Can learning health systems help organisations deliver personalised care?

CORRESPONDENCE Open Access Can lea ning heal h sys ems help o ganisa ions deli e pe sonalised ca e? B igh I. Nwa u 1,2,3 , Cha les F iedman 4 , John Halamka 5 and Aziz Sheikh 1,6,7* Abs ac The e is inc easing in e na ional policy and clinical in e es in de eloping lea ning heal h sys ems and deli e ing p ecision medicine, which i is hoped will help educe a ia ion in he quali y and sa e y o ca e, imp o e e iciency, and lead o inc easing he pe sonalisa ion o heal hca e. Al hough elian on simila policies, in o ma ics ools, and da a science and implemen a ion esea ch capabili ies, hese wo majo ini ia i es ha e hus a la gely p og essed in pa allel. In his opinion piece, we a gue ha hey should be conside ed as complemen a y, syne gis ic ini ia i es whe eby he c ea ion o lea ning heal h sys ems in as uc u e can suppo and ca alyse he deli e y o p ecision medicine ha maximises he bene i s and minimises he isks associa ed wi h ea men s o indi idual pa ien s. We illus a e his syne gy by conside ing he example o ea men s o as hma, which is now ecognised as an umb ella e m o a he e ogeneous g oup o ela ed condi ions. Keywo ds: P ecision medicine, P4 medicine, Pe sonalised medicine, S a i ied medicine, Lea ning heal h sys em, As hma In oduc ion The Human Genome P ojec and he subsequen se- quencing o he human genome d ama ically ede ined ou unde s anding o disease p ocesses, diagnosis, he a- peu ics, and p e en ion [1–4]. These ad ances laid he ounda ions o P esiden Obama’s launch in 2015 o he P ecision Medicine Ini ia i e, which aims o in eg a e he as and e e -inc easing quan i ies o genomic, biological, heal h, adminis a i e, en i onmen al, and be- ha iou al da a on indi iduals in o de o achie e mo e indi idually ailo ed decision making and pe sonalised heal hca e [5]. This Ini ia i e has gene a ed conside able en husiasm, bu i has also a ac ed some skep icism [6, 7]. The e ha e been high p o ile demons a ions o he p omise o p ecision medicine in, o example, cys ic ib osis and cance [8–12]. The disco e y and clea e unde s anding o he cys ic ib osis ansmem- b ane conduc ance egula (CFTR) gene a ian s as he cause o cys ic ib osis led o he de elopmen o i aca - o as a a ge ed d ug o pa ien s wi h cys ic ib osis [8–10]. Simila ly, he success o he ABL1 kinase inhibi o ima inib o ch onic myeloid leukaemia p o- ided a clea pla o m o he ield o oncology o mo e owa ds applica ion o molecula classi ica ion and a ocus on gene ic s a egies o cance diagnosis and he apeu ics [8, 11, 12]. P og ess is also now being made in o he disease a eas— o ins ance, in he ield o ca diology [13] and ischaemic s oke [14]. Pa allel o he p og ess made in p ecision medicine is he apid accumula ion o adminis a i e, heal hca e, and public heal h da a, pa icula ly h ough elec onic heal h eco ds (EHRs) esul ing om clinical encoun e s, heal h insu ance claims, and medica ion p esc ip ion da abases. Beyond adi ional da a cap u ing app oaches, he apid de elop- men s in echnology a e also making i possible o impo - an heal h da a o be cap u ed h ough pe sonal de ices, such as mobile phones, wea able de ices, ac i i y moni o uni s, and o he eme ging echnologies ha can be used o collec pe sonal da a in eal ime. These de elopmen s ha e g ea ly en iched he heal hca e da a space, and his, coupled wi h inc easing capabili ies in p ocessing, linking, and analysing hese dispa a e da a sou ces, is opening up new possibili ies o u ilise da a o imp o e human heal h [15, 16]. Th ough ad ances in high h oughpu compu ing and p edic i e algo i hms, machine lea ning is p o iding a pla o m o de elop ele an compu a ional algo i hms (e.g. * Co espondence: [email p o ec ed] 1 K e ing Resea ch Cen e, Depa men o In e nal Medicine, Uni e si y o Go henbu g, Go henbu g, Sweden 6 B igham and Women’s Hospi al/Ha a d Medical School, Bos on, MA, USA Full lis o au ho in o ma ion is a ailable a he end o he a icle Medicine and he Fu u e o Heal h © The Au ho (s). 2017 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e (h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed. Nwa u e al. BMC Medicine (2017) 15:177 DOI 10.1186/s12916-017-0935-0 decision ees and nes ed analy ic s uc u es) ha enable d awing in e ences om aw da a, in eal ime, wi hou he need o human inpu [15, 16]. One o he key oci o p ecision medicine lies in ede in- ing disease pa hogenesis, pa icula ly a he genomic and gene ic le els [8, 17, 18]. The e has in con as been a less p og ess in ansla ing hese insigh s in o ou ine heal hca e p ocesses ha op imise he apeu ic and p e- en i e s a egies [8, 19]. Clea ly, massi e genomic da a wi h po en ial o imp o e heal hca e decision making a e con inuously being gene a ed, bu he use ulness o such da a canno uly be app ecia ed un il hey a e success ully in eg a ed in o he heal hca e sys em in o de o imp o e human heal h [8, 19, 20]. Scien i ic b eak h oughs emain incomple e un il hey a e success ully, ou inely imple- men ed in clinical se ings [20]. P ecision medicine has hus a