Komawa Labarai
Bidi'aAI Understanding takaitaccen bayani

Buɗaɗɗen bututun OCR yana rage jinkirin bayanan rajistar cutar kansa

Masu bincike a Jami'ar Minnesota sun haɓaka bututun OCR mai buɗe ido wanda ke fitar da ƙimar gwajin kwayoyin halitta daga rahotannin asibiti tare da daidaiton kashi 97%, mai yuwuwar maye gurbin shigar da bayanan mai jinkirin.

4 min readRead the linked source
Source-provided image accompanying Open-source OCR pipeline reduces cancer registry data latency
Tushen tusheAn rubuta tushen tushe
Mawallafi
bioengineer.org
Tushen hanyar haɗin gwiwa
bioengineer.orghttps://bioengineer.org/ai-reads-the-charts-how-open-source-ocr-could-slash-cancer-data-delays/
Nau'in tushe
Tushen da aka haɗa - ba a kafa matsayin tushen farko ba.
MaganaFahimtar wannan a cikin daƙiƙa 60

Fara a nan

Mabuɗin sharuddan

OCR (Gane Halayen Na gani)
Fasahar da ke canza rubutu a cikin hotuna ko dubawa zuwa rubutu mai iya karanta na'ura.
Bututu
Tsarin aiki da aka ba da umarnin aiwatarwa, matakan ƙira, da matakan aiwatarwa.
Latency
Lokaci tsakanin aika buƙatu da karɓar fitowar samfurin.
Gwada kankaAI Model An Bayyana Tambayoyi

Me ya faru

Wata ƙungiyar bincike da Jami'ar Minnesota ke jagoranta ta nuna cewa bututun bututun buɗe ido na buɗe ido (OCR) na iya fitar da ƙimar sake dawowa daidai daga rahoton cutar kansar nono Oncotype DX. Binciken, wanda aka buga a cikin Ciwon Ciwon daji & Sarrafa, ya yi amfani da haɗin haɗin Tesseract da injunan EasyOCR don aiwatar da rahotanni 675, cimma ƙimar yarjejeniya ta 97% tare da abstraction ɗan adam. Tsarin sarrafa kansa ya fi saurin shigar da littafin hannu na gargajiya, wanda galibi yana fuskantar jinkiri na watanni 12 zuwa 18.

Ƙungiyar binciken, ciki har da Qianyun Luo da Schelomo Marmor, sun ƙera wani bututun 'hybrid OCR' (H-OCR) don magance rashin jin daɗi a cikin rajistar ciwon daji. Sakamakon gwajin genomic, irin su Oncotype DX maƙiyan maimaitawa, ana yawan adana su a cikin bayanan kiwon lafiya na lantarki (EHRs) azaman hotunan da ba a tsara su ba maimakon bayanan da za a iya karantawa na na'ura, suna buƙatar rubutun hannu.

Bututun H-OCR yana amfani da EasyOCR don gano rubutu mai zurfi na tushen ilmantarwa da Tesseract azaman injin faɗuwa don gane halaye. An ƙirƙiri wannan hanyar haɗin gwiwar don yin amfani da ikon EasyOCR don gano bayanan da'irar kan rahotanni yayin amfani da saurin Tesseract da ƙwarewar gano halayen.

A cikin saitin ingantattun rahotanni na 675, bututun H-OCR ya cimma yarjejeniya 97% tare da ka'idojin tunani, wanda ya zarce ƙimar daidaiton 91% na abstraction na yau da kullun na rajista. Tsarin sarrafa kansa ya kammala aikin a cikin kusan sa'o'i 4.9, idan aka kwatanta da tsawan watannin da aka yi na tafiyar da aikin hannu na yanzu.

Binciken ya gano cewa amincewar bututun ya gano yadda za a iya samun kurakurai, yana mai nuna cewa aiwatarwa a nan gaba za su iya amfani da waɗannan maki don tantance abubuwan da ba su da tabbas don bitar ɗan adam, ta yadda za a rage haɗarin ɓarna a asibiti.

