Zikomo pochezera Nature.com.Mtundu wa msakatuli womwe mukugwiritsa ntchito uli ndi chithandizo chochepa cha CSS.Kuti mupeze zotsatira zabwino, tikupangira kugwiritsa ntchito mtundu watsopano wa msakatuli wanu (kapena kuzimitsa mawonekedwe ofananirako mu Internet Explorer).Pakadali pano, kuti tiwonetsetse kuti tikuthandizidwa mosalekeza, tikuwonetsa tsambalo popanda masitayelo kapena JavaScript.
Mano amatengedwa ngati chizindikiro cholondola kwambiri cha zaka za thupi la munthu ndipo nthawi zambiri amagwiritsidwa ntchito poyesa zaka zakubadwa.Tinali ndi cholinga chotsimikizira kuyerekezera kwa zaka za migodi poyerekezera kulondola kwa kuyerekezera ndi kachitidwe ka magulu azaka za 18 ndi njira zachikhalidwe komanso kuyerekezera zaka zazaka zamigodi.Ma radiograph okwana 2657 adasonkhanitsidwa kuchokera kwa nzika zaku Korea ndi Japan zazaka 15 mpaka 23.Adagawidwa m'magulu ophunzitsira, iliyonse ili ndi ma radiograph aku Korea 900, ndi mayeso amkati okhala ndi ma radiograph aku Japan 857.Tidafanizira kulondola kwamagulu komanso kuchita bwino kwa njira zachikhalidwe ndi mayeso amitundu yamigodi ya data.Kulondola kwa njira yachikhalidwe pamayesero amkati ndipamwamba pang'ono kuposa chitsanzo cha migodi ya data, ndipo kusiyana kwake kuli kochepa (kutanthauza zolakwika mtheradi
Kuphunzira pamakina ndi mtundu wa luntha lochita kupanga (AI) lomwe limaphunzira mobwerezabwereza ndikuyika data yochulukirapo, limathetsa mavuto palokha, ndikuyendetsa madongosolo a data.Kuphunzira pamakina kumatha kupeza njira zobisika zobisika mumagulu akulu a data6.Mosiyana ndi zimenezi, njira zachikale, zomwe zimakhala zogwira ntchito komanso zowononga nthawi, zimatha kukhala ndi malire pochita zinthu zambiri zovuta zomwe zimakhala zovuta kuzikonza pamanja7.Chifukwa chake, kafukufuku wambiri wachitika posachedwa pogwiritsa ntchito umisiri waposachedwa wapakompyuta kuti achepetse zolakwika za anthu ndikukonza bwino deta yamitundumitundu8,9,10,11,12.Makamaka, kuphunzira mozama kwagwiritsidwa ntchito kwambiri posanthula zithunzi zachipatala, ndipo njira zosiyanasiyana zoyezera zaka mwa kusanthula ma radiographs zanenedwa kuti zimathandizira kulondola komanso kuchita bwino kwa kuyerekezera zaka13,14,15,16,17,18,19,20 .Mwachitsanzo, Halabi et al 13 adapanga njira yophunzirira makina pogwiritsa ntchito ma convolutional neural network (CNN) kuti athe kuyerekeza zaka za chigoba pogwiritsa ntchito ma radiograph a manja a ana.Kafukufukuyu akupereka chitsanzo chomwe chimagwiritsa ntchito kuphunzira pamakina pazithunzi zachipatala ndikuwonetsa kuti njirazi zitha kukonza kulondola kwa matenda.Li et al14 akuyerekeza zaka kuchokera pazithunzi za X-ray za m'chiuno pogwiritsa ntchito kuphunzira mozama kwa CNN ndikuziyerekeza ndi zotsatira zopumira pogwiritsa ntchito kuyerekezera kwa siteji ya ossification.Adapeza kuti njira yophunzirira mozama ya CNN idawonetsa magwiridwe antchito azaka zofananira ndi momwe zimakhalira.Kafukufuku wa Guo et al. [15] adawunika momwe ukadaulo wa CNN umathandizira kutengera ma orthophotos a mano, ndipo zotsatira za mtundu wa CNN zidatsimikizira kuti anthu adachita bwino kuposa momwe amagawira zaka.
