Kwenzekeni
Ithimba labacwaningi eliholwa yiNyuvesi yaseMinnesota libonise ukuthi ipayipi le-hybrid open-source optical character recognition (OCR) lingakhipha ngokunembile izikolo zokuphindaphinda kwe-genomic emibikweni yomdlavuza webele we-Oncotype DX askeniwe. Ucwaningo, olushicilelwe kuCancer Causes & Control, lusebenzise inhlanganisela yezinjini ze-Tesseract kanye ne-EasyOCR ukucubungula imibiko engama-675, kwazuza isilinganiso sesivumelwano esingu-97% ngokukhipha ngesandla komuntu. Isistimu ezenzakalelayo isebenze kakhulu ngaphezu kwejubane lokungena kokubhalisa okwenziwa ngesandla, okuvame ukubhekana nokulibaziseka kwezinyanga eziyi-12 kuya kweziyi-18.
Ithimba labacwaningi, elihlanganisa i-Qianyun Luo kanye ne-Schelomo Marmor, lenze ipayipi le-'hybrid OCR' (H-OCR) ukuze libhekane nokubambezeleka ezindaweni ezibhalisiwe zomdlavuza. Imiphumela yokuhlolwa kwe-Genomic, efana ne-Oncotype DX recurrence score, ivamise ukugcinwa kumarekhodi ezempilo e-electronic (EHRs) njengezithombe ezingahleliwe eziskeniwe kunedatha efundeka ngomshini, edinga ukulotshwa ngesandla.
Ipayipi le-H-OCR lisebenzisa i-EasyOCR ekutholeni umbhalo walo osekelwe ekujuleni kokufunda kanye ne-Tesseract njengenjini yokubuyela emuva yokubonwa kwezinhlamvu. Le ndlela eyingxubevange yakhelwe ukuthuthukisa ikhono le-EasyOCR lokwenza izichasiselo eziyindilinga zibe ezasendaweni emibikweni kuyilapho kusetshenziswa isivinini se-Tesseract namandla okuqaphela umlingiswa.
Kusethi yokuqinisekisa yemibiko engama-675, ipayipi le-H-OCR lizuze isivumelwano esingu-97% namazinga ereferensi enziwa mathupha, okwenza ngaphezu kwezinga lokunemba elingu-91% lokukhishwa kokubhalisa okujwayelekile. Inqubo ezenzakalelayo iqedele umsebenzi cishe emahoreni angu-4.9, uma kuqhathaniswa nokubambezeleka okuthatha izinyanga okujwayelekile kokugeleza komsebenzi okwenziwa ngesandla kwamanje.
Ucwaningo luthole ukuthi amaphuzu okuzethemba epayipi akhombe ngempumelelo amaphutha angaba khona, okuphakamisa ukuthi ukusetshenziswa kwesikhathi esizayo kungase kusebenzise lawa maphuzu ukumaka izingcaphuno ezingaqinisekile ukuze zibuyekezwe umuntu, ngaleyo ndlela kuncishiswe ubungozi bokungahlukaniswa kahle okubalulekile komtholampilo.
Imininingwane yomthombo: bioengineer.org ↗
Kungani kubalulekile
Ukuthembela kwamanje ekulotshweni okwenziwa ngesandla kwedatha ye-genomic kudala ibhodlela elibalulekile ocwaningweni lomdlavuza kanye ne-oncology enembile. Ngokuzenzakalela ukukhishwa kwedatha ehlelekile kuma-PDF angahlelekile, le ndlela yomthombo ovulekile ivumela amasistimu ezempilo ukuthi agcwalise okubhalisiwe komdlavuza cishe ngesikhathi sangempela. Lolu shintsho lubalulekile ekuthuthukiseni ikhwalithi yobufakazi bomhlaba wangempela, lunikeze amandla ucwaningo olusheshayo lokuqhathanisa, kanye nokuhlinzeka ngedatha ehlelekile edingekayo ukuze kuqeqeshwe amamodeli e-AI esikhathi esizayo ku-oncology. Ngenxa yokuthi amathuluzi awumthombo ovulekile futhi angasebenza ngaphakathi kwezindawo ezithobela i-HIPAA, le ndlela inikeza isixazululo esingabizi kakhulu, esikwazi ukukhiqizwa kabusha sezikhungo ezicindezelwe yizinsiza ukwenza ingqalasizinda yazo yedatha ibe yesimanje ngaphandle kwesoftware yobunikazi.