Bayanan tushe: bioengineer.org ↗

Me ya sa yake da mahimmanci

Dogaro na yanzu akan rubutun hannu don bayanan genomic yana haifar da ƙulli mai mahimmanci a cikin binciken kansa da daidaitaccen ilimin oncology. Ta hanyar sarrafa sarrafa bayanan da aka tsara daga PDFs marasa tsari, wannan hanyar buɗaɗɗen hanya tana ba da damar tsarin kiwon lafiya su cika rajistar cutar kansa a kusa da ainihin lokaci. Wannan sauye-sauye yana da mahimmanci don inganta ingancin shaida na ainihi, yana ba da damar bincike mai tasiri cikin sauri, da samar da bayanan da aka tsara don horar da samfurin AI na gaba a cikin ilimin oncology. Saboda kayan aikin buɗaɗɗen tushe ne kuma suna iya aiki a cikin mahalli masu dacewa da HIPAA, wannan hanyar tana ba da ƙarancin farashi, mafita mai sauƙi ga cibiyoyi masu ƙuntata albarkatu don haɓaka kayan aikin bayanan su ba tare da software na mallakar mallaka ba.

Genomic biomarkers suna da mahimmanci don daidaitaccen ilimin oncology da auna inganci, duk da haka amfanin su a halin yanzu yana iyakance ta lokacin da ake ɗauka don samar da su a cikin bayanan bincike. Rage wannan jinkirin yana ba da damar ƙarin samar da shaida akan lokaci.

Amfani da kayan aikin buɗaɗɗen kayan aikin kamar Tesseract da EasyOCR yana tabbatar da cewa mafita ta sami dama ga ƙananan ko cibiyoyin ciwon daji waɗanda ke iya rasa kasafin kuɗi don tsada, software na cire bayanan likita na mallakar mallaka.

Binciken ya nuna cewa fa'idodin wannan kamawa ta atomatik ana amfani da su gabaɗaya a cikin ɗimbin majinyata, kamar yadda masu binciken suka gano cewa ƙididdige ƙididdiga na marasa lafiya da abubuwan asibiti ba su yi hasashen kurakuran cirewa ba.

Ta hanyar canza abubuwa na binary da ba a tsara su ba zuwa bayanan da aka tsara, wannan bututun yana samar da ingantattun bayanai, bayanan tsayin daka waɗanda ake buƙata don horarwa da kuma tace AI na gaba da ƙirar koyan inji a cikin sararin oncology.

Interactive Mechanism

Ingantacciyar hanyar sadarwa: Yadda A zahiri yake Aiki

Bincika fasahar da ke bayan wannan ci gaban ta hanyar mu'amala.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Duba ra'ayi na hulɗa+10 Points
AI Models Explained Quiz

Which component of an AI application is the machine-learning model itself?

Abin kallo na gaba

Bincike na gaba zai buƙaci magance girman wannan bututun fiye da daidaitaccen tsarin Oncotype DX. Marubutan sun lura cewa rahotannin jeri na gaba-gaba na somatic, waɗanda ke nuna babban bambanci tsakanin masu siyarwa, suna gabatar da ƙalubale mai rikitarwa don hakar ta atomatik. Bugu da ƙari, yayin da binciken ya inganta bututun a kan tsarin bayanan da aka dawo da shi, ana buƙatar gwaji mai yiwuwa a cikin ayyukan aikin rajista na asibiti don tabbatar da amincin sa a cikin ainihin duniya, saitunan girma mai girma.

Masu binciken sun gano a sarari buƙatar gwada bututun akan ƙarin nau'ikan takardu daban-daban, kamar rahotannin jeri na gaba-gaba, waɗanda suka bambanta sosai ta hanyar mai siyarwa da tsari.

An gudanar da binciken a tsarin kiwon lafiya guda ɗaya; Dole ne aikin nan gaba ya kimanta aikin bututun a cikin cibiyoyi da yawa don tabbatar da cewa ya kasance mai ƙarfi a kan bambance-bambancen ingancin dubawa, ƙudurin fax, da daidaitawar EHR daban-daban.

Haɗin kai mai zuwa cikin ayyukan aikin yin rajista shine mataki na gaba na ma'ana don tantance idan tsarin yana kiyaye daidaito da ingancin sa yayin sarrafa bayanan asibiti masu shigowa cikin ainihin-lokaci.

Jagorori masu alaƙa & tambayoyin tambayoyi

AI Model ya bayyanaAI horoMakomar AIGwada abin da kuka sani - gwada gwajin AI kyautaNemo kalmar AI a cikin ƙamus ɗin muBi samfurin AI na sakin tracker
An sami wannan yana da amfani?