DIS­RUPTED DE­LIV­ERY

Tech­nol­ogy is chang­ing the way providers de­liver care.

Health Data Management - - INSIDE - By John Mor­ris­sey

New tech­nolo­gies are rad­i­cally trans­form­ing pa­tient care.

In­for­ma­tion tech­nol­ogy is trans­form­ing health­care, in­creas­ingly be­ing used to rad­i­cally im­prove care and change many of the ways in which or­ga­ni­za­tions have tra­di­tion­ally prac­ticed medicine. Dig­i­tal ap­proaches are chang­ing how physi­cians and health­care sys­tems di­ag­nose dis­eases, treat pa­tients and mon­i­tor their con­di­tions on an on­go­ing ba­sis. These new it­er­a­tions are com­ing rapidly, as tech­nol­ogy en­ables care to be­come vir­tual and pa­tient-cen­tric.

Var­i­ous tech­nolo­gies such as ar­ti­fi­cial in­tel­li­gence, nat­u­ral lan­guage pro­cess­ing, med­i­cal de­vices con­nected via the In­ter­net of Things, smart­phone-based apps and more are giv­ing doc­tors myr­iad op­tions for re­vamp­ing pa­tient care.

This is dis­rupt­ing com­mon no­tions of health­care and, at the same time, coun­ter­act­ing neg­a­tive per­cep­tions some clin­i­cians may have had about in­for­ma­tion tech­nol­ogy to date, says Lyle Berkowitz, MD, a med­i­cal in­for­mati­cist and IT en­trepreneur based in Chicago.

Con­tentions that health­care IT has un­duly bur­dened physi­cians are “silly,” he says, be­cause most prob­lems are a re­sult of the way tech­nol­ogy im­ple­men­ta­tions have been done, as well as the rel­a­tively ba­sic pur­poses be­hind health­care IT.

Emerg­ing so­lu­tions that recon­cep­tu­al­ize care de­liv­ery are catch­ing on be­cause they cut straight to the need to use IT to truly im­prove care—us­ing tech­nol­ogy’s strengths to elim­i­nate plod­ding, la­bor-in­ten­sive pro­cesses that make health­care ex­pen­sive and un­nec­es­sar­ily bur­den­some.

“Too of­ten, in­no­va­tion starts with cool things we think we can do, as op­posed to [start­ing with] the prob­lems we need to solve,” says Andy Slavitt, former ad­min­is­tra­tor of the Cen­ter for Medi­care and Medi­care Ser­vices who has be­come a noted health re­form ad­vo­cate.

Slavitt now sees tech­nol­ogy be­ing used to im­prove care and sup­port clin­i­cians in their work—and he be­lieves that trend will con­tinue. “We need to in­no­vate in ways that give physi­cians and clin­i­cians and pa­tients more time to get the bet­ter re­sult,” he adds.

In­creas­ingly, those in­no­va­tions are com­ing to mar­ket to­day, dis­rupt­ing norms in health­care de­liv­ery and bring­ing new ef­fi­ciency and ca­pa­bil­i­ties to provider or­ga­ni­za­tions. The use of these tech­nolo­gies holds prom­ise for as­sist­ing clin­i­cians in mak­ing care de­ci­sions and im­prov­ing pa­tient health.

Ar­ti­fi­cial in­tel­li­gence

Clin­i­cians and hos­pi­tal ad­min­is­tra­tors are adopt­ing com­put­ing tech­nol­ogy to process data and make de­ci­sions based on the in­sights de­rived from it.

The grow­ing ac­cep­tance of smart tech­nol­ogy—ar­ti­fi­cial in­tel­li­gence, ma­chine learn­ing, pre­dic­tive an­a­lyt­ics and deep learn­ing—in­volves mak­ing use of some of the best ca­pa­bil­i­ties of com­put­ing power. These ad­vanced tech­nolo­gies can or­ga­nize data and de­rive in­sights from that in­for­ma­tion, thus help­ing sup­port clin­i­cians as they make cru­cial care de­ci­sions.

AI is in­creas­ingly be­ing in­cor­po­rated into imag­ing prod­ucts to as­sist ra­di­ol­o­gists by im­prov­ing the speed and ac­cu­racy of their di­ag­nos­tic work.

For ex­am­ple, Royal Philips has de­vel­oped In­tel­liS­pace Por­tal 9.0, the lat­est edi­tion of its com­pre­hen­sive, ad­vanced vis­ual anal­y­sis and quan­tifi­ca­tion plat­form. The prod­uct helps ra­di­ol­o­gists de­tect, di­ag­nose and fol­low up on treat­ment of dis­eases, while us­ing new ma­chine learn­ing ca­pa­bil­i­ties to sup­port the physi­cian. An­other of its prod­ucts, Il­lumeo, is an imag­ing and in­for­mat­ics tech­nol­ogy that uses adap­tive in­tel­li­gence to re­de­fine and en­hance how ra­di­ol­o­gists work with med­i­cal im­ages.

