Yu yon "dijital twin"? Di laytst teknɔlɔji we go chenj drɔg tɛst (Digital Twin)

Yu dɔn ɛva wɔnda aw lɔng i tek fɔ mek da pil we yu tek dis mɔnin rich yu an? Fɔs, dɛn kin du bɔku risach na lɛbɔretri dɛn. Dɔn dɛn kin tɛst am pan animal dɛn. Bɔt bifo dɛn gri fɔ yuz dɛn na mɔtalman, dɛn fɔ tɛst am pan mɔtalman. Dis na wetin wi kɔl klinik trial. Na wan rili kɔmplikɛt ɛn dia dia prɔses.

Fɔ tɔk am simpul wan, wetin na klinik trial?

Fɔ tɔk am simpul wan, wan klinik trial kin wok lɛk dis: Risach pipul dɛn kin tek pipul dɛn we gɛt di sik we di nyu mɛrɛsin we dɛn dɔn mek de tɔch. Dɔn dɛn kin sheb dɛn volontia ya to tu grup dɛn we dɛn nɔ pik.

1. Dɛn kin gi di fɔs grup di nyu mɛrɛsin we dɛn fɔ tɛst.

2. Dɛn gi di sɛkɔn grup , di ``Kɔntrol Grup'', wan plesibo we nɔ gɛt ɛni aktif ingridiɛnt ɛn we tan lɛk di drɔg. Wi de kol dis ``Placebo''.

If, we dis trayal dɔn, di sayn dɛm fɔ di wan dɛm we tek di nyu mɛrɛsin dɔn bɛtɛ pas di wan dɛm we tek di plasɛbo, den dis pruv se di nyu drɔg de wok fayn.

Wan pan di big prɔblɛm dɛn we kin apin we dɛn de du dis kayn trial na fɔ fɛn inof pasɛnt dɛn we fit fɔ di trial. Sɔntɛnde ivin dɔktɔ dɛn nɔ kin no us trial dɛn we rayt fɔ dɛn pasɛnt dɛn. Dɔn bak, di kwaliti dɛn we di sik pipul dɛn we want fɔ tek pat kin gɛt nɔ kin mach wit wetin di trayal nid. Bɔt atifishal intɛlijɛns (AI) kin mek dis wok izi pasmak.

Mit yu "Digital Twin".

Imajin se yu gɛt yu yon kɔmpyuta mɔdel, dat na dijital kɔpi fɔ yusɛf. Na dat dɛn kɔl ``Digital Twin''. Dis na jɔs ``Kɔmpyuta Mɔdal'' we de simul wan rial-wɔl ɔbjɛkt ɔ sistɛm na kɔmpyuta. Bɔt we dɛn si dɛn wit nɔmba, dɛn kin biev jɔs lɛk di rial tin. Di bɛst ɛgzampul na we wan ɔksijɛn tank na di Apollo 13 spɛshal ship bin bɔm, NASA ɛp fɔ mek am fayn wit wan ``Digital Twin'' we na di sem spɛshal ship we dɛn bil na di wɔl.

Naw, sayɛnsman dɛn kin mek wan ``Digital Twin`` fɔ mɔtalman, we dɛn gi am inof ``data``. Dis de yuz wan atifishal intɛlijɛns teknɔlɔji we dɛn kɔl ``Mashin Lanin``. Fɔ tɔk am simpul wan, dis min se yu fɔ gi kɔmpyuta program bɔku bɔku data ɛn lɛ i lan frɔm am.

Dɛn kin mek Dijital Twin fɔ di pasɛnt dɛn bay we dɛn de tren dɛn Mashin Lanin mɔdel ya we dɛn de yuz data frɔm di klinik trial dɛn we dɛn bin dɔn du bifo ɛn ɛni pasɛnt in pasɔnal mɛdikal rɛkɔd. Dɔn di mɔdel kin tɔk aw di pɔsin in wɛlbɔdi go chenj ova di kɔs fɔ di trayal if dɛn gi dɛn plasɛbo.Dat min se, dɛn mek wan kɔntrol grup we dɛn mek wit kɔmpyuta (Simulated Control Group) fɔ da sem sikman de.

