An­a­lyt­ics Ris­ing

Providers are em­brac­ing SaaS mod­els to meet value de­mands.

Health Data Management - - INSIDE FEATURES - By Greg Sla­bod­kin

Cloud-based an­a­lyt­ics sup­port providers in of­fer­ing bet­ter care.

As health­care or­ga­ni­za­tions in­creas­ingly take on risk-based con­tracts, cloud-based an­a­lyt­ics are en­abling providers to break down data si­los and gain vis­i­bil­ity into care de­liv­ery pro­vided across their en­ter­prises in order to achieve bet­ter clin­i­cal out­comes at the best pos­si­ble value.

By har­ness­ing an­a­lyt­ics in the cloud, these or­ga­ni­za­tions are lever­ag­ing tools that help de­liver ac­tion­able in­sights ac­cessed through the In­ter­net with­out hav­ing to in­vest in their own costly, on-premise in­fra­struc­ture. This kind of soft­ware-asa-ser­vice en­sures that providers are able to store, ac­cess and an­a­lyze a plethora of clin­i­cal, claims, risk strat­i­fi­ca­tion and other data re­sult­ing in best prac­tices and strate­gies for qual­ity im­prove­ment.

Gain­ing shared sav­ings

Ac­count­able Health Part­ners (AHP), a clin­i­cally in­te­grated net­work of hos­pi­tals and physi­cians in Rochester, N.Y., has suc­cess­fully moved to new re­im­burse­ment mod­els us­ing a cloud-based data re­pos­i­tory and an­a­lyt­ics plat­form from Ar­ca­dia Health­care So­lu­tions to sup­port its pop­u­la­tion health man­age­ment ini­tia­tives.

AHP’s cloud-based pop­u­la­tion health plat­form an­a­lyzes the data to cal­cu­late hun­dreds of qual­ity mea­sures, cost uti­liza­tion, risk scores and clin­i­cal gaps. As a re­sult, its physi­cians make bet­ter de­ci­sions and sup­port stronger col­lab­o­ra­tions with health plans on risk-based con­tracts and pay-for-per­for­mance qual­ity pro­grams.

Ac­cord­ing to LaRon Rowe, direc­tor of in­for­ma­tion man­age­ment at AHP, the ac­count­able care or­ga­ni­za­tion in­cludes the Univer­sity of Rochester Med­i­cal Cen­ter and more than 40 in­de­pen­dent com­mu­nity pri­mary care prac­tices.

While the Univer­sity of Rochester Med-

ical Cen­ter has an Epic elec­tronic health record sys­tem, Rowe points out that the more than 40 other prac­tices use 10 dif­fer­ent EHRs—which pre­sented some tech­ni­cal chal­lenges. How­ever, AHP taps into the Ar­ca­dia plat­form in the cloud, en­abling the ACO to cen­trally plan and drive clin­i­cal out­comes across their het­ero­ge­neous, geo­graph­i­cally dis­persed net­work of hos­pi­tals and ru­ral in­de­pen­dent physi­cians.

“One of the first things that AHP did was we en­tered into a shared-sav­ings agree­ment with one of our com­mer­cial pay­ers here in town,” says Rowe. “We needed to not just mea­sure cost but qual­ity. We also have risk mod­el­ing built into the sys­tem that helps us come up with a risk score based on an al­go­rithm, al­low­ing our care man­agers to fig­ure out who has the po­ten­tial of be­ing the sick­est pa­tients and to co­or­di­nate ap­pro­pri­ate care.”

Un­der a three-year Ac­count­able Cost and Qual­ity Ar­range­ment (ACQA) with Ex­cel­lus BlueCross BlueShield, qual­ity mea­sures are re­ported via the Ar­ca­dia an­a­lyt­ics dash­board in an ef­fort to share re­spon­si­bil­ity for pro­vid­ing co­or­di­nated care to pa­tients to im­prove qual­ity in­di­ca­tors—such as can­cer screen­ing rates, hy­per­ten­sion and di­a­betes con­trol—and re­duce un­nec­es­sary health­care costs.

“It’s a pop­u­la­tion health sys­tem that helped us as a net­work mea­sure truly how we’re do­ing on qual­ity, ag­gre­gat­ing data into a sin­gle source of truth that en­ables suc­cess across dis­parate prac­tices,” adds Rowe. “We’re also able to in­te­grate claims data from those lo­cal pay­ers and bring it into a sin­gle sys­tem with clin­i­cal data.”

