How AI Vendors Can Navigate the Health Care Industry

How I Vendors Cn Nvigte the Helth Cre Industry

I hs the potentil to significntly improve the qulity nd cost of helth cre. But s compnies design new offerings, they must tke into ccount the obstcles they will encounter in persuding s, regultors, nd pyers to ccept their offerings. This rticle identifies those obstcles nd how specific kinds of business models cn overcome them.

The doption of I in helth cre is being n by n exponentil growth of helth dt, the brod vilbility of computtionl power, nd foundtionl dvnces in mchine lerning techniques. I hs lredy demonstrted the potential to create value by reducing costs, expnding ccess, nd improving qulity. But in order for I to relize its trnsformtive potentil t scle, its proponents need business models optimized to best cpture tht vlue.

I changes the rules of business nd, s ever, there re some unique considertions in helth cre. In order to understnd these, we studied AI across 15 sets of use cases

Filure in helth cre is costly. Users of I solutions in helth cre re, therefore, more risk-verse thn their counterprts in other sectors. They require more evidence before rolling out I pplictions. This plces burdens on product development, lengthens sles cycles, nd slows doption rtes. Firms cn sidestep these issues by deploying business models tht shre in the downside risk of their I solutions.

Exscientia, which is the phrmtech compny behind the first two AI-designed molecules submitted for human trials, ddresses this by entering co-development rrngements with its phrm s tht is tie the mount it is p to how successful its molecules turn out to be down the rod. This business model mens tht Exscienti is tking on significnt portion of the risk nd is closer to those used by trditionl drug discovery firms thn it is to technology business models such s Software-as-a-Service (SaaS). While Exscienti’s business model requires more initil cpitl thn fee-bsed ppros, it llows the compny to cpture more gins when drug succeeds.

Helth systems nd pyers re lso wry of the flood of pitches they receive from I vendors nd re reluctnt to plunge full stem hed with them. Insted, they will often strt pilot projects with these vendors, which cretes dilemm: The success of I depends on nlyzing dt t scle, but pilots, by definition, re sub-scle. To ddress this chllenge nd ccelerte doption, I vendors need to ddress this risk version through their business model. t minimum, they need to be willing to put their fees t risk to show they hve t lest some skin in the gme, nd idelly they should lso be willing nd ble to tke finncil hit if their product fils to deliver s promised. s their solution mtures, however, t-risk pricing will become less necessry to close sle, but vendors whose solutions hve proven trck record should consider still using t-risk pricing in order to chrge higher prices.

There re mny structurl brriers tht inhibit the doption of new technology in helth cre, including high level of regultion, significnt mrket concentrtion, nd vested interests in existing incentive structures. While I could ultimtely brek through these brriers, mny compnies will benefit initilly from designing their business models to fit in the current prdigm.

For instnce, most cre delivery in the United Sttes tody is still compensted on the bsis of the volume of ctivity (fee for service). There re entire systems of relted billing codes for hospitl procedures, clinic visits, dignostics, nd lbs tht hve been designed round ssumptions of resources nd costs ssocited with products nd services provided by humns. Rther thn trying to chnge this system, I dignostics compnies should tke the esier pth of trying to get pyers to set up reimbursement codes similar to those used today for human radiologists.

n lterntive, of course, is to go directly to consumers. This is Apple’s approach. It hs chosen to cpture the vlue of its helth I offerings such s those tht monitor arrhythmia and falls by chrging premium price for the pple Wtch. Others such s mentl helth chot Woebot mrket directly to consumers. We expect to see mny other direct-to-consumer I-enbled helth cre offerings in moleculr dignostics, remote ptient monitoring, helth coching, nd other res.

Obtining sufficient quntities of high-qulity dt is mjor chllenge in helth cre. Tht’s becuse such dt often resides in different orgniztions nd its qulity vries.

One wy to overcome this chllenge is to use one side of business model to fund the curtion nd preprtion of dt librries. Tempus, for exmple, provides dt integrtion services to cdemic reserch centers nd hospitls, which gives it ccess to huge high-qulity librry of multi-modl dt (clinicl, rdiology, pthology) nd it offers genetic testing services to generte genomic dt. The other side of its business uses I on this dt to derive insights for providers to improve clinicl cre for specific ptients nd to life science compnies for reserch purposes.

core element of the vlue proposition of other compnies such s Lumiata nd Clarify Health is providing pltforms to ddress the curtion of dt for their s. Lumit’s offering is bsed on cpbility pckges with different levels of dt nd modeling support, while Clrify Helth’s is pckged by use cse. Both models, though, re bsed on effectively spreding the high cost of building I-redy dtsets mong mny pyer, provider, nd life science s.