lacked he ools o close he loop om genomic disco e y o clinical applica ion [20]. This missing link can be illed by a hough ul syne gy o p ecision medicine and he p inciples o lea ning heal h sys ems (LHSs) [20]. LHSs, as will be desc ibed in u he de ail below, in oke as a undamen al p ecep cyclical p ocesses ha con e da a o knowledge, b ing knowledge o p ac ice, and e- u n he esul s o implemen a ion in o new da a and in- sigh s, which eed subsequen i e a ions o he cycle. In his opinion piece, we use as hma, which is now ecognised as being an umb ella e m o a he e ogeneous g oup o ela ed condi ions, as an exempla o illus a e he po en ial syne gis ic ela ionship be ween p ecision medicine and LHS in imp o ing heal hca e p ocesses and clinical decision making. We a gue ha by capi alising on he unde pinning ing edien s o implemen a ion science, he app oaches endemic o p ecision medicine and LHS can be success ully in eg a ed in o de o imp o e heal h- ca e and suppo ailo ed clinical decision making o he indi idual pa ien . In addi ion, we discuss some o he eme ging unde pinning issues ha need o be ha nessed in achie ing a syne gis ic in eg a ion o p ecision medicine and LHSs. These include, bu a e no limi ed o, iden i y- ing and accessing ele an da a se s ha need o be access- ible o analysis; ad ancing cu en elec onic da a cap u e sys ems o accommoda e he ull ange o ele an ou - come da a; achie ing da a s anda disa ion and ha monisa- ion; ad ancing compu a ional capabili ies o meaning ully in e oga e and analyse hese dispa a e da a se s; imple- men ing me hods o sys ema ic managemen and eed- back o knowledge c ea ed om da a analy ics; and c ea ing go e nance mechanisms ha allow o secu e, us wo hy use and epea ed euse o hese da a. De ini ion o concep s and ans o ma ional goals Lea ning heal h sys ems The LHS ocusses on app oaches o cap u e da a om clinical encoun e s and o he heal h- ela ed e en s, analyse he da a o gene a e new knowledge, and hen apply his knowledge o con inuously in o m and im- p o e heal h decision making and p ac ice [21–24]. This in u n equi es policies, in as uc u e, and go e nance mechanisms o suppo da a-d i en heal h lea ning and imp o emen [20, 21]. The LHS capi alises on he ad- ances made in compu a ional science in o de o de- elop algo i hms o in e oga e heal h da a, in e p e hese da a, and hen eed back he gained knowledge o he heal hca e sys ems in o de o imp o e he quali y and sa e y o ca e [21–24]. The LHS allows iden i ica ion o a - isk pa ien s, enhances s a i ica ion o he popula- ion acco ding o di e en isk p o iles, p o ides deci- sion suppo ools o clinicians, and p o ides a pla o m and in as uc u e o acili a ing he unde aking o clin- ical ials [20–24]. Mo e speci ically, wi hin he con ex o clinical ials, he LHS in as uc u es a e now being used o suppo e icien ec ui men , assess eligibili y conside - a ions, unde ake poin -o -ca e andomisa ion, assess ou - comes, and enhance long- e m ollow-up [25–28]. These bene i s a e no con ined o clinical ials; a he , he LHS in as uc u e can be used o add ess a b oad a ay o heal h p oblems, as is discussed in de ail elsewhe e [29]. As illus a ed in Fig. 1, he LHS can be iewed as a cyclical p ocess unde aken by a mul i-s akeholde communi y sha - ing in e es in sol ing a pa icula heal h- ela ed p oblem. Each cycle begins wi h con e sion o da a o knowledge (D2K), ollowed by applica ion o his acqui ed new know- ledge o ans o m p ac ice (K2P). The cap u e o p ac ice changes and he consequences o hese changes gene a e new da a, comple e he cycle, and ini ia e he nex i e a ion. Successi e i e a ions o he cycle aim o con inue o iden i y bes p ac ices and imp o e ou comes. The LHS ex ends p inciples o con inuous quali y imp o emen h ough he inclusion o go e nance and in as uc u e ha enables sys em imp o emen o occu wi h economies o scope and scale [30, 31]. A socio- echnical in as uc u e also enables he LHS o unc ion wi h economies o scale and scope. Fig. 1 F amewo k o a lea ning heal h sys em. Adap ed om F iedman e al. Yea b Med In o m. 2017;26:16–23 [57] wi h copy igh pe mission g an ed by he Publishe Nwa u e al. BMC Medicine (2017) 15:177 Page 2 o 8 By suppo ing mul iple lea ning cycles wi h se ices, in- cluding policy and echnology, he cos o execu ing N lea ning cycles is a less han N imes he cos o execu ing one cycle [32]. Because he in as uc u e p o- ides se ices ha anscend biomedical domains— ha is, he in as uc u e suppo ing an as hma-o ien ed LHS p o- ides he same se ices as a cance -o ien ed LHS—lea ning cycles can be di ec ed a any heal h p oblem. One eal-li e example o he LHS amewo k is he PINCER (pha macis -led in o ma ion echnology in e - en ion) ial, which aimed o p e en and co ec medi- ca ion e o s in gene al p ac ice [33]. Gene al p ac ice clinical sys ems we e sea ched using compu e ised p e- sc ibing sa e y indica o s o iden i y pa ien s a isk o medica ion e o s on he basis o hei p esc ip ions. Wi h pha macis suppo , he iden i ied e o s we e ac ed upon and co ec ed. The sys em was implemen ed by an expe