He says AHP is also fo­cused on op­ti­miza­tion for its value-based con­tracts by im­prov­ing its work­flows to bet­ter cap­ture qual­ity mea­sures, adding that when data is shared and pre­sented in a trans­par­ent, work­flow-rel­e­vant for­mat, physi­cians trust the pa­tient in­for­ma­tion seen at the point of care.

More than claims data

Beth Is­rael Dea­coness Care Or­ga­ni­za­tion (BIDCO), a Mas­sachusetts ACO and value-based hos­pi­tal and physi­cian net­work, has also im­ple­mented the Ar­ca­dia plat­form to en­able real-time, cloud-based an­a­lyt­ics aimed at im­prov­ing pop­u­la­tion by in­te­grat­ing both clin­i­cal and claims data.

“What we’re try­ing to do is get away from man­ag­ing risk pop­u­la­tions sim­ply through claims data—which, in our case, has lagged any­where from 90 to 150 days,” says Bill Gil­lis, BIDCO’s CIO.

BIDCO mem­bers, which in­clude nine hos­pi­tals and about 2,600 physi­cians, cur­rently op­er­ate more than 40 dif­fer­ent EHR sys­tems, and try­ing to in­te­grate data from all the var­i­ous hos­pi­tals and physi­cian groups across Mas­sachusetts was a daunt­ing chal­lenge. “We’ve got a very het­ero­ge­neous net­work from an EHR and clin­i­cal in­for­ma­tion sys­tem per­spec­tive,” adds Gil­lis.

How­ever, us­ing Ar­ca­dia’s EHR in­te­gra­tion process and tech­nol­ogy—called Data Con­nect—the or­ga­ni­za­tion has been able to ex­tract the clin­i­cal data needed to meet risk con­tract re­quire­ments for qual­ity out­comes and fi­nan­cial per­for­mance.

“The EHRs in our net­work now pro­vide real-time data for us to ac­tu­ally do some­thing with the an­a­lyt­ics,” Gil­lis says. “We get that data nightly in a batch process, and we pull it into our pop­u­la­tion health plat­form, which gets mar­ried with claims data as well as sched­ul­ing in­for­ma­tion and [ad­mis­sion, dis­charge and trans­fer] to give a real pic­ture to our care teams of what’s go­ing on in the net­work with our pa­tients.”

“If you’re a physi­cian or hos­pi­tal that’s in a value-based con­tract, those in­sights are im­por­tant to en­sur­ing that health plans are mea­sur­ing you in a timely man­ner and that all those im­por­tant in­sights around qual­ity are cap­tured in their an­a­lyt­ics so that your out­comes are not un­der­stated or mis­stated,” says Eric Sul­li­van, se­nior vice pres­i­dent of in­no­va­tion and data strate­gies at Ino­valon, a cloud-based data an­a­lyt­ics vendor fo­cused on health­care.

Gil­lis notes that BIDCO “has no ac­tual in­ter­nal in­fra­struc­ture, aside from the desk­tops, PCs and net­work com­po­nents in the closet—there are no servers here run­ning data­bases; ev­ery­thing we have is up in the cloud.” He asks: “Why would I own this in­fra­struc­ture, build it and have to main­tain it when I can go to the cloud?”

“As de­mand goes up and needs in­crease, be­ing able to scale the an­a­lyt­ics quickly is def­i­nitely some­thing a cloud­based so­lu­tion has helped us with ver­sus an on-premise sys­tem,” adds Rowe.

Sul­li­van agrees that scal­a­bil­ity and con­fig­ura­bil­ity are ma­jor ben­e­fits of the cloud. “You as a provider don’t have to down­load soft­ware and get up­dates—that au­to­mat­i­cally oc­curs,” he says.

When it comes to cy­ber­se­cu­rity, Gil­lis doesn’t worry about data stored in the cloud. He rec­om­mends that providers go with a rep­utable cloud vendor that is a “known and trusted en­tity” in the in­dus­try.

Data be­come in­sights

In many ways, cloud com­put­ing is be­ing trans­formed through ar­ti­fi­cial in­tel­li­gence. It’s no sur­prise that cloud gi­ants like Ama­zon, Google, IBM and Mi­crosoft are all mak­ing AI tools a part of their ser­vice of­fer­ings.

