Iingcali ze-AI zixoxa ngendlela yokudibanisa i-AI eqinileyo kwinkonzo yezempilo, isizathu sokuba intsebenziswano phakathi kwezifundo ezahlukeneyo ibalulekile, kunye namathuba e-AI yokuvelisa kuphando.
UFeifei Li kunye noLloyd Minor banike intetho yokuvula kwi-RAISE Health Symposium yokuqala kwiStanford University School of Medicine ngoMeyi 14. USteve Fish
Uninzi lwabantu abathinjwe bubukrelekrele bokwenziwa baye baba nohlobo oluthile lwexesha elithi "aha", bevula iingqondo zabo kwihlabathi lamathuba. Kwi-RAISE Health Symposium yokuqala ngoMeyi 14, uLloyd Minor, MD, umphathi weSikolo sezonyango kwiYunivesithi yaseStanford kunye nosekela-mongameli wemicimbi yezonyango kwiYunivesithi yaseStanford, wabelane ngembono yakhe.
Xa omnye umntwana oselula onomdla wacelwa ukuba ashwankathele oko akufumanisileyo malunga nendlebe yangaphakathi, wajika waya kwi-generative artificial intelligence. “Ndabuza, ‘Yintoni i-superior canal dehiscence syndrome?’ UMinor uxelele abathathi-nxaxheba abamalunga nama-4,000 kwi-symposium. Kwimizuzwana embalwa, kwavela iziqendu ezininzi.
“Zilungile, zilungile kakhulu,” utshilo. “Ukuba olu lwazi luqokelelwe kwinkcazo emfutshane, echanekileyo ngokubanzi kwaye ebekwe phambili ngokucacileyo yesi sifo. Oku kuyamangalisa kakhulu.”
Abaninzi babenomdla ofanayo nokaMinor kwisiganeko sesiqingatha sosuku, esasiyimvelaphi yesicwangciso se-RAISE Health, iprojekthi eyaqaliswa yiStanford University School of Medicine kunye neStanford Institute for Human-Centered Artificial Intelligence (HAI) ukukhokela ukusetyenziswa kobukrelekrele bokwenziwa. kuphando lwezonyango, imfundo, kunye nokhathalelo lwezigulane. Izithethi zihlolisise ukuba kuthetha ukuthini ukusebenzisa ubukrelekrele bokwenziwa kunyango ngendlela engeyonto nje kuphela eluncedo koogqirha kunye nososayensi, kodwa ecacileyo, enobulungisa nelinganayo kwizigulane.
“Sikholelwa ukuba le yiteknoloji ephucula amandla abantu,” utshilo uFei-Fei Li, uprofesa wesayensi yekhompyutha kwiStanford School of Engineering, umlawuli weRAISE Health kunye neprojekthi yeMinor kunye nomlawuli-mlingane weHAI. Isizukulwana emva kwesinye, ubuchwepheshe obutsha bungavela: ukusuka kulandelelwano olutsha lweemolekyuli zee-antibiotics ukuya kwimephu yezinto eziphilayo kunye nokutyhila iindawo ezifihlakeleyo zebhayoloji esisiseko, i-AI ikhawulezisa ukufunyanwa kwesayensi. Kodwa ayikuko konke oku okuluncedo. “Zonke ezi zicelo zinokuba nemiphumo ebengalindelekanga, kwaye sifuna izazinzulu zekhompyutha eziphuhlisa kwaye zisebenzise [ubukrelekrele bokwenziwa] ngokuthembekileyo, zisebenzisana nabantu abahlukeneyo abachaphazelekayo, ukusuka koogqirha kunye neengcali zokuziphatha… ukuya kwiingcali zokhuseleko nakwezinye iindawo,” utshilo. “Amanyathelo afana neRAISE Health abonisa ukuzinikela kwethu koku.”
Ukuhlanganiswa kwamacandelo amathathu eStanford Medicine—iSikolo sezoNyango, iStanford Health Care kunye neStanford University School of Child Health Medicine—kunye nonxibelelwano lwayo nezinye iindawo zeStanford University kuyibeke kwimeko apho iingcali zijongene nophuhliso lwengqondo yokwenziwa. imiba yolawulo kunye nokuhlanganiswa kwicandelo lezempilo kunye nezonyango. Ingoma ithi.