eam, who used s uc u ed ac i i ies, in he o m o e.g. educa ion, eedback, and oppo uni ies o sha ed lea ning, o engage clinicians and pha macy eams o e ec he a ge ed imp o emen s. Imp o e- men s we e hen measu ed using anonymised ou inely eco ded da a om gene al p ac ices collec ed a h ee mon hly ime poin s. Using appealing illus a i e s a is- ics and g aphs, eedback was hen p o ided back o he Clinical Commissioning G oup and gene al p ac ices on a con inuous basis o assess hei pe o mance and enhance con inuous imp o emen (h ps://sapc.ac.uk/ con e ence/2017/abs ac /implemen ing-pince -in e en ion-eas -midlands- educe-p esc ibing-e o s). A key ea- u e o he LHS app oach illus a ed by his wo k is he mul i-s akeholde ’lea ning communi y’. Addi ionally, wi hin he con ex o as hma, we a e cu - en ly implemen ing a p o o ype LHS in Sco land, which has been de eloped wi h ele an s akeholde s (i.e. pa- ien ep esen a i es, clinicians, indus y, cha i y, policy- make s, and academics). We a e in e oga ing he EHRs o sampled gene al p ac ices ac oss Sco land in o de o e alua e each gene al p ac ice’s pe o mance agains na- ional ca e benchma ks o as hma [34]. We plan o c e- a e a con inuous in as uc u e wi h a eedback (K2P) mechanism ha will allow s akeholde s o unde ake eal- ime moni o ing o he p og ess being made ac oss impo an indica o s o as hma ca e a he gene al p ac- ice le el. Th ough linkage o he anonymised clinical da a o o he demog aphic, social, and en i onmen al da a se s, pa ien s a isk o as hma a acks (s uc u ed by a ious popula ion segmen s such as gende , age g oups, e hnici y, socio-economic s a us, e c.) will be iden i ied. Findings will hen be ed back o clinicians using no el isualisa ion ools ha will allow ailo ed, ac ionable decisions o imp o e as hma con ol and e- duce exace ba ions (h ps://clinical ials.go /c 2/show/ NCT03000491). P ecision medicine The epo o he P ecision Medicine Ini ia i e Wo king G oup o he Ad iso y Commi ee o he Na ional Ins i u e o Heal h (NIH) de ined p ecision medicine as “an ap- p oach o disease ea men and p e en ion ha seeks o maximise e ec i eness by aking in o accoun indi idual a iabili y in genes, en i onmen , and li es yle”[17]. The o e a ching goal has been desc ibed as p o iding a clea e unde s anding o he de elopmen and exp ession o dis- ease h ough a “p ecise delinea ion o he molecula , en i - onmen al, beha iou al, and o he ac o s ha con ibu e o heal h and disease”in o de o enhance mo e a ge ed diagnosis, ea men , and p e en ion s a egies [8, 17, 18]. Whils p ecision medicine has so a been cen ed a ound he use o genomic da a o achie e indi idualised clinical decision making, he P4 (i.e. p edic i e, pe sonalised, p e- en i e, and pa icipa o y) pa adigm, o en linked o p eci- sion medicine, highligh s he need o ake a mo e holis ic sys ems app oach in o de o suppo pe sonalised heal h- ca e decision making [35–39]. This ex ended pa adigm is dependen on in eg a ion o gene ic, pheno ypic, adminis- a i e, and socio ypic da a. Achie ing in eg a ion o hese da a ypes will equi e he c ea ion o deeply cha ac e ised na ional longi udinal coho s ha a e con inuously mined o suppo deli e y o pe sonalised ca e and o imp o e ca e p ocesses [35–39]. Mos concep ions o p ecision medicine lea e implici he knowledge- o-p ac ice p o- cesses so clea ly called ou in he LHS. P ecision medicine ini ia i es using as hma as an illus a i e example The concep o ‘one size does no i all’, undamen al o p ecision medicine, has been ein o ced in he con ex o as hma, indica ing ha he adi ional unde - s anding o as hma as a single disease elemen is ou - da ed [40]. Ra he , accumula ed e idence now shows ha as hma is a he e ogeneous disease cha ac e ised by a iabili y in i s pa hophysiologic mechanisms (endo- ypes) and unde lying pa ien s’cha ac e is ics (pheno- ypes) [40–43]. The he e ogenei y o as hma also mani es s in a iabili y o clinical ou comes ha in lu- ence ea men op ions [40–43]. Whils as hma has been desc ibed as a condi ion wi h bo h ype 2 and non- ype 2 immune esponses, i is he ype 2 immune esponse endo ypes ha ha e been bes cha ac e ised [40, 41]. Type 2 immune esponse endo ypes unde lie he common a opic as hma pheno ypes, cha ac e ised by eosinophilic ai way in lamma ion, inc ease in ype 2 cy okine le els, and aspi in-exace ba ed espi a o y e- sponses [40]. Subendo ypes o ype 2 immune esponse endo ype include he in e leukin (IL)-5-high, IL-13- high, and immunoglobulin E (IgE)-high endo ypes [40]. F om p og ess made in genomic p o iling, i has been demons a ed ha sub ypes o as hma a e ela ed o Nwa u e al. BMC Medicine (2017) 15:177 Page 3 o 8 dis inc molecula pe u ba ions. Tailo ing ea men o di e en as hma sub ypes/endo ypes may imp o e clin- ical ou comes o as hma and educe he isk o ad e se e en s [40, 44, 45]. Fo example, pa ien s wi h ype 2 im- mune esponse as hma endo ypes appea o espond o ea men a ge ing he IL-5, IL-13, and IgE-media ed pa hophysiological pa hways [40]. Se e al as hma bio- ma ke s ha e also been iden i ied and a e cu en ly be- ing used o a