When it comes to AI, Rowe sees tremen­dous po­ten­tial for ma­chine learn­ing and nat­u­ral lan­guage pro­cess­ing to pro­vide real-time anal­y­sis of clin­i­cal data and claims data for quicker in­sights. In par­tic­u­lar, he ob­serves, NLP holds great prom­ise for un­lock­ing the value of vast troves of un­struc­tured data hid­den within EHRs.

Cloud-based an­a­lyt­ics vendor Ino­valon is col­lab­o­rat­ing with the Univer­sity of Mary­land’s Cen­ter for Health In­for­ma­tion and De­ci­sion Sys­tems as part of its on­go­ing de­vel­op­ment of NLP, ma­chine learn­ing and deep learn­ing so­lu­tions. One goal is to ad­vance the abil­ity to per­form ul­tra­high-speed anal­y­sis of un­struc­tured data con­tained within raw clin­i­cal doc­u­men­ta­tion, such as that found within EHRs.

“More than two-thirds of what is re­ally clin­i­cally rel­e­vant is in un­struc­tured text and un­cod­i­fied fields,” says Sul­li­van. “If you can have a ma­chine scroll through a 300-page med­i­cal record in a frac­tion of a se­cond and iden­tify five po­ten­tial places where the physi­cian is in­di­cat­ing that they may have some di­a­betes com­pli­ca­tions, that is in­cred­i­bly more ef­fi­cient than hav­ing a hu­man try to iden­tify those ar­eas.”

“Ar­ti­fi­cial in­tel­li­gence can be used to help with dis­ease pre­dic­tion, iden­tify high-risk pa­tients and pre­ven­ta­tive ther­a­pies,” con­tends Véronique Grenon, vice pres­i­dent of risk an­a­lyt­ics of The Risk Au­thor­ity Stan­ford, as well as the direc­tor of risk an­a­lyt­ics for Stan­ford Health Care and Stan­ford Chil­dren’s Health. “Risk man­age­ment can also help with automating and op­ti­miz­ing hos­pi­tal op­er­a­tions.”

Iden­ti­fy­ing risk

The Risk Au­thor­ity Stan­ford, cre­ated from the hos­pi­tal risk man­age­ment depart­ment serv­ing the Stan­ford Univer­sity School of Medicine, Stan­ford Health Care and Stan­ford Chil­dren’s Health, has de­vel­oped a plat­form for hos­pi­tals that lever­ages ma­chine learn­ing and NLP al­go­rithms to clas­sify mil­lions of data points into cat­e­gories that can iden­tify key ar­eas of risk.

TRA Stan­ford’s plat­form, called In­novence Pulse, pro­vides a suite of tools that de­liver ev­i­dence-based data on de­mand to the hos­pi­tal in­dus­try. The risk man­age­ment soft­ware an­a­lyzes dis­parate data sets such as in­ci­dent re­ports, loss runs, pa­tient com­plaints and net pa­tient rev­enue so users can man­age ex­pec­ta­tions and claims in a given area, while tak­ing ac­tions to ad­dress is­sues and pre­vent fu­ture losses.

“It uses ma­chine learn­ing and nat­u­ral lan­guage pro­cess­ing to read the un­struc­tured free text of an event and then cat­e­go­rizes that event through our Stan­ford Risk Lex­i­con, which pro­vides ac­cu­rate risk clas­si­fi­ca­tions, de­scrip­tions and re­ports that make in­for­ma­tion ac­tion­able,” says Ran­dall Smith, prod­uct man­ager for In­novence Pulse. For in­stance, in the case of an in­fec­tion that goes through a cer­tain unit of a hos­pi­tal, Smith says the plat­form can iden­tify that emerg­ing risk and en­able the provider or­ga­ni­za­tion to in­ter­vene be­fore it po­ten­tially af­fects more pa­tients.

Bea­z­ley, an un­der­writer of hos­pi­tal pro­fes­sional li­a­bil­ity in­surance, is us­ing In­novence Pulse to iden­tify trends in past claims and gen­er­ate ac­tion­able data in real time to in­crease pa­tient safety. Its claims data­base is one of the largest in the in­surance in­dus­try and in­cludes nearly 900,000 unique loss records dat­ing back two decades.

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.