Ra­di­ol­ogy pro­fes­sion­als are in­creas­ingly see­ing new IT ca­pa­bil­i­ties ap­plied to im­age assess­ment. Early tests have sug­gested that these ap­proaches are at least as ac­cu­rate as clin­i­cians. For ex­am­ple, a re­search team led by Case Western Re­serve Univer­sity stud­ied whether a deep-learn­ing net­work ap­proach could iden­tify in­va­sive forms of breast can­cer. The net­work was “trained” by down­load­ing 400 biopsy im­ages from mul­ti­ple hos­pi­tals; then, the net­work was asked

to an­a­lyze 200 im­ages from The Can­cer Genome At­las and Univer­sity Hos­pi­tals Cleve­land Med­i­cal Cen­ter. Ac­cord­ing to Anant Mad­ab­hushi, pro­fes­sor of bio­med­i­cal en­gi­neer­ing at Case Western Re­serve and co-au­thor of the study, the net­work scored 100 per­cent ac­cu­racy in de­ter­min­ing the pres­ence or ab­sence of can­cer on whole slides.

AI, ma­chine learn­ing and other forms of ap­ply­ing com­put­ing to in­tel­li­gence are be­ing com­bined with other tech­nolo­gies and an­a­lyt­ics to help providers an­tic­i­pate pa­tient prob­lems and then head them off.

For ex­am­ple, Hamil­ton Health Sciences, based in Hamil­ton, On­tario, is us­ing tech­nol­ogy from Toronto-based ThoughtWire, which com­bines data stream­ing from In­ter­net of Things de­vices with AI for a va­ri­ety of pur­poses. In the realm of pa­tient care, the tech­nol­ogy is able to fore­warn staff of po­ten­tial Code Blue calls, us­ing a grad­ing scale to launch pre­emp­tive in­ter­ven­tions be­fore pa­tients are even in dan­ger. Code calls have been re­duced, and the or­ga­ni­za­tion has set a goal of even­tu­ally elim­i­nat­ing them al­to­gether. As a re­sult, pa­tient survival is en­hanced by in­ter­ven­ing be­fore heart ar­rests re­quire heroic mea­sures, says Mark Far­row, vice pres­i­dent and chief in­for­ma­tion of­fi­cer at Hamil­ton Health.

The ThoughtWire tech­nol­ogy also uses AI to as­sist ad­min­is­tra­tors. For ex­am­ple, it is able to cal­cu­late pro­jected staffing needs, as­sist­ing Hamil­ton Health in sched­ul­ing nurses to spe­cific units, Far­row notes.

Other AI tar­gets

Thus, ar­ti­fi­cial in­tel­li­gence can con­trib­ute on sev­eral lev­els to dis­rupt cur­rent chal­lenges in health­care. But it can make an im­me­di­ate im­pact,by first res­cu­ing clin­i­cians from over­work, says San­tosh Mo­han, who chairs the HIMSS In­no­va­tion Com­mit­tee and heads an ini­tia­tive at athenahealth called “More Dis­rup­tion Please” Labs. Mount­ing ad­min­is­tra­tive tasks “are not the best use of a doc­tor’s time, and [physi­cians] aren’t great at do­ing as­sem­bly-line check lists any­way,” says Mo­han. As a re­sult, he sees great value in us­ing tech­nol­ogy to com­put­er­ize and au­to­mate some of this “low-level work.”

For ex­am­ple, Berkowitz is co-founder and chief med­i­cal of­fi­cer of healthfinch, a new com­pany that seeks to squeeze time and cost out of some of the most or­di­nary tasks in clin­i­cal prac­tice.

“Our fo­cus is not on the most dif­fi­cult, com­plex 5 per­cent of pa­tients,” says Berkowitz, who’s also CEO of Fu­tureHealth, a pre­ci­sion medicine-based pri­mary care prac­tice and direc­tor of the Szol­losi Health­care In­no­va­tion Pro­gram at North­west­ern Medicine. “Our fo­cus is on the 80 to 90 per­cent of pa­tients who

just need rou­tine care. And if we can au­to­mate and del­e­gate a lot of that care, we can give doc­tors back a lot of time to be able to fo­cus their at­ten­tion on the more com­plex cases.”

The ob­jec­tive is to clear away rou­tine work—such as re­fill re­quests, emails about in­com­ing pa­tients and lab test or­ders—that pile up in physi­cians’ elec­tronic in­boxes says, Jonathan Baran, CEO of healthfinch.

