This makes sense to help us predict but like all algorithms out there it’s a tool and other factors should be reviewed as well. The algorithm will be published in an the journal of Applied Statistics. This was a joint effort from the University of Washington and MIT and is pretty much the first time an algorithm as such is being used in a medical setting. What makes this algorithm different than others is the fact that it shares information with experiences of other patients with similar problems.
It can give information relative to potential medical conditions that could arise in the future, and again this is a tool and not a 100% rule here. Now the next question that arises is what will insurance companies pay to get a hold of this and base their payment algorithms on such data? You can almost bet they will be right there first in line for the ability to have more data. Stokes are one of the conditions that the algorithm is said to have the ability to give information. The algorithm is also slated to be “patient friendly” so it’s not technical terminology just for clinicians.
This uses a standard Bayesian methodology which is what is used with most predictive models today. Again if a tool as such is used wisely and not over cooked with too many prevention recommendations, as Netflix doesn’t get theirs right all the time either, it could in fact be a good guidance too and furthermore, add in some genomic information and the value could rise here. This is not the first time I have blogged about Netflix and how their algorithm methodologies can be useful. Pharmaceutical companies have also engaged with the same type of algorithms, again for additional information relative to molecular drugs.
What Can Pharma Learn by Using a “Netflix” Algorithm – Molecular Ranking and the Drugs Consumers Will Take
Used as a tool and not a model to not include compensation by insurers this could change a lot of the pictures in the office of the doctor in future. With insurer use of some of their algorithms we end up with more of an attack of the killer algorithms when profits enter the picture sometimes. It happens every day on servers that run 24/7 making life impacting decisions about all of us and we don’t really know if the math and code is good as you can read that in the news every day in both healthcare and the financial areas, i.e. JP Morgan and their troubled math. Here’s more links about “killer” algorithms that deny and rely on “flawed” data for profit and they kind of take advantage of the naïve consumer in several ways too.
Attack of the Killer Algorithms–Digest & Links for All 30 Chapters–How Math and Crafty Formulas Today Running on Servers 24/7 Make Life Impacting Decisions About You–Updated 5-03-2012
A good example of an over sell of this though exists with FICO that I hope someone calls them on it with using your credit score and other data to determine if you will take your medications and they are selling this to pharma and insurance companies with scoring you.
Heck, even the algorithm that Netflix uses to determine what you will like is only 60% on target so look at what a fleecing this is with misuse of numbers. They use predictive modeling with some very far fetched mismatched and flawed data so again if this is kept in the clinical arena where it belongs and not put out there for shear profit, we may get somewhere.
I would almost bet that this algorithm will end up in front of lawmakers just as the “gene” bill did to not allow insurers and other for profit companies cash in and discriminate against patients due to the fact that “the algorithm says”. It will be entertained for sure in the “data rich” world we live in today and other substantiated reports will more than likely be created to justify such use as that’s the way the algorithm game gets played today sadly, for profit and not always for better care. BD
Analyzing medical records from thousands of patients, statisticians have devised a statistical model for predicting what other medical problems a patient might encounter.
Like how Netflix recommends movies and TV shows or how Amazon.com suggests products to buy, the algorithm makes predictions based on what a patient has already experienced as well as the experiences of other patients showing a similar medical history.
"This provides physicians with insights on what might be coming next for a patient, based on experiences of other patients.
It also gives a predication that is interpretable by patients," said Tyler McCormick, an assistant professor of statistics and sociology at the University of Washington.
The statisticians used medical records obtained from a multiyear clinical drug trial involving tens of thousands of patients aged 40 and older. The records included other demographic details, such as gender and ethnicity, as well as patients' histories of medical complaints and prescription medications.
They came up with a statistical modeling technique that is grounded in Bayesian methods, the backbone of many predictive algorithms. McCormick and his co-authors call their approach the Hierarchical Association Rule Model and are working toward making it available to patients and doctors.