The Case for Computer-Based Health Care

A participant works on a computer during the Algorithm competition at the Imagine Cup 2006 in Agra, some 200 kms from New Delhi, 08 August 2006. University students representing 42 countries participated in a Microsoft technology contest showcasing innovations they hoped would help people live healthier lives.
National Journal
Darius Tahir
See more stories about...
Darius Tahir
Oct. 16, 2013, 5:37 a.m.

The vic­tory of Wat­son, an ar­ti­fi­cial-in­tel­li­gence sys­tem de­signed to dom­in­ate the quiz show Jeop­ardy!, over the coun­try’s best nerds in 2011 may not be the equal of John Henry strug­gling against a steam-powered drill in the an­nals of man versus ma­chine. But the re­place­ment of Jeop­ardy!‘s hu­man com­pet­it­ors with a com­puter al­gorithm may sig­nal a trend that could soon spread through the health care sec­tor as Obama­care is im­ple­men­ted.

That’s the proph­ecy of ven­ture cap­it­al­ist Vinod Khosla. Khosla, a prom­in­ent Sil­ic­on Val­ley in­vestor, has pre­dicted that com­puters will re­place 80 per­cent of what doc­tors do in a couple of dec­ades. The shift could counter an­oth­er health-sec­tor trend: stag­nant pro­ductiv­ity, which the Af­ford­able Care Act aims to ad­dress with fin­an­cial in­cent­ives for ef­fect­ive, ef­fi­cient care, and which could en­cour­age a move to­ward di­git­al doc­tor­ing.

Between 1990 and 2010, pro­ductiv­ity in the health care sec­tor de­clined by 0.6 per­cent an­nu­ally as em­ploy­ment in­creased by 2.9 per­cent, ac­cord­ing to Robert Kocher, now a ven­ture cap­it­al­ist at Ven­rock, in an Oc­to­ber 2011 ed­it­or­i­al in the New Eng­land Journ­al of Medi­cine. In­creas­ing pro­ductiv­ity might bridge this dis­con­nect, and com­puters could be part of the solu­tion.

Khosla, who sup­ports the move to com­puter-based health care, notes the hu­man frailties that weak­en doc­tors’ dia­gnoses and treat­ment: The brain is biased, for­get­ful, and lim­ited. As a res­ult, dia­gnoses are of­ten in­con­sist­ent. Khosla cites a study in which psy­cho­lo­gists were asked to dia­gnose pa­tients’ ma­jor de­press­ive dis­order. On a scale where 0 rep­res­en­ted total dis­agree­ment and 1 rep­res­en­ted total agree­ment, the psy­cho­lo­gists rated 0.3.

Hu­man brains take in less data than their di­git­al coun­ter­parts. “It’s a simple fact that most doc­tors couldn’t pos­sibly read and di­gest all of the latest 5,000 re­search art­icles on heart dis­ease,” Khosla writes. “In fact, most of the av­er­age doc­tor’s med­ic­al know­ledge is from when they were in med­ic­al school, and cog­nit­ive lim­it­a­tions pre­vent them from re­mem­ber­ing the 10,000+ dis­eases hu­mans can get.” As the amount of in­form­a­tion in­creases—there’s more re­search, and more sensors to col­lect it—di­git­al sup­port pro­cessing the data could be a big help.

Khosla pre­dicts that com­puters will take over large por­tions of the med­ic­al pro­cess, leav­ing hu­mans to do em­path­ic tasks, such as re­as­sur­ing and coach­ing pa­tients. The be­gin­nings of that trend are here already. Sev­er­al start-ups of­f­load doc­tors’ tasks onto com­puters; Eye­N­etra, which Khosla has in­ves­ted in, uses soft­ware and a device that at­taches to a smart­phone to de­term­ine the strength of pre­scrip­tion lenses a pa­tient re­quires.