Some I compnies tht hve scored erly successes hve focused on nrrow use-cses such s in rdiology nd pthology, where dt is less siloed. Even in such pplictions, though, compnies need to tke into ccount tht I dt costs re not one nd done. There will be ongoing dt costs to customize lgorithms for different popultions nd s.

The use of I is fraught with ethical considerations and associated risks. This is true in helth cre s well where use cses in ptient enggement, cre delivery, nd popultion helth re prticulrly prone to issues such s bis, filure to get pproprite ptient consent, nd violtions of dt privcy. I purveyors must proctively mitigte these risks or they will fce significnt lsh from clinicins, ptients, nd policymkers.

Bis in society is reflected in historicl helth dt nd, when not corrected, cn cuse I systems to mke bised decisions on, for instnce, who gets access to care management services or even life-saving organs for transplants. STAT found tht of 6 products clered by the U.S. Food nd Drug dministrtion (FD) from 202 to 2020 just seven reported the rcil mkeup nd just 3 reported the gender split of their study popultions. This will chnge: The FDA is developing regulatory approaches to reduce bis nd is proposing tht firms monitor nd periodiclly report on the rel-world performnce of their lgorithms.

Consequently, firms need to ensure tht the choices they mke — the s nd prtners they work with, the composition of their dt science tems (i.e., their diversity), nd the dt they collect — ll contribute to minimizing bis. Some compnies re lredy mking such chnges. For exmple, Google Health, which is working on I to revolutionize cncer screening by promising improved performnce with n lmost tenfold reduction in cost, is not only vlidting the lgorithm’s performnce in different clinicl settings but is lso mking lrge investments to ensure tht the lgorithm performs equitbly cross different rcil groups.

Helth cre is littered with exmples of best prctices tht tke mny yers to be dopted even fter being proven superior. Even I pplictions tht hve institutionl buy-in still need to get clinicins nd other frontline workers to use them, nd the pinful rollout of electronic helth records in the United Sttes over the lst decde or so, which hs mde helth cre workers wry of new informtion technology, hs only mde this job hrder. I pplictions cn be perceived s especilly thretening becuse they require chnges in fmilir workflows, impinging on clinicins’ utonomy, nd cn be seen s thret to jobs or income.

Consequently, in ddition to investing in product development, dt preprtion, nd supportive services, I compnies need to invest in chnge mngement. This includes using design thinking in the development of the product, strong trining nd onbording progrm, nd sensitive communictions (e.g., tht focuses on the benefits nd ddresses concerns bout the impcts on people’s jobs).

I is not perfect; in some situtions — especilly those tht re complex — it will fil. In helth cre, where diseses re cused by intercting genetic, socil, nd behviorl fctors, there is gret complexity. So it should not be surprising tht I in helth cre is more likely to fil thn it is in mny other industries nd the cost of filure — for instnce, misdignosis, filed drug cndidte, or mistke in prescribing mediction — is much higher.

Therefore, it is often necessry to involve humns in the loop to ccept or reject decisions mde by I. Firms building nd selling I-bsed systems need to fctor the cost of this humn expertise into their pricing. One compny tht hs done this is AliveCor, whose direct-to-consumer electrocrdiogrm (EKG) device uses I to interpret EKG redings tht consumer tkes by using reltively chep device pired with cell phone pp. When the I sees n “edge cse” (n uncommon cse tht it might not hve seen before) or finds n issue tht requires clinicin’s input, it prompts the user to consider hving clinicin tke second look — for fee of course.

Where it is not possible to pss on this dded cost of the humn intervention, compnies should limit the scope of the product. Buoy Health took this pproch with its populr I-bsed symptom checker. Its I chot engges ptient nd suggests likely dignoses long with nvigtion to the most pproprite point of cre, which could be telehelth, urgent cre, the room, or the ptient’s primry cre doctor. In ech of these cses, Buoy is choosing to let others provide the costly humns in the loop, llowing it to mintin low-cost model.

I hs enormous potentil in helth cre. But to succeed with their offerings, compnies need to tilor their business models to the chrcteristics of their prticulr offering. One size does not fit ll.

How I Vendors Cn Nvigte the Helth Cre Industry



Happiness is for those who plan well and pursue. A profound among us have been proven those who have true dream to live for likely REALIZED IT. It is just simply the person working toward the DREAM days and night until accomplishment. There is a phrase of efficiency a head of you. Steps and obstacles at first seem tremendous. However, just with some times those difficult steps and challenges are so easy performance for you. There are also plenty of tools including VISUALIZATIONS and helps are around you.

You May Also Like…


Submit a Comment