Maphunziro ambiri oyerekeza zaka pogwiritsa ntchito kuphunzira pamakina amagwiritsa ntchito njira zophunzirira mozama13,14,15,16,17,18,19,20.Kuyerekeza zaka kutengera kuphunzira mozama kumanenedwa kukhala kolondola kuposa njira zachikhalidwe.Komabe, njira imeneyi imapereka mwayi wochepa wosonyeza maziko a sayansi a zaka zakubadwa, monga zizindikiro za zaka zomwe zimagwiritsidwa ntchito poyerekezera.Palinso mkangano wamalamulo pa omwe amayendera.Choncho, kuyerekezera zaka zozikidwa pa kuphunzira mozama kumakhala kovuta kuvomereza ndi akuluakulu oyang'anira ndi oweruza.Data mining (DM) ndi njira yomwe ingatulukire osati zoyembekezeredwa komanso zosayembekezereka monga njira yopezera mgwirizano wothandiza pakati pa kuchuluka kwa deta6,21,22.Kuphunzira pamakina nthawi zambiri kumagwiritsidwa ntchito popanga data, ndipo migodi ya data ndi kuphunzira pamakina zimagwiritsa ntchito njira zazikuluzikulu zofananira kuti zipezeke mu data.Kuyerekeza zaka pogwiritsa ntchito kakulidwe ka mano kumatengera kuwunika kwa oyesa kukhwima kwa mano omwe akuwunikira, ndipo kuwunikaku kumawonetsedwa ngati gawo la dzino lililonse lomwe mukufuna.DM ingagwiritsidwe ntchito kusanthula mgwirizano pakati pa gawo lowunika mano ndi zaka zenizeni ndipo ili ndi kuthekera kolowa m'malo mwa kusanthula kwachikhalidwe.Chifukwa chake, ngati tigwiritsa ntchito njira za DM pakuyerekeza zaka, titha kugwiritsa ntchito kuphunzira pamakina poyerekezera zaka zazamalamulo popanda kuda nkhawa ndi mlandu.Kafukufuku wofananiza angapo adasindikizidwa paza njira zina zomwe zingatheke m'njira zachikhalidwe zomwe zimagwiritsidwa ntchito pofufuza zazamalamulo ndi njira zozikidwa pa EBM zodziwira zaka zamano.Shen et al23 adawonetsa kuti mtundu wa DM ndi wolondola kuposa njira yachikhalidwe ya Camerer.Galibourg et al24 anagwiritsa ntchito njira zosiyanasiyana za DM kuti adziŵe zaka malinga ndi muyezo wa Demirdjian25 ndipo zotsatira zake zinasonyeza kuti njira ya DM inapambana kwambiri ndi njira za Demirdjian ndi Willems poyerekezera zaka za anthu a ku France.
Kuyerekeza zaka zamano za achinyamata aku Korea komanso achichepere, njira ya Lee 4 imagwiritsidwa ntchito kwambiri pakufufuza zaku Korea.Njirayi imagwiritsa ntchito kusanthula kwakanthawi (monga kubwereza kangapo) kuti iwunike ubale womwe ulipo pakati pa anthu aku Korea ndi zaka zakubadwa.Mu kafukufukuyu, njira zoyezera zaka zomwe zimapezedwa pogwiritsa ntchito ziwerengero zachikhalidwe zimatchedwa "njira zachikhalidwe."Njira ya Lee ndi njira yachikhalidwe, ndipo kulondola kwake kwatsimikiziridwa ndi Oh et al.5;komabe, kugwiritsa ntchito kwa kuyerekezera zaka kutengera mtundu wa DM muzochita zamalamulo zaku Korea kukadali kokayikitsa.Cholinga chathu chinali kutsimikizira mwasayansi kuthekera kothandiza pakuyerekeza zaka kutengera mtundu wa DM.Cholinga cha kafukufukuyu chinali (1) kuyerekeza kulondola kwa mitundu iwiri ya DM poyerekezera zaka za mano ndi (2) kuyerekeza momwe machitidwe a 7 DM ali ndi zaka 18 ndi omwe anapezedwa pogwiritsa ntchito njira zowerengera zachikhalidwe Kukula kwachiwiri. ndi molars lachitatu m'nsagwada zonse ziwiri.
