THe following is worth digesting, regardless of agreement, as it lists several of the current technological efforts which could improve medical treatement and cost savings if used either as a job aide to physicians or in conjuntion with diagnostic decisions by physicians to determine ‘after the fact’ accuracy via regressions and reporting. I was very happy to learn that people started with analysis of childhood diseases. It does speak to the need though to apply this to adult diseases, and it also really, really opens up the ability to globally improve resource dedication to treatment (rather than $ dedicated toward medical analysis). ISABEL is automating differential diagnosis for pediatric illness, for positive identification of disease. Whether this is possible to scale out to generalized medicine, the important question is whether software can track actions and reduce physicians wasting cycles/$ on impossible paths of investigation, and further to allow/require rare/low probability analyses from being executed without proper foundation and historical artifacts in the software which demonstrate why a physician thought that a rare course of analysis was worth pursuit…then know if that intuition was right (in which case work the analysis into software next time). The technology also does a very important thing to prevent doctor errors: Confirmation bias, which doctors have in large amount, is reduced by computer aided decision-support. Doctors aren’t bad folks, everyone seems to have confirmation bias- but it’s really bad when your neurologist is wrong about what disease you have…it’s a lot worse than when your lawn-person mis-guesses the weeds you have.
http://www.isabelhealthcare.com/info/SB111627938932135070.html – cool WSJ 2005 article on why physicians don’t like decision analysis software, but patients might like it a lot – since they survive more often. Also, insurance companies and payors certainly would like doctors to be using decision support software; since faster and more effective treatment saves money. Also funny in that the term decision support software is being used to couch its actual function.
History and background of why isabel got created. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1172209 However, the claims in the article (Trials in four hospitals have found that in 95 out of 100 paediatric cases, the Isabel tool came up with the correct diagnosis. More extensive trials are planned for August of this year.) may not hold up because when I dug out the definition of ‘success’ it was in fact that Isabel only (still cool) presented the solution within a list of 40 diseases, that said, it was still a very strong indicator that if you can positively ID things, you can DEFINITELY preclude stupid things well (which prevents expensive, wasteful tests). Doctors lockig into the wrong diagnosis (confirmation bias) is reduced using software to confront their human tendency to see what they think confirms their diagnosis.
The company which originally produced this, and what they want to accomplish: http://us.isabelhealthcare.com/info/ourmission.htm – his daughter was nearly misdiagnosed to death.
I cannot get to this article, but the reference repeats through 3 different articles. http://linkinghub.elsevier.com/retrieve/pii/S1067502703001701
Physicians, and potential software person, need to focus some analysis on this 5 page paper, The structure they built here is cool: http://cmbi.bjmu.edu.cn/news/report/2004/medinfo2004/pdffiles/papers/5158Ramnaryan.pdf – The paper details building the taxonmony which allows Isabel to grind info….I don;’t know that the case is made for positive diagnosis, but for preventing misdiagnosis the case is really strong! I would see working with idabel’s healthcare people to simply tap what doctors SHOULDN’T do….Note: precluding bad behavior is actually MUCH EASIER and drives cost savings and human happiness. The taxonomy (cladistics if you’d prefer) and 30+ GB of data they’ve taken into the structure says they have operational analysis started (maybe not finished in the paper)
http://www.jamia.org/cgi/content/abstract/10/6/563 – Medical informatics paper on decision support – also most likely a magazine which would
interest folks in this area.
For econ tools and reading, I really like:
good resource for analytics and informatics/econometrics–http://www.analyticbridge.com/profile/JimVarrialeQUANTstercom?xgp=friend
supercrunchers background material: http://islandia.law.yale.edu/ayers/indexhome.htm
Levitt’s NYTimes opinion from Freakonomics author: http://freakonomics.blogs.nytimes.com/author/slevitt/
naturalrationality is worth checking out only if you like econ tools applied to evolution, cognition, and other biology/psychology topics: http://naturalrationality.blogspot.com/.