da Masu binciken Leken Artificial (HUKUMAR LAFIYA TA DUNIYA) daga Google da ke aiki tare da Asibitin Jami'ar Arewa maso Yamma sun kirkiro samfurin AI wanda zai iya gano kansar huhu. Dangane da bayanai daga Hukumar Lafiya ta Duniya, kansar huhu (muguwar cuta a cikin huhu) na daya daga cikin cututtukan da ke saurin kashe mutane a duniya, suna kashe sama da mutane miliyan biyu a shekara kuma suna kashe mutane da yawa.kamar kansar nono.
Don taimakawa ƙwararrun likitocin kiwon lafiya, algorithms da kwamfutoci na iya taimakawa haɓaka ingantattun hanyoyi a fannin kiwon lafiya.
Koyaya, don waɗannan kayan aikin su zama masu amfani, dole ne kowa ya iya zama mai sauƙin fahimta da fahimta, likitoci da majiyyata, koda ba tare da ilimin fasaha ko kwamfuta ba.
A zahiri, ya kamata a sani cewa aikin duk na'urorin dijital ya dogara da shirye-shiryen kwamfuta da bayanai.
Kalmar "hankali na wucin gadi" yana nuna cewa waɗannan na'urori suna da ikon yin tunani da kansu. Idan an tsara su daidai, na'urori masu fasaha zasu iya kimanta bayanan da aka bayar kuma canza tsari ko sigogi "a kan tashi." Idan aka basu isassun bayanai, zasu iya 'koyo' da kuma gyara lambar su dangane da waɗannan sabbin sifofin.
A cikin shekaru uku da suka gabata, ƙungiyoyin Google suna amfani da AI don matsalolin kiwon lafiya, daga bincikar cututtukan ido zuwa tsinkayar sakamakon haƙuri a cikin bayanan likita.
A yau muna raba sabon bincike wanda ke nuna yadda AI zata iya hango kansar huhu ta hanyoyin da zasu kara samun damar rayuwa ga mutane da yawa masu hadari a duniya.
Ilimin Artificial don inganta rayuwar
Cikakken bayani game da binciken da aka buga 20 Mayu a Magungunan Yanayi, an yi amfani da samfurin ilmantarwa mai zurfi don hango ko mai haƙuri yana da ciwon huhu na huhu, samarda kasadar cutar sankarar huhu da gano wurin da cutar sankarar huhu take.
"Ta hanyar nuna cewa zurfin ilmantarwa na iya ƙara ƙayyadadden bayani ba tare da sadaukar da hankali ba, muna fatan samar da ƙarin bincike da tattaunawa game da rawar da AI ke iya takawa wajen sauya sikelin-fa'idar faɗakarwar gano cutar kansa." , zamu iya karantawa akan shafin Google.
"Tsarin ilimin kere kere na amfani da 3D mai zurfin ilmantarwa don yin nazarin dukkan jikin jikin na'urar daukar hoton kirji, da kuma faci bisa dabarun gano abubuwan da ke gano yankuna masu mummunan rauni," in ji Shreeva Shetty. , manajan fasaha na Google.
Ta hanyar nazarin hoto daya, samfurin ya gano cutar kansa (a matsakaita 5%) mafi yawan lokuta fiye da rukuni na ƙwararrun masana mutum shida kuma sun fi 11% iya rage ƙage na ƙarya (ba gaskiya ba ne sakamakon yanke shawara a cikin zaɓen fidda gwani, ya bayyana tabbatacce, inda ainihin ba shi da kyau)
Masu aikin rediyo galibi suna ganin ɗaruruwan hotuna na 2D a cikin CT sau ɗaya, kuma cutar kansa na iya zama ƙarama da wahalar ganowa. Mun kirkiro abin kirki wanda ba zai iya samar da hasashen gaba daya kawai na cutar sankarar huhu ba (wanda aka kalle shi a cikin 3D) amma kuma zai iya gano muguwar cuta a cikin huhu (huhunan nodules).
Hakanan samfurin zai iya yin la'akari da bayanai daga sikanin da ya gabata, mai amfani wajen tsinkayar haɗarin cutar sankarar huhu saboda ƙimar girman nodules na huhu na iya zama alamun rashin lafiya.
Wadannan sakamakon farko suna da ban ƙarfafa, amma kara karatu zai kimanta tasiri da amfani a cikin aikin asibiti.
A cikin bincikenmu, mun lakafta kararrakin gano CT guda 45,856 (wasu kuma an gano cutar kansa) daga bayanan binciken NIH da aka kafa daga National National Lung Trial Trial da Northwest University. Mun ƙaddamar da sakamakon tare da saiti na biyu kuma mun gwada sakamakonmu tare da lasisin masanan rediyo na Amurka 6.
Google ya sanar cewa zai samar da samfurin ta hanyar Google Cloud Healthcare API yayin da yake ci gaba da ƙarin gwaji da gwaji tare da ƙungiyoyin abokan tarayya.
Source: https://www.blog.google/