Njira ndi zopatuka mulingo wanthawi yanthawi ndi siteji ndi mtundu wa dzino zikuwonetsedwa pa intaneti mu Supplementary Table S1 (seti yophunzitsira), Supplementary Table S2 (seti ya mayeso amkati), ndi Supplementary Table S3 (seti ya mayeso akunja).Miyezo ya kappa yodalirika ya intra- ndi interobserver yomwe idapezedwa kuchokera pamaphunzirowa inali 0.951 ndi 0.947, motsatana.Makhalidwe a P ndi 95% nthawi yodalirika pamakhalidwe a kappa akuwonetsedwa patebulo lowonjezera pa intaneti S4.Mtengo wa kappa umatanthauziridwa kuti "pafupifupi bwino", mogwirizana ndi zomwe Landis ndi Koch26.
Poyerekeza mean mean absolute error (MAE), njira yachikale imapambana pang'ono ndi DM model kwa amuna ndi akazi komanso mu test test ya amuna akunja, kupatulapo multilayer perceptron (MLP).Kusiyana pakati pa chitsanzo chachikhalidwe ndi chitsanzo cha DM pamayeso a mkati mwa MAE anali zaka 0.12-0.19 kwa amuna ndi zaka 0.17-0.21 kwa akazi.Kwa batire yoyesera yakunja, zosiyana ndizochepa (zaka 0.001-0.05 kwa amuna ndi zaka 0.05-0.09 kwa akazi).Kuonjezera apo, root mean square error (RMSE) ndi yotsika pang'ono kusiyana ndi njira yachikale, ndi kusiyana kochepa (0.17-0.24, 0.2-0.24 kwa mayeso amkati mwa amuna, ndi 0.03-0.07, 0.04-0.08 kwa mayeso akunja).).MLP ikuwonetsa kuchita bwinoko pang'ono kuposa Single Layer Perceptron (SLP), kupatula ngati mayeso akunja a akazi.Kwa MAE ndi RMSE, mayeso akunja amakhala apamwamba kuposa mayeso amkati omwe amaperekedwa kwa amuna ndi akazi ndi mitundu yonse.Zonse za MAE ndi RMSE zikuwonetsedwa mu Table 1 ndi Chithunzi 1.
MAE ndi RMSE ya miyambo yachikhalidwe ndi migodi ya data.Kutanthauza cholakwika mtheradi MAE, muzu zikutanthauza square cholakwa RMSE, single wosanjikiza perceptron SLP, multilayer perceptron MLP, chikhalidwe CM njira.
Kugwira ntchito kwamagulu (ndi kudulidwa kwa zaka 18) kwamitundu yachikhalidwe ndi ya DM kunawonetsedwa potengera kukhudzika, kutsimikizika, mtengo wolosera wabwino (PPV), mtengo wolosera molakwika (NPV), ndi malo omwe ali pansi pa olandila opareshoni curve (AUROC) 27 (Table 2, Chithunzi 2 ndi Supplementary Figure 1 pa intaneti).Pankhani ya kukhudzika kwa batire yoyesera mkati, njira zachikhalidwe zidachita bwino kwambiri pakati pa amuna komanso zoyipa pakati pa azimayi.Komabe, kusiyana kwa magulu a machitidwe pakati pa njira zachikhalidwe ndi SD ndi 9.7% kwa amuna (MLP) ndi 2.4% yokha ya amayi (XGBoost).Pakati pa zitsanzo za DM, logistic regression (LR) inawonetsa kukhudzika kwabwinoko mwa amuna ndi akazi.Ponena za kutsimikizika kwa mayeso amkati, zidawoneka kuti mitundu inayi ya SD idachita bwino mwa amuna, pomwe yachikhalidwe idachita bwino mwa akazi.Kusiyana kwamagulu amagulu a amuna ndi akazi ndi 13.3% (MLP) ndi 13.1% (MLP), motero, kusonyeza kuti kusiyana kwa machitidwe a magulu pakati pa zitsanzo kumaposa kukhudzidwa.Pakati pa zitsanzo za DM, makina othandizira vekitala (SVM), mtengo wosankha (DT), ndi zitsanzo za m'nkhalango (RF) zinachita bwino kwambiri pakati pa amuna, pamene LR chitsanzo chinachita bwino kwambiri pakati pa akazi.AUROC yachitsanzo chachikhalidwe ndi mitundu yonse ya SD inali yokulirapo kuposa 0.925 (oyandikana nawo wapafupi kwambiri (KNN) mwa amuna), kuwonetsa magwiridwe antchito pakusankha zitsanzo zazaka 1828.Pamayeso akunja, panali kuchepa kwa magwiridwe antchito potengera kukhudzidwa, kutsimikizika ndi AUROC poyerekeza ndi mayeso amkati.Komanso, kusiyana kwa kukhudzika ndi tsatanetsatane pakati pa machitidwe a magulu a zitsanzo zabwino kwambiri komanso zoipitsitsa zinali zochokera ku 10% mpaka 25% ndipo zinali zazikulu kusiyana ndi kusiyana kwa mayesero amkati.