ge ea men op ions in ela ion o ype 2 immune- ela ed in lamma ion. Fo example, blood eo- sinophilia has been shown as a a ge ed bioma ke linked o co icos e oids, as well as an i-IL-4-, an i-IL- 13-, and an i-IL-5- a ge ed ea men ; spu um eosinophil le els a e also used as bioma ke s o p edic ing ea men esponses o inhaled s e oids and an i-IL-13 and an i-IL-5 ea men ; whils se um pe ios in le els a e linked o an i- IL-13 he apy, pe ios in p esen in b onchial issue has been shown as a bioma ke o eosinophilic ai way in lam- ma ion [40]. Se e al o he bioma ke s ha e also been shown o p edic ea men esponse o pa ien s wi h ype 2 immune esponse- ela ed as hma in lamma ion. Howe e , whils mos as hma bioma ke s a e being used o esea ch pu poses, hei clinical applica ion emains unce ain, as hey a e ye o be alida ed and quali ied—i.e. linking many a ailable bioma ke s wi h clea clinical end- poin s is s ill a a ious s ages o de elopmen [40]. In conside ing he applica ion o p ecision medicine o achie e pe sonalised he apy o pa ien s wi h as hma, he PRACTALL collabo a ion (join expe g oup o he Eu opean Academy o Alle gy and Clinical Immunology [EAACI] and he Ame ican Academy o Alle gy, As hma & Immunology [AAAAI]) ou lined a h ee-s ep ap- p oach (Fig. 2) [40]. As a i s s ep, he e is a need o co ec ly es ablish and e i y he diagnosis o as hma, in- cluding adequa e ea men o any co-mo bidi y. This should be ollowed by he es ablishmen o he unde - lying as hma pheno ype based on he physical cha ac e - is ics o he pa ien s. Thi d, he unde lying as hma endo ype needs o be asce ained, as a clea unde s and- ing o he pa hophysiological mechanism o a pa icula endo ype is c ucial o deli e ing a ge ed pe sonalised ea men . Finally, he e is a need o alida e known bioma ke s ha may be ela ed o as hma se e i y and clinical ou comes, as his will be essen ial o u he de- elopmen o p ima y and seconda y as hma p e en ion s a egies [36]. O e all, his model lea es he K2P ame- wo k (see Fig. 1) implici and, as such, his example does no e lec he ull po en ial o an in eg a ion o p eci- sion medicine and LHS. Link be ween p ecision medicine and LHS: heo e ical amewo k Whils he p ima y ocus o p ecision medicine has hus a been on he disco e y side o science, he e is also a need o pay a en ion o how o ansla e he disco e ies being made in o clinical p ac ice. This should be ollowed by e o s o sys ema ically e alua e ou comes a an indi- idual pa ien le el and i e a e he p ocess, as needed, wi h he o e all goal o ensu ing ha he disco e ies made a he genomic le el a e used o imp o e clinical ca e o Fig. 2 PRACTALL sugges ed s eps o implemen ing p ecision medicine in as hma. AHR ai way hype - esponsi eness, BM bioma ke s. Rep oduced om Mu a o e al. J Alle gy Clin Immunol. 2016;137:1347–58 [40] Nwa u e al. BMC Medicine (2017) 15:177 Page 4 o 8 indi idual pa ien s (see Fig. 1). The Na ional Academy o Medicine, in i s epo based on he Round able on T ans- la ing Genomic-Based Resea ch o Heal h, p o ided a sp ingboa d o discussion on he s a egies o ansla ing genomic b eak h oughs in o clinical p ac ice, emphasising he need o capi alise on he unde pinning heo e ical amewo ks in inno a ion sciences [46]. The pa hways o ansla ing esea ch indings o he clinical se ing a e complex, o en inhibi ed by known and unknown ba ie s. I is insu icien o know whe he a pa icula in e en ion wo ks in con olled en i onmen s. Heal h imp o emen equi es ha he in e en ion wo ks and can be cos - e ec i ely deli e ed in eal-wo ld se ings [20]. The LHS can p o ide he concep ual amewo k and in as uc u e pla o m o ensu e he eal- ime i e a i e applica ion o b eak h oughs in p ecision medicine o a eal-wo ld heal hca e se ing, bu he pa hway o achie - ing his may no be s aigh o wa d [20]. Chambe s and colleagues ha e sugges ed a heo e ical amewo k h ough which p ecision medicine and LHS can be syn- e gis ically in eg a ed, emphasising ha implemen a ion science may p o ide he missing link o his complex con e gence [20]. Implemen a ion science has been de- ined as “ he scien i ic s udy o me hods o p omo e he sys ema ic up ake o esea ch indings and o he e idence-based p ac ices in o ou ine p ac ice, and, hence, o imp o e he quali y and e ec i eness o heal h se ices and ca e”[37]. Heal hca e deli e y is complex, wi h a numbe o ba ie s inhibi ing he eme gence o high e iciency heal h sys ems. Implemen a ion science he e o e ocusses on iden i ying and modi ying po en ial ac o s, bo h indi idual and policy ac o s, ha end o in luence he up ake and applica ion o esea ch indings in o eal-wo ld clinical p ac ice [44–50]. We ha e buil on his amewo k o p opose he model depic ed in Fig. 3 (illus a ed o as hma). The p oposed amewo k indica es ha he ul ima e goal o a syne gy be ween p ecision medicine and LHS is o achie e a con inuously imp o ed heal h sys em in which p ecision medicine is conside ed a co e ing edien o a lea ning heal h cycle, which includes de elopmen o imp o ed and be e a ge ed diagnosis and he apy, allowing o be e decision making o each pa ien p esen ing a he