Ama-Genomic biomarker abalulekile ekuhlolweni kwe-oncology ngokunemba kanye nokukalwa kwekhwalithi, kodwa ukusetshenziswa kwawo okwamanje kunqunyelwe isikhathi esithathwayo ukuze atholakale kudathabhesi yocwaningo. Ukunciphisa lokhu kubambezeleka kuvumela ukukhiqizwa kobufakazi obengeziwe ngesikhathi.
Ukusetshenziswa kwamathuluzi omthombo ovulekile njenge-Tesseract kanye ne-EasyOCR kuqinisekisa ukuthi isixazululo siyafinyeleleka ezikhungweni zomdlavuza ezincane noma ezivimbelwe yizinsiza ezingase zintule isabelomali sesofthiwe ebizayo, ephathelene nokukhipha idatha yezokwelapha.
Ucwaningo lubonise ukuthi izinzuzo zalokhu kuthwebula okuzenzakalelayo zisebenza kabanzi kuzo zonke iziguli ezihlukene, njengoba abacwaningi bathola ukuthi izibalo zesiguli nezici zomtholampilo azizange zibikezele ngokuphawulekayo amaphutha okukhipha.
Ngokuguqula izinto kanambambili ezingahlelekile zibe idatha ehlelekile, leli payipi lihlinzeka ngekhwalithi ephezulu, idatha yobude edingekayo ukuze kuqeqeshwe futhi kucwengwe i-AI yesikhathi esizayo kanye namamodeli okufunda emishini esikhaleni se-oncology.
I-Interactive Mechanism: Indlela Esebenza Ngayo Ngempela
Hlola ubuchwepheshe obuyisisekelo ngemuva kwalokhu kuthuthukiswa ngokuhlanganyela.
Which component of an AI application is the machine-learning model itself?
Ongakubuka ngokulandelayo
Ucwaningo lwesikhathi esizayo luzodinga ukubhekana nokwanda kwaleli payipi ngale kwefomethi ye-Oncotype DX esezingeni. Ababhali baphawule ukuthi imibiko yokulandelana kwesizukulwane esilandelayo somatic, ebonisa ukuhlukahluka okukhulu kubathengisi bonke, yethula inselele eyinkimbinkimbi yokukhipha okuzenzakalelayo. Ukwengeza, ngenkathi ucwaningo luqinisekisa ipayipi ngokuqhathaniswa nedathasethi yedatha, ukuhlolwa okulindelwe ngaphakathi kokugeleza komsebenzi wokubhaliswa komtholampilo kuyadingeka ukuze kuqinisekiswe ukwethembeka kwakho emhlabeni wangempela, izilungiselelo zevolumu ephezulu.
Abacwaningi bahlonze ngokusobala isidingo sokuhlola ipayipi ezinhlotsheni zamadokhumenti ahlukahlukene, njengemibiko yokulandelana kwesizukulwane esilandelayo somatic, ehluka kakhulu ngomthengisi nefomethi.
Ucwaningo lwenziwe ohlelweni olulodwa lwezempilo; umsebenzi wesikhathi esizayo kufanele uhlole ukusebenza kwepayipi kuzo zonke izikhungo eziningi ukuze kuqinisekiswe ukuthi lihlala liqinile uma liqhathaniswa nokwehluka kwekhwalithi yokuskena, ukulungiswa kwefeksi, nokuhlelwa kwe-EHR okuhlukile.
Ukuhlanganiswa okulindelwe ekugelezeni komsebenzi wokubhalisa bukhoma kuyisinyathelo esilandelayo esinengqondo sokunquma ukuthi uhlelo lugcina ukunemba nokusebenza kahle kwalo lapho lucubungula idatha yomtholampilo engenayo ngesikhathi sangempela.