Kukhudzika ndi kutsimikizika kwa mitundu ya migodi ya data poyerekeza ndi njira zachikhalidwe zomwe zidadutsa zaka 18.KNN k oyandikana nawo wapafupi, makina opangira SVM othandizira, LR logistic regression, mtengo wachigamulo wa DT, nkhalango yachisawawa ya RF, XGB XGBoost, MLP multilayer perceptron, njira yachikhalidwe ya CM.
Gawo loyamba mu phunziroli linali kuyerekezera kulondola kwa zaka za zaka za mano zopezedwa kuchokera ku zitsanzo zisanu ndi ziwiri za DM ndi zomwe zinapezedwa pogwiritsa ntchito njira zachikhalidwe.MAE ndi RMSE adayesedwa m'mayesero amkati mwa amuna ndi akazi, ndipo kusiyana pakati pa njira yachikhalidwe ndi chitsanzo cha DM kuyambira 44 mpaka 77 masiku a MAE ndi 62 mpaka 88 masiku a RMSE.Ngakhale kuti njira yachikhalidwe inali yolondola pang'ono mu phunziroli, n'zovuta kunena ngati kusiyana kwakung'ono koteroko kuli ndi tanthauzo lachipatala kapena lothandiza.Zotsatirazi zikuwonetsa kuti kulondola kwa zaka zakubadwa kwa mano pogwiritsa ntchito mtundu wa DM kuli pafupifupi kofanana ndi njira yakale.Kuyerekeza kwachindunji ndi zotsatira za maphunziro am'mbuyomu kumakhala kovuta chifukwa palibe kafukufuku yemwe wayerekeza kulondola kwa zitsanzo za DM ndi njira zachikhalidwe zowerengera pogwiritsa ntchito njira yomweyi yojambulira mano azaka zofanana ndi kafukufukuyu.Galibourg et al24 anayerekezera MAE ndi RMSE pakati pa njira ziwiri zachikhalidwe (Demirjian method25 ndi Willems method29) ndi zitsanzo za 10 DM mwa anthu a ku France a zaka 2 mpaka 24.Iwo adanena kuti mitundu yonse ya DM inali yolondola kuposa njira zachikhalidwe, ndi kusiyana kwa zaka 0.20 ndi 0.38 mu MAE ndi 0.25 ndi zaka 0.47 mu RMSE poyerekeza ndi njira za Willems ndi Demirdjian, motsatira.Kusiyana pakati pa mtundu wa SD ndi njira zachikhalidwe zomwe zawonetsedwa mu kafukufuku wa Halibourg zimaganizira malipoti ambiri30,31,32,33 oti njira ya Demirdjian siyiyerekeza molondola zaka zamano mwa anthu ena kupatula aku Canada aku France komwe kafukufukuyu adachokera.mu phunziro ili.Tai et al 34 adagwiritsa ntchito algorithm ya MLP kulosera zaka za mano kuchokera ku zithunzi za 1636 zaku China za orthodontic ndikuyerekeza kulondola kwake ndi zotsatira za njira ya Demirjian ndi Willems.Adanenanso kuti MLP ili ndi zolondola kwambiri kuposa njira zachikhalidwe.Kusiyana kwa njira ya Demirdjian ndi njira yachikhalidwe ndi <0.32 zaka, ndipo njira ya Willems ndi zaka 0.28, zomwe ziri zofanana ndi zotsatira za kafukufuku wamakono.Zotsatira za maphunziro apitawa24,34 zimagwirizananso ndi zotsatira za kafukufuku wamakono, ndipo kuyerekezera kwa zaka za chitsanzo cha DM ndi njira yachikhalidwe ndizofanana.Komabe, kutengera zotsatira zomwe zaperekedwa, titha kunena mosamala kuti kugwiritsa ntchito mitundu ya DM kuyerekeza zaka kungalowe m'malo mwa njira zomwe zilipo kale chifukwa chosowa kuyerekezera ndi kuyerekezera maphunziro am'mbuyomu.Maphunziro otsatiridwa pogwiritsa ntchito zitsanzo zazikulu akufunika kuti atsimikizire zotsatira zomwe zapezedwa mu kafukufukuyu.