“Sikwindawo entle yokuba ngabavulindlela kuphuhliso kunye nokuphunyezwa okunenkathalo kobukrelekrele bokwenziwa, ukusuka ekufumaneni izinto ezisisiseko zebhayoloji ukuya ekuphuculeni uphuhliso lwamayeza nokwenza iinkqubo zovavanyo lweklinikhi zisebenze ngakumbi, ukuya kuthi ga ekunikezelweni kweenkonzo zempilo. Unyango. Indlela inkqubo yezempilo esekwe ngayo,” utshilo.
Izithethi ezininzi zigxininise ingcamango elula: ukugxila kumsebenzisi (kule meko, isigulane okanye ugqirha) kwaye yonke enye into iya kulandela. “Ibeka isigulane kwindawo ephambili kuyo yonke into esiyenzayo,” utshilo uGqr. Lisa Lehmann, umlawuli we-bioethics kwiBrigham and Women's Hospital. “Kufuneka siqwalasele iimfuno zabo kunye nezinto eziza kuqala kubo.”
Ukusuka ekhohlo ukuya ekunene: Umsasazi weendaba ze-STAT uMohana Ravindranath; uJessica Peter Lee woPhando lweMicrosoft; uSylvia Plevritis, uprofesa wesayensi yedatha yezonyango, uxoxa ngendima yobukrelekrele bokwenziwa kuphando lwezonyango.
Izithethi ezazikwiphaneli, ezaziquka uLehmann, ingcali yezonyango kwiYunivesithi yaseStanford uMildred Cho, MD, kunye neGosa eliyiNtloko leKlinikhi likaGoogle uMichael Howell, MD, ziphawule ubunzima beenkqubo zezibhedlele, zigxininisa imfuneko yokuqonda injongo yazo ngaphambi kokuba naluphi na ungenelelo. Zisebenzise kwaye uqinisekise ukuba zonke iinkqubo eziphuhlisiweyo ziyabandakanya kwaye umamele abantu ezenzelwe ukubanceda.
Isitshixo esinye kukucaca: kwenza kucace ukuba idatha esetyenziselwa ukuqeqesha i-algorithm ivela phi, yintoni injongo yokuqala ye-algorithm, kunye nokuba idatha yesigulane esizayo iya kuqhubeka nokunceda i-algorithm ifunde na, phakathi kwezinye izinto.
“Ukuzama ukuxela kwangaphambili iingxaki zokuziphatha ngaphambi kokuba zibe nzulu [kuthetha] ukufumana indawo efanelekileyo apho wazi khona ngokwaneleyo malunga netekhnoloji ukuze ube nokuzithemba kuyo, kodwa kungekhona ngaphambi kokuba [ingxaki] isasazeke ngakumbi kwaye uyisombulule ngokukhawuleza.” , utshilo uDenton Char. Umviwa weSayensi yezonyango, uNjingalwazi oNcedisayo kwiSebe le-Anesthesiology yabantwana, i-Perioperative Medicine kunye ne-Pain Medicine. Elinye inyathelo elibalulekileyo, utsho, kukuchonga bonke abachaphazelekayo abanokuchaphazeleka yitekhnoloji kunye nokugqiba indlela bona ngokwabo abangathanda ukuphendula ngayo loo mibuzo.
UJesse Ehrenfeld, MD, umongameli we-American Medical Association, uxoxa ngezinto ezine eziqhuba ukusetyenziswa kwaso nasiphi na isixhobo sezempilo sedijithali, kuquka nezo ziqhutywa bubukrelekrele bokwenziwa. Ngaba iyasebenza? Ngaba oku kuya kusebenza kwiziko lam? Ngubani ohlawulayo? Ngubani onoxanduva?
UMichael Pfeffer, MD, igosa eliyintloko lolwazi kwiStanford Health Care, ukhankanye umzekelo wakutshanje apho uninzi lweengxaki luvavanyiwe phakathi kwabahlengikazi kwizibhedlele zaseStanford. Iingcali zonyango zixhaswa ziimodeli ezinkulu zolwimi ezibonelela ngeenkcazo zokuqala zemiyalezo engenayo yezigulane. Nangona le projekthi ingagqibelelanga, oogqirha abancedise ekuphuhliseni ubuchwepheshe baxela ukuba le modeli inciphisa umthwalo wabo womsebenzi.
“Sisoloko sigxila kwizinto ezintathu ezibalulekileyo: ukhuseleko, ukusebenza kakuhle kunye nokubandakanywa. Singoogqirha. Sifunga ukuba “asizukulimaza,” utshilo uNina Vasan, MD, uprofesa oncedisayo wezonyango lwezengqondo kunye nesayensi yokuziphatha, ojoyine uChar kwaye uPfeffer wajoyina iqela. “Le mayibe yindlela yokuqala yokuvavanya ezi zixhobo.”
UNigim Shah, MBBS, Ph.D., uNjingalwazi wezeMithi kunye neSayensi yeDatha yeBiomedical, uqale ingxoxo ngenani elimangalisayo nangona benikwe isilumkiso esifanelekileyo kubaphulaphuli. “Ndithetha ngokubanzi nangamanani, kwaye ngamanye amaxesha badla ngokuba ngqo kakhulu,” utshilo.
Ngokutsho kukaShah, impumelelo ye-AI ixhomekeke kubuchule bethu bokuyikhulisa. “Ukwenza uphando lwesayensi olufanelekileyo kwimodeli kuthatha malunga neminyaka eli-10, kwaye ukuba nganye yeenkqubo ezili-123 zobudlelwane kunye ne-residency zifuna ukuvavanya nokusasaza imodeli kuloo nqanaba lobunzima, kuya kuba nzima kakhulu ukwenza isayensi echanekileyo njengoko okwangoku sihlela imizamo yethu kwaye [sivavanya]] Kuya kubiza i-$138 yeebhiliyoni ukuqinisekisa ukuba zonke iisayithi zethu zisebenza ngokuchanekileyo,” utshilo uShah. “Asinakukwazi ukuhlawulela oku. Ngoko ke kufuneka sifumane indlela yokwandisa, kwaye kufuneka sandise kwaye senze isayensi elungileyo. Izakhono zobuchule zikwindawo enye kwaye izakhono zobuchule zikwindawo enye, ngoko ke siza kuyidinga loo hlobo lobudlelwane.”
Usekela Dean uYuan Ashley kunye noMildred Cho (owayenguMvuzo) baya kwi-RAISE Health Workshop. USteve Fish
Ezinye izithethi kwintlanganiso zithe oku kungafezekiswa ngokusebenzisana phakathi kukarhulumente nabucala, njengomyalelo wakutshanje we-White House Executive Order on the Secure, Secure and Trustworthy Development and Use of Artificial Intelligence kunye ne-Consortium for Healthcare Artificial Intelligence (CHAI).
“Intsebenziswano phakathi korhulumente nabucala enefuthe elikhulu iphakathi kwezemfundo, icandelo labucala kunye necandelo likarhulumente,” utshilo uLaura Adams, umcebisi omkhulu kwiNational Academy of Medicine. Uqaphele ukuba urhulumente unokuqinisekisa ukuba uluntu luyathembana, kwaye amaziko ezonyango ezemfundo anokubonelela ngokusemthethweni, kwaye ubuchule bobuchwephesha kunye nexesha lekhompyutha zinokubonelelwa licandelo labucala. “Sonke singcono kunaye nabani na kuthi, kwaye siyaqonda ukuba… asinakuthandazela ukuqonda amandla [obukrelekrele bokwenziwa] ngaphandle kokuba siqonde indlela yokusebenzisana omnye nomnye.”
Izithethi ezininzi zithe i-AI ikwanegalelo kuphando, nokuba izazinzulu ziyayisebenzisa ukuphonononga imfundiso yebhayoloji, ukuxela kwangaphambili ulandelelwano olutsha kunye nezakhiwo zeemolekyuli zokwenziwa ukuxhasa unyango olutsha, okanye nokubanceda bashwankathele okanye babhale amaphepha esayensi.
“Eli lithuba lokubona izinto ezingaziwayo,” utshilo uJessica Mega, MD, ingcali yentliziyo kwiSikolo sezonyango kwiYunivesithi yaseStanford kunye nomseki weAlphabet's Verily. UMega ukhankanye imifanekiso ye-hyperspectral, ebamba iimpawu zomfanekiso ezingabonakaliyo emehlweni omntu. Ingcinga kukusebenzisa ubukrelekrele bokwenziwa ukubona iipatheni kwiislayidi ze-pathology abantu abangaziboniyo ezibonisa isifo. “Ndikhuthaza abantu ukuba bamkele izinto ezingaziwayo. Ndicinga ukuba wonke umntu apha uyazi umntu onesifo esithile ofuna into engaphaya koko sinokukubonelela namhlanje,” utshilo uMejia.
Ababeneqela bavumile ukuba iinkqubo zobukrelekrele bokwenziwa ziya kubonelela ngeendlela ezintsha zokuchonga nokulwa nokwenza izigqibo ezicalu ...
“Impilo ayikokhathalelo lwezonyango nje kuphela,” kuvuma iiphaneli ezininzi. Izithethi zigxininise ukuba abaphandi badla ngokungazihoyi izinto ezibangela impilo yoluntu, ezifana nesimo sentlalo-ntle, ikhowudi yeposi, inqanaba lemfundo, kunye nohlanga nobuhlanga, xa beqokelela idatha ebandakanyayo kwaye beqesha abathathi-nxaxheba kwizifundo. “I-AI isebenza kuphela njengedatha eqeqeshwe ngayo le modeli,” utshilo uMichelle Williams, uprofesa we-epidemiology kwiYunivesithi yaseHarvard kunye noprofesa oncedisayo we-epidemiology kunye nempilo yabemi kwiYunivesithi yaseStanford School of Medicine. “Ukuba senza oko sikulwela ukukwenza, siphucula iziphumo zempilo kwaye sisuse ukungalingani, kufuneka siqinisekise ukuba siqokelela idatha esemgangathweni ngokuziphatha kwabantu kunye nokusingqongileyo kwentlalo nendalo.”
UNatalie Pageler, MD, unjingalwazi wezonyango lwabantwana kunye nonyango, uthe idatha edityanisiweyo yomhlaza idla ngokungabandakanyi idatha yabasetyhini abakhulelweyo, nto leyo edala ucalucalulo olungenakuphepheka kwiimodeli kwaye ikhulisa ukungalingani okukhoyo kunyango lwempilo.
UGqr. David Magnus, uprofesa wezonyango lwabantwana kunye nezonyango, uthe njengayo nayiphi na iteknoloji entsha, ubukrelekrele bokwenziwa bunokwenza izinto zibe ngcono ngeendlela ezininzi okanye buzenze zibe mbi ngakumbi. UMagnus uthe umngcipheko kukuba iinkqubo zobukrelekrele bokwenziwa ziya kufunda ngeziphumo zempilo ezingalinganiyo eziqhutywa zizinto ezichaphazela impilo yoluntu kwaye ziqinise ezo ziphumo ngemveliso yazo. “Ubukrelekrele bokwenziwa sisipili esibonisa uluntu esiphila kulo,” utshilo. “Ndiyathemba ukuba ngalo lonke ixesha sinethuba lokukhanyisa umba othile—ukuphakamisa isipili kuthi—kuya kusebenza njengesikhuthazo sokuphucula imeko.”
Ukuba awukwazanga ukuya kwiworkshop ye-RAISE Health, unokufumana ingxelo erekhodiweyo yeseshoni apha.
ISikolo sezonyango seYunivesithi yaseStanford yinkqubo yokhathalelo lwempilo edibeneyo yezemfundo equka iSikolo sezonyango seYunivesithi yaseStanford kunye neenkqubo zokunikezelwa kweenkonzo zempilo zabantu abadala nabantwana. Ziyayiqonda yonke into enokwenzeka kwi-biomedicine ngophando oluhlangeneyo, imfundo kunye nokhathalelo lwezigulane zonyango. Ukuze ufumane ulwazi oluthe kratya, ndwendwela i-med.stanford.edu.
Imodeli entsha yobukrelekrele bokwenziwa inceda oogqirha kunye nabahlengikazi kwiSibhedlele saseStanford ukuba basebenzisane ukuphucula ukhathalelo lwezigulane.
Ixesha leposi: Julayi-19-2024