clinical ca e le el. E idence de eloped using p ecision medicine app oaches, o example, he es ablishmen o ea men op ions a ge ing speci ic subendo ypes o pheno ypes o as hma, he e o e needs o be assessed on whe he such disco e ies can be ansla ed o imp o e clinical ou comes o as hma and on whe he i is easible o iden i y subg oups o as hma pa ien s o whom such a po en ial ea men op ion can be applied. Mo ing e i- dence om p ecision medicine based on gene ic a ge s will equi e linking genomic da a o clinical elec onic heal h da a [51]. This p ocess has e hical, policy, eco- nomic, and echnical dimensions ha need o be ad- d essed, and a e illus a i e o he ange o challenges ha s and be ween he cu en s a e o heal h sys ems and hei po en ial o become lea ning sys ems capable o ealising he ull po en ial o p ecision medicine. One impo an solu ion is he es ablishmen o Sa e Ha ens Fig. 3 A amewo k o a syne gy be ween p ecision medicine and LHS: he ole o implemen a ion science as a ca alys o he link. Rep oduced wi h pe mission om Chambe s e al. JAMA. 2016. 10;315(18):1941-2 [58] Copy igh © (2016) Ame ican Medical Associa ion. All igh s ese ed Nwa u e al. BMC Medicine (2017) 15:177 Page 5 o 8 ha p o ide a secu e in as uc u e o unde aking da a- in ensi e esea ch and clinical ca e. Conce ning p i acy and da a sa e y, Tai sman and colleagues ecen ly iden i- ied h ee le els o sa egua ds: physical, elec onic, and hu- man capi al [52]. A he physical le el, pa ien da a should be ea ed wi h he u mos le el o con iden iali y includ- ing making pa ien -physician consul a ion as p i a e as possible and enabling a secu e s o age o all pa ien da a. A he elec onic le el, access o pa ien da a should be use -au hen ica ed: sys ems should be p o ec ed wi h ap- p op ia e i ewalls, an i i us p og ams, and necessa y en- c yp ions; and s o age ha dwa e should be made as secu e as possible. A he human capi al le el, hi ing o da abase pe sonnel should ollow ca e ul e ing and backg ound checks; pe sonnel should unde ake app op ia e aining on in o ma ion go e nance, da a secu i y, and pa ien p i acy, and hey should be equipped wi h necessa y da a sha ing and secu i y p o ocols [52]. The amewo k embodied in Fig. 3 highligh s he s eng hs o implemen a ion science, which can se e as a ca alys o achie ing a syne gy be ween p ecision medicine and LHS. Th ough implemen a ion science, undamen al elemen s o heal hca e se ings, he cul u es ha unde pin hem, and he speci ics o each con ex can be iden i ied, analysed, and app op ia ely managed. By in ol ing and pa ne ing wi h all ele an s akeholde s, implemen a ion science capi alises on co e s a egies (planning, educa ion, inancing, es uc u ing, quali y managemen , and awa e- ness o policy con ex s) o achie e i s goal and o suppo a success ul link be ween p ecision medicine and LHS, so ha p omising indings gene a ed om genomic da a can be success ully ansla ed in o eal-wo ld se ings o im- p o e as hma ca e. This syn hesis, i achie ed, will ealise a pa adigm shi esul ing in, among o he changes, he pe cep ion o genomic da a and in o ma ion de- i ed om hem as pa o da a suppo ing ou ine ca e. Such a pa adigm shi equi es con inuous edu- ca ion o pa ien s, heal hca e p o ide s, and he gen- e al popula ion on he eme ging de elopmen s in ou unde s anding o disease and he impac hey may ha e on ou ine clinical p ac ice [13, 19]. Unde pinning in as uc u e and s a egies o scaling up Achie ing he goal o p ecision medicine ha has ele- ance o day- o-day heal hca e p ocesses and decision making, h ough he applica ion o LHS p inciples and wi h implemen a ion science as a b idge, will equi e a sha ed digi al in as uc u e o bo h p omo e pe sonalisa- ion and con inuously imp o e he o e all sys em [35, 53]. One o he i s undamen al challenges is o iden i y da a asse s (i.e. genomic, clinical, pheno ypic, socio ypic, and en i onmen al da a se s) ha a e ele an o he heal h p oblems being add essed [35, 53]. F om he heal hca e sys em pe spec i e, he e is a need o enhance cu en elec onic da a sys ems in ways ha enable hem o cap- u e and in eg a e mul iple sou ces o ou come da a, such as adminis a i e, social, and pa ien - epo ed ou comes/ expe iences (PROMS/PREMS) da a [35, 53]. The sou ces and na u e o hese da a se s a e usually dispa a e; hence, he e is a need o es ablishmen o amewo ks o da a s anda disa ion, ha monisa ion, ans o ma ion, and link- age [53]. Such ools and analysis pla o ms will bene i om being ans e able ac oss con ex s, adap able o mul iple compu a ional pla o ms, and open sou ce in o de o allow o u he de elopmen al inpu s om all s akeholde s, which will be essen ial in ealising hei ull po en ial in ad ancing he ield [53]. These ools should also suppo meaning ul isualisa ion o da a in ways ha will allow benchma king and assessing pe sonal isk. App oaches o analysis o obse a ional da a need o be obus , wi h case-mix adjus men s and p opensi y sco e echniques; whils hese ha e usually been achie ed h ough con en ional equen is s a is ical app oaches, whe e pos- sible, hey should be complemen ed by a i icial in elligence- based app oaches (e.g. machine lea ning) ha p o ide meaning ul lea ning om non-dimensional complex da a se s. These da a sys ems and he ools needed o con- inuously in e oga e hem mus be complemen ed by ’knowledge o p ac ice’in as uc u es (e.g. Ape i a [h ps://ape i a.com/], Semedy [h p://www.semedy.