Pakati pa maphunziro omwe amayesa kulondola kwa SD pakuyerekeza zaka zamano, ena adawonetsa kulondola kwambiri kuposa kafukufuku wathu.Stepanovsky et al 35 adagwiritsa ntchito mitundu 22 ya SD pama radiographs apanoramic a 976 okhala ku Czech azaka 2.7 mpaka 20.5 ndikuyesa kulondola kwa mtundu uliwonse.Anayesa kukula kwa mano okhazikika a 16 kumtunda ndi kumunsi kumanzere kumanzere pogwiritsa ntchito ndondomeko yamagulu yomwe inaperekedwa ndi Moorrees et al 36.MAE imachokera ku 0.64 mpaka zaka 0.94 ndipo RMSE imachokera ku 0.85 mpaka zaka 1.27, zomwe ziri zolondola kuposa zitsanzo ziwiri za DM zomwe zimagwiritsidwa ntchito mu phunziroli.Shen et al23 adagwiritsa ntchito njira ya Cameriere kuyerekeza zaka zamano za mano asanu ndi awiri okhazikika kumanzere kwa mandible okhala kum'mawa kwa China azaka za 5 mpaka 13 ndikufanizira ndi zaka zomwe zimaganiziridwa pogwiritsa ntchito kusinthika kwa mzere, SVM ndi RF.Adawonetsa kuti mitundu yonse itatu ya DM ili ndi zolondola kwambiri poyerekeza ndi njira yachikhalidwe ya Cameriere.MAE ndi RMSE mu kafukufuku wa Shen anali otsika kuposa omwe ali mu DM model mu kafukufukuyu.Kuwonjezeka kwatsatanetsatane kwa maphunziro a Stepanovsky et al.35 ndi Shen et al.23 ikhoza kukhala chifukwa cha kuphatikizidwa kwa maphunziro achichepere mu zitsanzo zawo zophunzirira.Chifukwa kuyerekezera zaka kwa omwe ali ndi mano omwe akukulirakulira kumakhala kolondola kwambiri pamene chiwerengero cha mano chikuwonjezeka panthawi ya kukula kwa mano, kulondola kwa njira yoyezera zaka kungasokonezedwe pamene ochita nawo kafukufuku ali aang'ono.Kuphatikiza apo, zolakwika za MLP pakuyerekeza zaka ndizocheperako kuposa za SLP, kutanthauza kuti MLP ndiyolondola kuposa SLP.MLP imawonedwa ngati yabwinoko pang'ono pakuyerekeza zaka, mwina chifukwa cha magawo obisika mu MLP38.Komabe, pali chosiyana ndi zitsanzo zakunja za amayi (SLP 1.45, MLP 1.49).Kupeza kuti MLP ndiyolondola kuposa SLP pakuwunika zaka kumafuna maphunziro owonjezera obwerera.