Mo­han says AI can make an im­me­di­ate im­pact in per­form­ing “in­tel­li­gent com­put­ing” in ar­eas such as sched­ul­ing, pre­dict­ing no-shows and can­cel­la­tions, de­ter­min­ing ap­pro­pri­ate ap­point­ment lengths, and tack­ling the drudgery of in­sur­ance preau­tho­riza­tions.

Re­mote mon­i­tor­ing

Ad­vanced IT also is mak­ing in­roads in pro­vid­ing care to pa­tients who are not in tra­di­tional health­care set­tings. As providers deal with shrink­ing re­im­burse­ments, they’re seek­ing to treat pa­tients in the least ex­pen­sive care set­ting pos­si­ble.

New tech­nolo­gies are en­abling or­ga­ni­za­tions to mon­i­tor these pa­tients off­site. Such ca­pa­bil­i­ties are cru­cial for or­ga­ni­za­tions that want to bet­ter man­age their care con­tin­uum and han­dle the fi­nan­cial risk as­so­ci­ated with pop­u­la­tion health and value-based care con­tracts.

In home health­care, for ex­am­ple, sub­ject­ing vi­tal sign and other mon­i­tored data to a “lightweight ma­chine learn­ing al­go­rithm,” AI can “pre­dict the on­set of an ad­verse event and per­haps avoid an ED visit or some other bad out­come,” says Michael Joseph, CEO of Prime Di­men­sions, a health­care con­sult­ing firm.

This class of tech­nol­ogy also can “mi­croseg­ment” pa­tients along lines of what works for a very spe­cific health his­tory and med­i­cal sit­u­a­tion, by map­ping back to records of sim­i­lar cases very quickly and ap­ply­ing a care path­way ac­cord­ingly, some­thing that couldn’t be done even five years ago, says Joseph.

An­other prob­lem with de­liv­er­ing care at pa­tient homes is that a nurse can’t be present to con­stantly ob­serve pa­tient com­pli­ance with med­i­ca­tion ad­min­is­tra­tion or other treat­ment rou­tines. And if a pa­tient lapses into a neg­a­tive out­come while on a treat­ment, it’s not clear from self-re­ported data whether the prob­lem was non­ad­her­ence to treat­ment or the treat­ment it­self didn’t work.

By us­ing a front-fac­ing cam­era func­tion of a smart­phone in com­bi­na­tion with so­phis­ti­cated al­go­rithms, a plat­form cre­ated by AiCure sim­u­lates the in­ter­ac­tion of nurses. It iden­ti­fies a spe­cific med held up to the phone, “ob­serves” the med go­ing into a pa­tient’s mouth and con­firms it was in­gested, even to the ex­tent of alert­ing to pos­si­ble “cheating” if the pa­tient’s uniquely rec­og­nized face drops out of the field of vi­sion at any point, says Laura Shafner, AiCure’s chief strat­egy of­fi­cer.

Some mon­i­tor­ing ap­proaches are geared to the hos­pi­tal set­ting, en­abling clin­i­cians to iden­tify changes in pa­tient con­di­tions in real time and make ad­just­ments in treat­ment ap­proaches or in­ter­ven­tions to head off de­te­ri­o­ra­tion in pa­tient con­di­tions.

For ex­am­ple, Medtronic is of­fer­ing a com­bi­na­tion of tech­nolo­gies that en­ables pa­tient mon­i­tor­ing soft­ware with wire­less mon­i­tor­ing de­vices and cus­tom­iz­a­ble clin­i­cal de­ci­sion sup­port mo­bile ap­pli­ca­tions. The com­pany con­tends that its Vi­tal Sync mon­i­tor­ing and CDS so­lu­tion can sim­plify time-in­ten­sive pa­tient care pro­cesses, help­ing clin­i­cians pre­vent or mit­i­gate harm­ful and po­ten­tially costly ad­verse events.

Medtronic’s so­lu­tion gath­ers pa­tient phys­i­o­log­i­cal data from a va­ri­ety of wire­less and bed­side de­vices made by Medtronic or other ven­dors. Vi­tal Sync is a connectivity and re­mote pa­tient mon­i­tor­ing soft­ware so­lu­tion from Medtronic that links elec­tronic med­i­cal record sys­tems to a va­ri­ety of med­i­cal de­vices, in­clud­ing ven­ti­la­tors, capnog­ra­phy mon­i­tors, pulse oxime­ters, depth-of-con­scious­ness mon­i­tors and other de­vices. Clin­i­cians can re­motely view re­sults on any web-en­abled de­vice, in­clud­ing smart­phones.

Pa­tient in­volve­ment

Dis­rup­tive tech­nol­ogy also gives pa­tients an op­por­tu­nity to take a more ac­tive role in check­ing on their health and par­tic­i­pat­ing in their care.

For ex­am­ple, Zip­no­sis is a vir­tual care plat­form that works with a client health sys­tem—us­ing its physi­cians and la­beled with the sys­tem’s name—and acts as “an en­trée and an en­try point” to the health sys­tem, says Jon Pearce, the its CEO.