- com/]) ha engage o ganisa ions, heal hca e p o es- sionals, and pa ien s ac i ely in heal h and heal hca e imp o emen [35, 53]. These s a egies will h i e in a ele an policy-enabling con ex : he issues o da a owne ship, secu i y, p i acy and anonymi y, and da a sha ing need o be ca e ully ou lined, wi h he con ibu- ion o all s akeholde s [53, 54]. Ou inc easing unde - s anding o disease p ocess and heal h now clea ly e eal he complex in e - ela ionships be ween gene ic, physio- logical, social, beha iou al, and heal h sys em de e mi- nan s. Al hough inhe en ly challenging, pa icula ly as he olume and complexi y o he a ailable da a inc ease, he e is a need o de elop da a p ocessing and analy ical capabili ies ha will help b ing oge he a ious ele an da a se s, in e oga e hese da a se s o unco e he unde - lying in e ac ions, and hen in eg a e he indings in o ou ine clinical ca e [55, 56]. Conclusions The po en ial o p ecision medicine in indi idualising heal hca e decision making will no be ully ealised un il eme ging b eak h oughs a e success ully in eg a ed in o ou ine clinical ca e. We ha e p oposed a amewo k o in eg a ing accumula ing genomic da a in o he clinical encoun e in a way ha allows o a con inuous lea ning heal h cycle and imp o emen o he quali y o ca e. We ha e emphasised he ole o implemen a ion science as Nwa u e al. BMC Medicine (2017) 15:177 Page 6 o 8 an impo an ca alys in linking p ecision medicine o he heal hca e sys em. Th ough pa ne ship wi h all ele an s akeholde s, implemen a ion science needs o suppo he ansla ing o indings gene a ed om p eci- sion medicine in o eal-wo ld heal hca e se ings. Fu - he mo e, in o de o enable scale-up o an in eg a ed sys em, he unde lying digi al in as uc u e needs o p omo e pe sonalisa ion and con inuously imp o e he o e all sys em need o be sha ed: his includes iden i ica- ion o all ele an da a; ad ancemen o cu en elec- onic da a cap u e sys ems o accommoda e o he sou ces o ou come da a (e.g. adminis a i e and social da a); es ablishmen o amewo ks o da a s anda disa- ion and ha monisa ion; c ea ion o amewo ks o e - ec i e linkage o he a ious da a se s in a secu e and ans e able way; ad ancemen o he compu a ional capabili ies o meaning ully in e oga e and analyse hese da a; de elopmen o decision suppo sys ems ha will enhance ac i e pa icipa ion o o ganisa ions, heal hca e p o essionals, and pa ien s in he heal hca e p ocesses; and inally he need o c ea e a policy-enabling con ex ha de ines and se s necessa y limi s o da a owne ship, secu i y, p i acy, and da a sha ing. Abb e ia ions AAAAI: Ame ican academy o alle gy, as hma & immunology; CFTR: Cys ic ib osis ansmemb ane conduc ance egula ; D2K: Da a o knowledge; EAACI: Eu opean academy o alle gy and clinical immunology; EHR: Elec onic heal h eco d; K2P: Knowledge o p ac ice; LHS: Lea ning heal h sys em; NIH: Na ional ins i u e o heal h; P2D: Pe o mance o da a; P4: P edic i e, pe sonalised, p e en i e, and pa icipa o y; PINCER: Pha macis -led in o ma ion echnology in e en ion; PRACTALL: P ecision medicine in pa ien s wi h alle gic diseases; PREMS: Pa ien - epo ed expe iences; PROMS: Pa ien - epo ed ou comes Funding BN and AS we e suppo ed by g an s om he As hma UK Cen e o Applied Resea ch and he Fa Ins i u e. Au ho s’con ibu ions AS concei ed his pape , con ibu ed o he w i ing o he manusc ip , and edi ed a numbe o d a s. BN led he w i ing o he manusc ip . CF w o e sec ions o he manusc ip and commen ed on hose sec ions he did no w i e. JH commen ed on d a s o he manusc ip . All au ho s app o ed he inal submi ed e sion o he pape . Compe ing in e es s BN, CF, and AS ha e no ele an compe ing in e es s. JH is a membe o he Boa d o O ion Heal hca e (New Zealand). Publishe ’sNo e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps and ins i u ional a ilia ions. Au ho de ails 1 K e ing Resea ch Cen e, Depa men o In e nal Medicine, Uni e si y o Go henbu g, Go henbu g, Sweden. 2 Wallenbe g Cen e o Molecula and T ansla ional Medicine, Ins i u e o Medicine, Uni e si y o Go henbu g, Go henbu g, Sweden. 3 As hma UK Cen e o Applied Resea ch, Ushe Ins i u e o Popula ion Heal h Sciences and In o ma ics, Uni e si y o Edinbu gh, Edinbu gh, UK. 4 Depa men o Lea ning Heal h Sciences, Uni e si y o Michigan, Ann A bo , MI, USA. 5 Be h Is ael Deaconess Medical Cen e /Ha a d Medical School, Bos on, MA, USA. 6 B igham and Women’s Hospi al/Ha a d Medical School, Bos on, MA, USA. 7 As hma UK Cen e o Applied Resea ch, Cen e o Medical In o ma ics, Ushe Ins i u e o Popula ion Heal h Sciences and In o ma ics, The Uni e si y o Edinbu gh, Te io Place, Edinbu gh EH8 9AG, UK. Recei ed: 27 Feb ua y 2017 Accep ed: 22 Augus 2017 Re e ences 1. Au ay C, Caul ield T, G i in JL, Khou y MJ, Lupski JR, Schwab M. F om genomic medicine o p ecision medicine: highligh s o 2015. Genome Med. 2016;8:12. 