Magwiridwe amtundu wa mtundu wa DM ndi njira yachikhalidwe pazaka 18 adafananizidwanso.Mitundu yonse yoyesedwa ya SD ndi njira zachikhalidwe pamayesero amkati zidawonetsa tsankho lovomerezeka kwa zitsanzo zazaka 18.Kumverera kwa amuna ndi akazi kunali kwakukulu kuposa 87.7% ndi 94.9%, motsatana, ndipo kutsimikizika kunali kwakukulu kuposa 89.3% ndi 84.7%.AUROC yamitundu yonse yoyesedwa imaposa 0.925.Monga momwe tikudziwira, palibe kafukufuku yemwe adayesa momwe machitidwe a DM amachitira m'magulu azaka 18 potengera kukhwima kwa mano.Titha kufanizitsa zotsatira za kafukufukuyu ndi kachitidwe ka m'magulu amitundu yozama yophunzirira pazithunzi za radiograph.Guo et al.15 anawerengetsera momwe machitidwe a m'gulu la CNN amaphunzirira mwakuya komanso njira yamanja yotengera njira ya Demirjian pamlingo wina wazaka.Chidziwitso ndi tsatanetsatane wa njira yamanja inali 87.7% ndi 95.5%, motero, komanso kukhudzidwa ndi tsatanetsatane wa chitsanzo cha CNN chinaposa 89.2% ndi 86.6%, motero.Iwo adatsimikiza kuti zitsanzo zozama zophunzirira zimatha kulowa m'malo kapena kupitilira kuwunika kwapamanja pakuyika malire azaka.Zotsatira za phunziroli zidawonetsa magwiridwe antchito ofanana;Amakhulupirira kuti kugawa pogwiritsa ntchito mitundu ya DM kungalowe m'malo mwa njira zachikhalidwe zowerengera zaka.Pakati pa zitsanzo, DM LR inali chitsanzo chabwino kwambiri chokhudzana ndi kukhudzidwa kwa chitsanzo cha amuna ndi kukhudzidwa ndi kutsimikizika kwa chitsanzo cha akazi.LR ili pamalo achiwiri mwapadera kwa amuna.Komanso, LR imatengedwa kuti ndi imodzi mwazosavuta kugwiritsa ntchito DM35 ndipo sizovuta komanso zovuta kukonza.Kutengera zotsatira izi, LR idawonedwa ngati njira yabwino kwambiri yopangira ana azaka 18 ku Korea.
Ponseponse, kulondola kwa zaka zakubadwa kapena machitidwe pamayeso akunja kunali kocheperako kapena kuchepera poyerekeza ndi zotsatira za seti ya mayeso amkati.Malipoti ena akuwonetsa kuti kulondola kwamagulu kapena kuchita bwino kumachepera pamene kuyerekeza zaka kutengera kuchuluka kwa anthu aku Korea kukugwiritsidwa ntchito kwa anthu aku Japan5,39, ndipo njira yofananira idapezeka mu kafukufukuyu.Kuwonongeka uku kudawonedwanso mu mtundu wa DM.Choncho, kuti muyese molondola zaka, ngakhale pogwiritsira ntchito DM pofufuza, njira zochokera ku deta ya anthu amtundu, monga njira zachikhalidwe, ziyenera kukondedwa5,39,40,41,42.Popeza sizikudziwika ngati zitsanzo zozama zamaphunziro zimatha kuwonetsa zofanana, maphunziro oyerekeza kulondola kwamagulu ndi luso pogwiritsa ntchito njira zachikhalidwe, zitsanzo za DM, ndi zitsanzo zakuya zophunzirira pazitsanzo zomwezo ndizofunikira kuti zitsimikizire ngati luntha lochita kupanga lingagonjetse kusiyana kwamitundu kumeneku pazaka zochepa.kuwunika.