Naw lɛ wi si aw dis de wok. Lɛ wi se dɛn pik pɔsin we nem "Samadhi" fɔ de na di grup we de gɛt nyu drɔg. Samadhi in ``Digital Twin,'' we na kɔmpyuta mɔdel, de na di kɔntrol grup. Di mכdel de prεdikt wetin go apin if dεn gi Samadhi wan ``Placebo'' insted כf di aktual drog. Ultimately, di difrεns bitwin Samadhi in aktual rispכns to di dכg εn wetin di mכdel prεdikt go apin if i tek di ``Placebo'' na mכsu fכ aw di drog de wok fכ Samadhi.

Lɛ wi luk dis tebul fɔ si klia wan di bɛnifit dɛn we pɔsin kin gɛt we i yuz ``Digital Twins'' instead ɔf tradishɔnal kɔntrol tim dɛn.

Pan ɔl we dis teknɔlɔji gud, nɔto prɔblɛm dɛn de?

Pan ɔl we dis teknɔlɔji gɛt bɔku prɔmis fɔ tumara bambay, dɛn nɔ dɔn yuz am bɔku bɔku wan yet. Sɔntɛm gud rizin dɛn de fɔ dis. As Dr. Daniel Neal, we na masta sabi bukman pan ``(Machine Learning),'' point out, dis model dem de dipen pan big amount of data. Dɔn bak, i rili at fɔ gɛt ay kwaliti data frɔm wan wan pipul dɛn.

Wan ɔda tin na di sayd ɛfɛkt dɛn we nɔ kin apin so ɔltɛm we wan mɛrɛsin kin mek pɔsin gɛt. Bɔku tɛm, dɛn nɔ kin kɔba dɛn tin ya bay di mɔdel we dɛn mek fɔ di kɔntrol grup.

Bɔt di big tin we Dɔktɔ Nil de fred na dat dis prɛdiktiv mɔdel de bays pan "biznɛs lɛk aw i kin bi." Imajin wan big tin we yu nɔ bin de ɛkspɛkt lɛk di COVID-19 pandemic we de chenj ɔlman in bihayvya ɛn mek pipul dɛn sik. I se: “Kɔntrol mɔdel dɛn nɔ de tek dis na kɔna.” Dɔn, bikɔs ɔf dɛn tin ya we di kɔntrol grup nɔ bin tek in kɔnt, di las rizɔlt dɛn fɔ di ɛkspiriɛns kin rili difrɛn.

So yu tink se dɛn kin rili yuz dis? Wetin go apin tumara bambay?

Eric Topol, we sabi aw fɔ yuz dijital tɛknɔlɔji fɔ kia fɔ wɛlbɔdi biznɛs, tink se di aidia fayn, bɔt di tɛm nɔ rayt yet. "A nɔ tink se klinik trial go chenj insay di nɛks tɛm. Bikɔs i nid bɔku layers fɔ data, pas jɔs wɛlbɔdi rɛkɛd, lɛk di jɛnɛtik sikyud, gut maykrobayɔm, envayrɔmɛnt data."

Akɔdin to Charles Fisher, we na di wan we mek Unlearn.AI, we na payɔnia fɔ dis teknɔlɔji, sɔlv dɛn dɔn ɔlrɛdi de fɔ di prɔblɛm dɛn we gɛt fɔ du wit data prayvesi ɛn bias. "Prayvesi izi. Wi de wok onli wit data we dem don olredi anonymized."

We i de tɔk bɔt bias, i se pan ɔl we dɛn nɔ dɔn sɔlv di prɔblɛm kpatakpata, i nɔ gɛt natin fɔ du wit di tin dɛn we dɛn dɔn stɔdi. Bikɔs di stɔdi dɛn na randomized, ivin if sɔm bias de na di data, i nɔ de afɛkt di rizɔlt. Bɔt sɔm masta sabi bukman dɛn nɔ gri wit dis aidia. Dɛn se, .Bikɔs yu de kɔmpia rial pɔsin to kɔmpyuta mɔdel, if ɛni mistek de na da mɔdel de, i kin afɛkt di rizɔlt dairekt wan.

Bɔt, Unlearn.AI dɔn ɔlrɛdi de wok wit famasi kɔmni dɛm fɔ disayn klinik trial fɔ nyurolɔjik sik dɛm lɛk Alzaima, Pakinsin, ɛn maltipul sklɛrosis. Dɛn bin pik dɛn sik ya as gud ples fɔ bigin bikɔs dɛn gɛt mɔ data pas ɔda wan dɛn. Fisher se if dɛn kin yuz di we aw dɛn de du tin wan de fɔ ɔl di sik dɛn, i go rili ridyus di tɛm we dɛn kin tek fɔ mek nyu mɛrɛsin dɛn kam na makit.