2. G een ED, Guye MS. Cha ing a cou se o genomic medicine om base pai s o bedside. Na u e. 2011;470:204–13. 3. In e na ional HapMap Conso ium. The in e na ional HapMap p ojec . Na u e. 2003;426:789–96. 4. Mogensen J, an Tin elen JP, Foks uen S, Ellio P, an Langen IM, Mede B, e al. The cu en ole o nex -gene a ion DNA sequencing in ou ine ca e o pa ien s wi h he edi a y ca dio ascula condi ions: a iewpoin pape o he Eu opean Socie y o Ca diology wo king g oup on myoca dial and pe ica dial diseases and membe s o he Eu opean Socie y o Human Gene ics. Eu Hea J. 2015;36:1367–70. 5. Collins FS, Va mus H. A new ini ia i e on p ecision medicine. N Engl J Med. 2015;372:793–5. 6. Khou y MJ, Galea S. Will p ecision medicine imp o e popula ion heal h? JAMA. 2016;316:1357–8. 7. Khou y MJ, E ans JE. A public heal h pe spec i e on a na ional p ecision medicine coho : balancing long- e m knowledge gene a ion wi h ea ly heal h bene i . JAMA. 2015;313:2117–8. 8. Jameson JL, Longo DL. P ecision medicine —pe sonalized, p oblema ic, and p omising. N Engl J Med. 2015;372:2229–34. 9. B ewing on JJ, McPhail GL, Clancy JP. Lumaca o alone and combined wi h i aca o : p eclinical and clinical ial expe ience o F508del CFTR co ec ion. Expe Re Respi Med. 2016;10:5–17. 10. Rehman A, Baloch NU, Janahi IA. Lumaca o -i aca o in pa ien s wi h cys ic ib osis homoqygous o Phe508del CFTR. N Engl J Med. 2015;373:1783. 11. Lindeman NI, Cagle PT, Beasley MB, Chi ale DA, Dacic S, Giaccone G, e al. Molecula es ing guideline o selec ion o lung cance pa ien s o EGFR and ALK y osine kinase inhibi o s: guideline om he College o Ame ican Pa hologis s, In e na ional Associa ion o he S udy o Lung Cancee , and Associa ion o Molecula Pa hology. J Tho ac Oncol. 2013;8:823–59. 12. Blumen hal G. Nex -gene a ion sequencing in oncology in he e a o p ecision medicine. JAMA Oncol. 2015;2:13–4. 13. An man EM, Loscalzo J. P ecision medicine in ca diology. Na u e Re Ca diol. 2016;13:591–602. 14. Ros anski SK, Ma shall RS. P ecision medicine o ischemic s oke. JAMA Neu ol. 2016;73:773–4. 15. Da cy AM, Louie AK, Robe s LW. Machine lea ning and he p o ession o medicine. JAMA. 2016;315:551–2. 16. Pa ikh RB, Kakad M, Ba es DW. In eg a ing p edic i e analy ics in o high- alue ca e: he dawn o p ecision deli e y. JAMA. 2016;315:651–2. 17. Na ional Resea ch Council. Towa d p ecision medicine: building a knowledge ne wo k o biomedical esea ch and a new axonomy o disease. Washing on, DC: Na ional Academies P ess; 2011. h p://www.nap.edu/ca alog/13284/ owa d-p ecision-medicine-building-a-knowledge-ne wo k- o -biomedical- esea ch. 18. Ashley EA. Towa ds p ecision medicine. Na Re Gene . 2016;17:507–22. 19. Dzau VJ, Ginsbu g GS. Realizing he ull po en ial o p ecision medicine in heal h and heal h ca e. JAMA. 2016;316:1659–60. 20. Chambe s DA, Fee o WG, Khou y MJ. Con e gence o implemen a ion science, p ecision medicine, and he lea ning heal h ca e sys em: a new model o biomedical esea ch. JAMA. 2016;315:1941–2. 21. F iedman CP, Wong AK, Blumen hal D. Achie ing a na ionwide lea ning heal h sys em. Sci T ansl Med. 2010;2:57cm29. 22. F iedman C, Rigby M. Concep ualising and c ea ing a global lea ning heal h sys em. In J Med In o m. 2013;82:e63–71. 23. Be ns ein JA, F iendman C, Jacobson P, Rubin JC. Ensu ing public heal h’s u u e in a na ional-scale lea ning heal h sys em. Am J P e Med. 2015;48:480–7. 24. F iedman C, Rubin J, B own J, Bun in M, Co n M, Ehe edge L, e al. Towa d a science o lea ning sys ems: a esea ch agenda o he high- unc ioning Lea ning Heal h Sys em. J Am Med In o m Assoc. 2015;22:43–50. Nwa u e al. BMC Medicine (2017) 15:177 Page 7 o 8 25. an S aa TP, Dyson L, McCann G, Padmanabhan S, Bela i R, Goldac e B, e al. The oppo uni ies and challenges o p agma ic poin -o -ca e andomized ials using ou inely collec ed elec onic eco ds: e alua ions o wo exempla ials. Heal h Technol Assess. 2014;18(43):1–146. 26. Williams JG, Cheung WY, Cohen DR, Hu chings HA, Longo MF, Russell IT. Can andomised ials ely on exis ing elec onic da a? A easibili y s udy o explo e he alue o ou ine da a in heal h echnology assessmen . Heal h Technol Assess. 2003;7:1–117. 27. G an A, U e J, Nicolson DJ, Hanley J, Sheikh A, McKins y B, e al. Accep abili y and pe cei ed ba ie s and acili a o s o c ea ing a na ional esea ch egis e o enable ‘di ec o pa ien ’en olmen in o esea ch: he Sco ish Heal h Resea ch Regis e (SHARE). BMC Heal h Se Res. 2013;13:422. 28. Nwa u BI, Soyi i IN, Simpson CR, G i i hs C, Sheikh A. Building a ec ui men da abase o as hma ials: a concep ual amewo k o he c ea ion o he UK Da abase o As hma Resea ch Volun ee s. T ials. 2016;17:264. 29. P ono os PJ, Ma hews SC, Chu e CG, Rosen A. C ea ing a pu pose-d i en lea ning and imp o ing heal h sys em: he Johns Hopkins Medicine quali y and sa e y expe ience. Lea n Heal h Sys. 2017;1:e10018. 30. Clegho n GD, Head ick LA. The PDSA cycle a he co e o lea ning in heal h p o essions educa ion. J Comm J Qual Imp o . 