This sum­mer, the Na­tion­al In­sti­tutes of Men­tal Health fun­ded a round of grants to ex­plore us­ing mo­bile devices to treat men­tal health prob­lems. Akili In­ter­act­ive, a video-game maker that re­ceived NIMH fund­ing, com­bines tasks that re­quire fine mo­tor skills and visu­al at­ten­tion. The games “ac­tu­ally be­come a very sens­it­ive meas­ure of cog­ni­tion,” com­pany cofounder Ed­die Mar­tucci said at a May con­fer­ence. Akili is also hop­ing to treat ma­jor de­press­ives, who tend to struggle with prob­lem solv­ing. Solv­ing prob­lems in a game might con­di­tion these pa­tients to bet­ter solve prob­lems in real life, without a doc­tor’s in­ter­ven­tion.

Oth­er al­gorithmic soft­ware aims to aid phys­i­cians’ de­cisions rather than re­place them. “Clin­ic­al de­cision sup­port soft­ware” ana­lyzes data, of­ten from a pa­tient’s elec­tron­ic health re­cords, and ad­vises doc­tors as they pre­scribe a treat­ment course. The soft­ware could note, for ex­ample, that two drugs shouldn’t be ad­min­istered to­geth­er due to their harm­ful in­ter­ac­tions.

Soft­ware can also be used to im­prove doc­tors’ ad­her­ence to clin­ic­al guidelines. The Health­Part­ners hos­pit­al sys­tem in Min­nesota found lim­ited suc­cess with a soft­ware tool that showed doc­tors how well their or­der of a scan for a pa­tient stacked up to Amer­ic­an Col­lege of Ra­di­ology guidelines. The tool brought only mod­est in­creases in the doc­tors’ or­der­ing of evid­ence-based scans, ac­cord­ing to a study pub­lished in the Amer­ic­an Journ­al of Man­aged Care in 2010, but there were no in­cent­ives for the doc­tors to or­der more of these ap­pro­pri­ate tests. Hos­pit­als could of­fer more re­wards to en­sure ad­her­ence to best prac­tices.

Health care work­ers have also star­ted per­form­ing em­path­ic tasks based on al­gorithmic ad­vice. Jef­frey Bren­ner, ex­ec­ut­ive dir­ect­or of non­profit Cam­den Co­ali­tion of Health­care Pro­viders, re­ceived a Ma­cAr­thur Found­a­tion “geni­us grant” for his ap­proach to de­liv­er­ing bet­ter care at lower cost. Bren­ner sent so­cial work­ers to cer­tain “hot spots,” which were iden­ti­fied by soft­ware as places where a small minor­ity of pa­tients con­sumes a dis­pro­por­tion­ate amount of health care re­sources. These pa­tients of­ten have chron­ic dis­eases that aren’t treated prop­erly, and these pa­tients fre­quently end up in the hos­pit­al. The goal of “hot spot­ting” is to in­ter­vene early be­fore prob­lems flare, mak­ing the pa­tient health­i­er des­pite us­ing few­er re­sources. So­cial work­ers can as­sist by form­ing re­la­tion­ships with pa­tients and help­ing them man­age their ill­nesses.

Pro­viders are ad­opt­ing this ap­proach in re­sponse to new pay­ment in­cent­ives ushered in by the 2010 health re­form law, which im­poses pen­al­ties on hos­pit­als with high read­mis­sion rates. Mount Sinai in New York is one in­sti­tu­tion send­ing so­cial work­ers to troubled pa­tients. A 600-pa­tient pi­lot study cut emer­gency-room vis­its in half between Septem­ber 2010 and May 2012, the hos­pit­al says.

So in­stead of be­ing re­placed—what John Henry fought against when he raced the steam drill—health care work­ers can per­haps work along­side the new com­puter over­lords.

What We're Following See More »
More People Watched Trump’s Acceptance Speech
1 days ago

Hillary Clinton hopes that television ratings for the candidates' acceptance speeches at their respective conventions aren't foreshadowing of similar results at the polls in November. Preliminary results from the networks and cable channels show that 34.9 million people tuned in for Donald Trump's acceptance speech while 33.3 million watched Clinton accept the Democratic nomination. However, it is still possible that the numbers are closer than these ratings suggest: the numbers don't include ratings from PBS or CSPAN, which tend to attract more Democratic viewers.