a machine learning program developed in partnership with Multicare that could help hospitals significantly lower the rates of readmission for heart failure - improving patient health outcomes, and saving hospitals money. The Risk-o-meter analyzes more than 100 factors, producing risk scores that help doctors make decisions on patient care throughout the treatment process, and even after the patient has left the hospital.

You don't have to be much of a techie to know that big data is a big buzz word these days, and with an Affordable Care Act emphasis on reducing hospital readmissions, this technology hits a couple of hot topics right now.

It's exciting to see UWT getting attention for this innovative project.

Read more at Gigaom.com



I work in the healthcare field and talk to hundreds of providers every month.  I know the industry.  This tool, in theory, is good.  But, the most important thing about medicine is that it is practiced; one size does not fit all.  That means that every patient is different.  Their tolerance for pain, anguish, or whatever else ails them is different from patient to patient.  This allows or disallows medical providers to choose their line of therapy tailored to the patient.  This tool is a way for hospital systems to create one size fits all formularies (roadmaps of care) for every disease state.

As well, I have met many surgeons who are basically cherry-pickers.  That is, they choose the easiest slam-dunk stuff to up their reputation in the field (and therefore their dollar value).  They simply will not attempt that risky cancer surgery, as an example, that only has a 10% chance of saving your ass even though the alternative is not attempting anything at all and watching you die.  This tool will help them avoid these risky “hail-Mary” surgeries so they can avoid all things risky.  In fact, I can totally foresee hospital systems regulating possible care metric in accordance to this sort of system.

This type of stuff is not all good.

July 15, 2014 at 1:10 pm / Reply / Quote and reply

0 | 0

Peter PeterRegistered

The very idea of a machine learning based system is that it isn’t “one size fits all” as you describe it. Sure, it’s only a tool. But it is one that can track a lot more variables in real time than any individual can keep track of themselves.

I see this as a good thing.

July 15, 2014 at 2:33 pm / Reply / Quote and reply

0 | 0


Absolutely it’s a good thing.  I agree.  But, I’ve seen things like this implemented in a way that makes a patient’s care tied to a mathematical equation - which is notoriously done in hospital systems to create formulary.  There’s so much more to a patient than that.  Mathematical equations ignore the reality of a patient’s lifestyle, psyche, ability to cope with disease, and on and on.  Things like this are a pitfall for people with engineer brains who make management decisions at hospitals - they forget that they’re ultimately dealing with people who are all different.

July 16, 2014 at 10:30 am / Reply / Quote and reply

0 | 0

Post A New Comment

Please enter the word you see in the image below: