Ukufika kobuhlakani bokwenziwa kwezempilo akuseyona into esiyibonayo kumafilimu esayensi; kuyiqiniso eliba ngokoqobo ngokushesha ezibhedlela, emitholampilo, nasemalebhu. Kusukela kuma-algorithms athola amathumba ngokunemba okumangalisayo kuya ezinhlelweni ezibikezela izingozi zenhliziyo nemithambo yegazi, i-AI idla idatha yezokwelapha, yezakhi zofuzo, kanye neyebhayomethrikhi. Amandla okusindisa izimpilo makhulu kakhulu, kodwa-ke, ingozi yokwenza amaphutha inkulu ngokulinganayo uma ingaphathwa ngokuqina okukhulu, njengoba sibhekene nolwazi lwabantu oluyimfihlo kakhulu.
Ukusungula izinto ezintsha kwezokwelapha kubalulekile, kodwa ukwenza kanjalo ngokungaboni akuyona inketho. Ukuvimbela la mathuluzi ukuthi angabi yikhanda elibuhlungu elingokomthetho noma, okubi nakakhulu, abe yingozi ezigulini, kubalulekile ukuthi ahambisane nohlaka lokulawula oluqinile nolunezimiso zokuziphatha . Akukhona nje ukusebenza kwekhodi, kodwa mayelana nokuhlonipha ubumfihlo, ukugwema ukucwasa ngokwe-algorithmic, kanye nokuhambisana nemithetho, nakuba ngezinye izikhathi ibonakala iyi-labyrinth, ekhona ukuvumela ubuchwepheshe buthuthuke ngaphandle kokunyathela amalungelo ayisisekelo ku-inthanethi nakwedijithali.
I-labyrinth yezomthetho: i-GDPR, uMthetho we-AI kanye ne-MDR
EYurophu, akukho mthetho owodwa, kodwa kunenethiwekhi yemithethonqubo ehambisanayo eyenzelwe ukumboza zonke izikhala. Umthetho Wokuvikelwa Kwedatha Ojwayelekile (i-GDPR) uyisisekelo; uqinisekisa ukuthi ulwazi olubucayi lwesiguli alugcini lapho kungafanele lufike khona. Lapha, imiqondo efana nemvume ecacile, ukungaziwa (ukususa noma yimuphi umkhondo wobunikazi), kanye nokuqamba igama eliyimfihlo (ukubethela idatha) kubalulekile kunoma yisiphi isikhungo sezempilo esifuna ukugwema izinhlawulo ezinkulu.
Ngakolunye uhlangothi, sinoMthetho we-AI osanda kwethulwa , ohlukanisa izinhlelo ze-AI ngokwengozi eziba nayo. Kwezokunakekelwa kwempilo, amathuluzi amaningi awela esigabeni esinobungozi obukhulu , njengezinhlelo zokuhlola izingozi eminyangweni yezimo eziphuthumayo noma ukuhumusha izithombe ze-radiological. Kulezi zimo, umthetho uqinile futhi udinga ukugadwa kwabantu njalo, idatha yokuqeqeshwa kwabamele, kanye nokucaca okuphelele kokuthi i-algorithm ifinyelela kanjani eziphethweni zayo.
Ekugcineni, i -Medical Device Regulation (MDR) iyaqala ukusebenza lapho i-AI ifakwa kudivayisi yezokwelapha. Lapha, ukuqinisekiswa kwezokwelapha kubalulekile; kumele kuboniswe ukuthi idivayisi isebenza kahle ezweni langempela nokuthi izibuyekezo zesofthiwe aziphazamisi ukusebenza kwayo. Ukuze umtholampilo ube nokuthula kwengqondo, kumele uhlanganise lawa maqoqo amathathu emithethonqubo, njengoba ukuhambisana nomunye ngenkathi ungawunaki omunye kungadala izikhala zomthetho eziyingozi.
Ingozi engabonakali: Ukucwasa kanye "neBhokisi Elimnyama"
Enye yezinto ezibangela ubuhlungu obukhulu kakhulu ukucwasa ngokwe-algorithm. I-AI ifunda kudatha esiyinikezayo, futhi uma leyo datha icwasa, umphumela uzoba obandlululayo. Isibonelo esicacile yilapho uhlelo lwengozi yenhliziyo luthatha kancane ingozi kwabesifazane ngoba luqeqeshwe kakhulu emlandweni wezokwelapha wabesilisa. Lokhu akuyona nje iphutha lobuchwepheshe; kuyingozi yezokwelapha engaholela ekuxilongweni okulibazisekile.
Okungeziwe kulokhu inkinga "yebhokisi elimnyama," lokho kungabonakali lapho i-AI ithi "lesi siguli sinesifo X" kodwa ingachazi ukuthi kungani. Kudokotela, ukwamukela ukuxilongwa ngaphandle kokucabanga kufana nokudlala i-roulette yaseRussia. Yingakho uMthetho we-AI ucindezela izindlela zokuchaza , okuvumela uchwepheshe ukuthi ahlole isinqumo futhi asichazele isiguli ngamagama acacile, kuyilapho ehlala egcina umthwalo wokugcina wokuxilongwa.
Umbono Womhlaba Wonke: IYurophu, i-US, kanye neShayina
Uma sibheka ngale kwemingcele yethu, sibona ukuthi isifunda ngasinye sinefilosofi yaso. E-United States, i -FDA ibeka phambili ukuqinisekiswa kobuchwepheshe ngaphambi kokuba umkhiqizo ukhishwe emakethe, okunikeza izinga eliphezulu lokuphepha kodwa okungenza ukuthi ukwamukelwa kuhambe kancane. Ngokuphambene nalokho, iShayina ithanda imodeli eguquguqukayo kakhulu futhi enolaka, ekhuthaza ukusungula izinto ezintsha okusheshayo, yize ngezinye izikhathi kulimaza ubumfihlo bedatha kanye nokungakhethi .
IYurophu izama ukulinganisela. Sifuna ukusungula izinto ezintsha, yebo, kodwa hhayi nganoma yiziphi izindleko. Ngokubeka ukuphepha nokuziphatha phambili, izwekazi lifuna ukuzibeka njengomholi ku-AI enomthwalo wemfanelo . Kudokotela waseYurophu, lokhu kusho ukuthi ukuhlanganisa ithuluzi laseMelika noma laseShayina kuzodinga ukubuyekezwa okuphelele ukuqinisekisa ukuthi liyahambisana ne-GDPR kanye noMthetho we-AI ngaphambi kokuba usetshenziswe emsebenzini wabo.
Imiphumela engokoqobo kanye nemithwalo yemfanelo yezokwelapha
Emisebenzini yansuku zonke, i-AI isivele iyithuluzi elibalulekile kwezobuchwepheshe ezifana ne-radiology, lapho ama-algorithms ethola khona amaqhubu emaphashini ngokunemba okungaphezu kuka-94%, noma ku-dermatology, ekhomba ama-melanoma iso lomuntu elingase liwaphuthe. Kodwa-ke, lokhu akusho ukuthi odokotela bangahoxa. Ukuqapha okusebenzayo kuyimpoqo : uchwepheshe kumele aqinisekise isiphakamiso ngasinye esivela emshinini, anikeze umongo wezokwelapha i-AI engasoze yaba nawo.
Ukuze kuphathwe konke lokhu, indima yeSikhulu Sokuvikela Idatha (i-DPO) iba yinto ebalulekile ezikhungweni zezempilo. Lochwepheshe akagcini nje ngokuqapha ukwephulwa kwedatha kodwa kumele futhi enze Ukuhlolwa Komthelela Wokuvikela Idatha (ama-DPIA) lapho kusetshenziswa ubuchwepheshe obusha. Ngaphezu kwalokho, kubalulekile ukuthi izikhungo zamukele izinqubo zokuphepha, njengomhlahlandlela ophelele wokuba yisakhamuzi sedijithali kanye nokuphepha kwe-cyber, ukuvimbela i-ransomware ukuthi ingaphazamisi ukunakekelwa kwezempilo.
Ukubheka esikhathini esizayo: Indawo Yedatha Yezempilo YaseYurophu
Isinyathelo esikhulu esilandelayo yi-European Health Data Area (EHDA), isinyathelo sokudala uhlelo lwedijithali lapho idatha engaziwa igeleza khona ngokuphephile phakathi kwamazwe. Lokhu kuzovumela ama-AI ukuthi aqeqeshwe kumasampula ahlukahlukene kakhulu , kuncishiswe ukucwaswa futhi kusheshiswe ucwaningo lwezokwelapha. Kodwa-ke, inselele izoba ukuhlanganisa lokhu ezibhedlela ngaphandle kokucindezela odokotela ngemithwalo yokuphatha.
Ukuze basinde kulolu shintsho, ochwepheshe bezokunakekelwa kwempilo badinga ukuqeqeshwa okukhethekile. Ukwazi ukusebenzisa isofthiwe akwanele; badinga ukuqonda ukuziphatha kwedatha kanye nokuphepha kwe-cyber . Izifundo ezikhethekile kanye nezinhlelo ze-master ku-AI yezempilo ziyavela ukuze kuvalwe leli gebe lolwazi, kuqinisekiswe ukuthi ubuchwepheshe buwusizo, hhayi umthwalo wezomthetho.
Izihluthulelo zokusetshenziswa okubushelelezi
Uma umtholampilo ufuna ukusebenzisa i-AI ngaphandle kokuphela enkantolo, kufanele uqale ngokuhlola izinhlelo zawo bese uzihlukanisa ngokwengozi. Kubalulekile ukuhlola izinkontileka nabaphakeli ukuqinisekisa ukuthi bangabakhiqizi abanomthwalo wemfanelo ngokoMthetho we-AI. Ngaphezu kwalokho, kunconywa ukusebenzisa amasethi edatha okwenziwa esigabeni sokuthuthukiswa nokuqeqeshwa, ngaleyo ndlela kugwenywe ukudalulwa kwedatha yangempela yesiguli ngokungadingekile.
Ikhwalithi yedatha iyigazi lokuphila kwe-AI; uma ufaka idatha engeyinhle, uzothola imiphumela emibi. Ngakho-ke, kubalulekile ukusebenzisa amazinga omhlaba wonke njenge-SNOMED noma i-LOINC futhi wenze ukuhlolwa kwedatha okuphelele ngaphambi kokuqalisa noma iyiphi imodeli. Kungale ndlela kuphela lapho ungaqinisekisa khona ukuthi ithuluzi lithembekile, liphephile, futhi, ngaphezu kwakho konke, liwusizo ekuthuthukiseni izimpilo zeziguli.
Ukuhlanganiswa kobuhlakani bokwenziwa kwezokwelapha kumele igxathu elibalulekile ekuqondeni kokuxilonga kanye nokuphathwa kwezibhedlela, uma nje izidingo ze-GDPR kanye noMthetho we-AI ziqondiswa ngobuhlakani. Isihluthulelo asisikho ekushintsheni isahlulelo sezokwelapha ngesahlulelo se-algorithmic, kodwa ekusithuthukiseni ngokubheka okuqinile kwabantu, ukuqedwa kobandlululo, kanye nokuvikelwa kobumfihlo okuqinile, ukuguqula ubunzima bokulawula kube isiqinisekiso sokuphepha kohlelo lwezempilo kanye nabasebenzisi balo.
Umbhali oshisekayo ngomhlaba wamabhayithi nobuchwepheshe ngokujwayelekile. Ngiyathanda ukwabelana ngolwazi lwami ngokubhala, futhi yilokho engizokwenza kule bhulogi, ngikubonise zonke izinto ezithakazelisayo kakhulu ngamagajethi, isofthiwe, ihadiwe, izitayela zobuchwepheshe, nokuningi. Inhloso yami ukukusiza ukuthi uzulazule emhlabeni wedijithali ngendlela elula nejabulisayo.