Whereas tra­di­tional telemedicine has to put both clin­i­cians and pa­tients in front of a cam­era at the same time, the Zip­no­sis plat­form works like this: A pa­tient with a sim­ple con­di­tion, such as a blad­der in­fec­tion, can go to the hos­pi­tal’s web­site and an­swer a se­ries of soft­ware-guided di­ag­nos­tic ques­tions, en­abling the ser­vice to cap­ture a full his­tory. It then ap­plies rules to sug­gest the ap­pro­pri­ate level of care for that pa­tient—rang­ing from on­line ad­vice with no need to talk to any­one, to a video con­sul­ta­tion to a clinic visit, which the ser­vice can sched­ule.

And smart­phones laced with other ad­vanced tech­nolo­gies are en­abling con­sumers to take more of their di­ag­nos­tic and care needs into their own hands.

For ex­am­ple, Duke Univer­sity re­searchers have de­vel­oped a hand­held de­vice for cer­vi­cal can­cer screen­ing that pro­duces high-qual­ity im­ages on a smart­phone or lap­top, part of an ini­tia­tive to make screen­ings more ac­ces­si­ble, easier to con­duct and less costly than stud­ies us­ing ex­pen­sive tra­di­tional equip­ment.

The wan­d­like de­vice, which is por­ta­ble and sim­ple to use, cap­tures high-qual­ity im­ages of the cervix. In fact, the pocket col­po­scope ri­vals the im­age qual­ity of the best col­po­scopes on the mar­ket but at a frac­tion of the weight, size and cost, con­tends Nimmi Ra­manu­jam, the Robert W. Carr, Jr., Pro­fes­sor of Bio­med­i­cal En­gi­neer­ing at Duke.

“The mor­tal­ity rate of cer­vi­cal can­cer should ab­so­lutely be zero per­cent be­cause we have all the tools to see and treat it— but, it isn’t,” says Ra­manu­jam. “Women do not re­ceive screen­ing or do not fol­low up on a pos­i­tive screen­ing to have col­poscopy per­formed. We need to bring col­poscopy to women so we can re­duce this.”

Not just for tech’s sake

Many ex­perts be­lieve that in­no­va­tion needs to con­tinue to fo­cus on mak­ing changes that im­prove health­care, and not just make clin­i­cians’ lives more com­plex.

Health­care’s plod­ding pro­cesses need rein­ven­tion, but in­no­va­tion with­out a plan of adop­tion won’t make a dif­fer­ence in care trans­for­ma­tion. Dig­i­tal dis­rup­tion only works if it solves long­stand­ing prob­lems and makes clin­i­cians’ lives easier, Slavitt says.

For ex­am­ple, the healthfinch ser­vice uses a cloud-based rules en­gine in­te­grated with an EHR to de­tect a task—re­fill re­quest, ap­point­ment or some other trig­ger—and re­view a patent’s chart for a va­ri­ety of ac­tiv­i­ties, such as the last ap­point­ment, medicines taken, re­cent prob­lems, re­cent lab tests and vi­tal signs. An al­go­rithm de­ter­mines what should be done, and routes the task to the right per­son.

With tech­nol­ogy de­signed to take the rules-based pro­to­col from the nurses and au­to­mate all the steps, the re­sult­ing work done by the sys­tem is handed off to the nurse, who then has the in­for­ma­tion to be as­sured that it met the pro­to­col and is ready for ac­tion, Berkowitz says. For ex­am­ple, he ex­plains, con­di­tions were met for a re­fill, con­di­tions weren’t met for rea­sons that re­quired ex­tra scru­tiny or physi­cian in­ter­ven­tion, or it was in a gray area in which not all data could be found but a nurse could fill in the rest.

All told, the au­to­ma­tion was able to re­duce the work­load of nurses by 80 per­cent, he says, which made it pos­si­ble to re­de­ploy the ex­cess nurs­ing staff to other cen­tral­ized ac­tiv­i­ties where they were needed.

What­ever the prom­ise of ad­vanced tech­nol­ogy, how­ever, dig­i­tal dis­rup­tion has its lim­its in the cur­rent cul­ture, ex­perts warn. Physi­cians have warmed to the idea of fol­low­ing ev­i­dence-based or rec­om­mended rules for di­ag­no­sis and treat­ment, but trust­ing con­clu­sions gen­er­ated by ar­ti­fi­cial in­tel­li­gence or ma­chine learn­ing is an­other mat­ter.

COVER PHO­TO­GRAPH BY JEFF SCIORTINO

PHO­TOG­RA­PHY BY JEFF SCIORTINO

Lyle Berkowitz, MD: Distrup­tive tech­nol­ogy can im­prove pa­tient care and make clin­i­cians’ lives easier.

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.