1996;22:206–12. 31. Leis JA, Shojania KG. A p ime on PDSA: execu ing plan-do-s udy-ac cycles in p ac ice, no jus in name. BMJ Qual Sa . 2017;26:572–7. 32. F iedman CP, Rubin JC, Sulli an KJ. Towa d an in o ma ion in as uc u e o Global Heal h Imp o emen . Yea b Med In o m. 2017;26:16–7. 33. A e y AJ, Rodge s S, Can ill JA, A ms ong S, C esswell K, Eden M, e al. A pha macis -led in o ma ion echnology in e en ion o medica ion e o s (PINCER): a mul icen e, clus e andomised, con olled ial and cos - e ec i eness analysis. Lance . 2012;379:1310–9. 34. B i ish Tho acic Socie y and Sco ish In e collegia e Guidelines Ne wo k. B i ish guideline on he managemen o as hma: a na ional clinical guideline. Edinbu gh: Heal hca e Imp o emen Sco land; 2016. 35. Hood L, F iend SH. P edic i e, pe sonalized, p e en i e, pa icipa o y (P4) cance medicine. Na Re Clin Oncol. 2011;8:184–1.87. 36. Flo es M, Glusman G, B ogaa d K, P ice ND, Hood L. P4 medicine: how sys ems medicine will ans o m he heal hca e sec o and socie y. Pe Med. 2013;10:565–76. 37. Hood L, Flo es M. A pe sonal iew on sys ems medicine and he eme gence o p oac i e P4 medicine: p edic i e, p e en i e, pe sonalized and pa icipa o y. N Bio echnol. 2012;29:613–24. 38. Hood L, Balling R, Au ay C. Re olu ionizing medicine in he 21s cen u y h ough sys ems app oaches. Bio echnol J. 2012;7:992–1001. 39. Hood L, Au ay C. Pa icipa o y medicine: a d i ing o ce o e olu ionizing heal hca e. Genome Med. 2013;5:110. 40. Mu a o A, Lemanske RF, Hellings PW, Akdis CA, Biede T, Casale TB, e al. P ecision medicine in pa ien s wi h alle gic diseases: ai way diseases and a opic de ma i is —PRACTALL documen o he Eu opean Academy o Alle gy and Clinical Immunology and he Ame ican Academy o Alle gy, As hma & Immunology. J Alle gy Clin Immunol. 2016;137:1347–58. 41. King HY, DeK uy RH, Ume su DT. The many pa hs o as hma: pheno ype shaped by inna e adap i e immuni y. Na Immunol. 2010;11:557–84. 42. Bo ish L, Culp JA. As hma: a synd ome composed o he e ogeneous diseases. Ann Alle gy As hma Immunol. 2008;101:1–8. 43. Bon ield TL, Ross KR. As hma he e ogenei y and he apeu ic op ions om he clinic o bench. Cu Opin Alle gy Clin Immunol. 2012;12:60–7. 44. S eilling K, Ch is enson SA. Ta ge ing ‘ ypes: p ecision medicine in pulmona y disease. Am J Respi C i Ca e Med. 2015;191:1093–4. 45. Ke s en ETG, Koppelman GH. Pha macogene ics o as hma: owa d p ecision medicine. Cu Opin Pulm Med. 2017;23:12–20. 46. Na ional Academy o Medicine. Round able on ansla ing genomic-based esea ch o heal h. h ps://www.na ionalacademies.o g/hmd/Ac i i ies/ Resea ch/GenomicBasedResea ch.aspx. Accessed 20 Jan 2017. 47. Eccles MP, Mi man BS. Welcome o implemen a ion science. Implemen Sci. 2006;1:1. 48. Fishe ES, Sho ell SM, Sa i z LA. Implemen a ion science: a po en ial ca alys o deli e y sys em e o m. JAMA. 2016;315:339–40. 49. Tabak RG, Khoong EC, Chambe s DA, B ownson RC. B idging esea ch and p ac ice: models o dissemina ion and implemen a ion esea ch. Am J P e Med. 2012;43:337–50. 50. Powell BJ, McMillen JC, P oc o EK, Ca pen e CR, Bunge AC, Glass JE, e al. A compila ion o s a egies o implemen ing clinical inno a ions in heal h and men al heal h. Med Ca e Res Re . 2012;69:123–57. 51. Ginsbu g G. Medicial genomics: ga he and use gene ic da a in heal hca e. Na u e. 2014;508:451–3. 52. Tai sman JK, G imm CM, Ag awal S. P o ec ing pa ien p i acy and da a secu i y. N Engl J Med. 2013;368:977–9. 53. Ha ey A, B and A, Holga e ST, K is iansen LV, Leh ach H, Palo ie A, e al. The u u e o echnologies o pe sonalized medicine. N Bio echnol. 2012;29: 625–33. 54. Gos in LO, Tu ek-B esina J, Powe s M, Kozlo R, Faden R, S einaue DD. P i acy and secu i y o pe sonal in o ma ion in a New Heal h Ca e Sys em. JAMA. 1993;270(20):2487–93. 55. Eisenbe g JNS, Desai MA, Le y K, Ba es SJ, Liang S, Naumo K, e al. En i onmen al de e minan s o in ec ious disease: a amewo k o acking causal links and guiding public heal h esea ch. En i on Heal h Pe spec . 2007;115:1216–23. 56. Gomez GB, Campos PE, Buendia C, Ca camo CP, Ga cia PJ, Segu a P, e al. S udying complex in e ac ions among de e minan s o heal hca e-seeking beha iou s: sel -medica ion o sexually ansmi ed in ec ion symp oms in emale sex wo ke s. Sex T ansm In ec . 2010;86:285–91. 57. F iedman CP, Rubin JC, Sulli an KJ. Towa d an in o ma ion in as uc u e o Global Heal h Imp o emen . Yea b Med In o m. 2017;26:16–23. 58. Chambe s DA, Fee o WG, Khou y MJ. Con e gence o Implemen a ion Science, P ecision Medicine, and he Lea ning Heal h Ca e Sys em: A New Model o Biomedical Resea ch. JAMA. 2016;315(18):1941–2. doi10.1001/ jama.2016.3867. • We accep p e-submission inqui ies • Ou selec o ool helps you o ind he mos ele an jou nal • We p o ide ound he clock cus ome suppo • Con enien online submission • Tho ough pee e iew • Inclusion in PubMed and all majo indexing se ices • Maximum isibili y o you esea ch Submi you manusc ip a www.biomedcen al.com/submi Submi you nex manusc ip o BioMed Cen al and we will help you a e e y s ep: Nwa u e al. BMC Medicine (2017) 15:177 Page 8 o 8