Tikuwonetsa kuti njira zachikhalidwe zitha kusinthidwa ndikuyerekeza zaka kutengera mtundu wa DM pakuyerekeza zaka zaku Korea.Tidazindikiranso kuthekera kogwiritsa ntchito kuphunzira pamakina pakuwunika zaka zakubadwa.Komabe, pali malire omveka bwino, monga chiwerengero chosakwanira cha omwe atenga nawo mbali mu phunziroli kuti adziwe bwino zotsatira zake, komanso kusowa kwa maphunziro apitalo kuti afanizire ndi kutsimikizira zotsatira za phunziroli.M'tsogolomu, maphunziro a DM akuyenera kuchitidwa ndi zitsanzo zokulirapo komanso anthu osiyanasiyana kuti ziwongolere magwiridwe antchito ake poyerekeza ndi njira zachikhalidwe.Kuti atsimikizire kuthekera kogwiritsa ntchito luntha lochita kupanga kuyerekeza zaka m'magulu angapo, maphunziro amtsogolo akufunika kuti afananize kulondola kwamagulu ndi magwiridwe antchito a DM ndi mitundu yophunzirira mozama ndi njira zachikhalidwe mu zitsanzo zomwezo.
Kafukufukuyu adagwiritsa ntchito zithunzi za 2,657 zojambulidwa kuchokera kwa akulu aku Korea ndi Japan azaka zapakati pa 15 mpaka 23.Ma radiograph aku Korea adagawidwa m'maseti ophunzitsira a 900 (zaka 19.42 ± 2.65) ndi ma seti 900 oyesa amkati (zaka 19.52 ± 2.59).Maphunzirowa adasonkhanitsidwa ku bungwe limodzi (Chipatala cha Seoul St. Mary's), ndipo mayeso omwe adayesedwa adasonkhanitsidwa ku mabungwe awiri (Seoul National University Dental Hospital ndi Yonsei University Dental Hospital).Tinasonkhanitsanso ma radiograph 857 kuchokera ku data ina yokhudzana ndi anthu (Iwate Medical University, Japan) kuti ayesedwe kunja.Ma Radiographs a maphunziro aku Japan (zaka 19.31 ± 2.60) adasankhidwa ngati mayeso akunja.Deta inasonkhanitsidwa mobwerezabwereza kuti afufuze magawo a chitukuko cha mano pazithunzithunzi za radiograph zomwe zimatengedwa panthawi ya chithandizo cha mano.Deta yonse yomwe inasonkhanitsidwa inali yosadziwika kupatula jenda, tsiku lobadwa ndi tsiku la radiograph.Njira zophatikizira ndi zopatula zinali zofanana ndi maphunziro omwe adasindikizidwa kale 4, 5.Zaka zenizeni za chitsanzocho zinawerengedwa pochotsa tsiku lobadwa kuyambira tsiku lomwe radiograph inatengedwa.Gulu lachitsanzo linagawidwa m'magulu asanu ndi anayi.Zaka ndi kugawidwa kwa kugonana kukuwonetsedwa mu Table 3 Phunziroli linachitidwa molingana ndi Declaration of Helsinki ndipo linavomerezedwa ndi Institutional Review Board (IRB) ya Seoul St. Mary's Hospital ya Catholic University of Korea (KC22WISI0328).Chifukwa cha kalembedwe ka kafukufukuyu, chilolezo chodziwitsidwa sichikanatha kupezeka kwa odwala onse omwe akuyesedwa ndi radiographic pofuna kuchiza.Chipatala cha Seoul Korea University St. Mary's Hospital (IRB) chinachotsa kufunikira kwa chilolezo chodziwitsidwa.
Magawo a chitukuko cha bimaxillary yachiwiri ndi yachitatu molars adawunikidwa molingana ndi Demircan criteria25.Dzino limodzi lokha linasankhidwa ngati dzino lamtundu umodzi likupezeka kumanzere ndi kumanja kwa nsagwada iliyonse.Ngati mano homologous mbali zonse anali pa misinkhu osiyana chitukuko, dzino ndi m`munsi siteji chitukuko anasankhidwa chifukwa cha kusatsimikizika mu zaka anayerekezera.Ma radiograph 100 osankhidwa mwachisawawa kuchokera pagulu la maphunziro adagoleredwa ndi owonera awiri odziwa zambiri kuti ayese kudalirika kwa interrobserver pambuyo pa precalibration kuti adziwe kukula kwa mano.Kudalirika kwa intraobserver kunayesedwa kawiri pakapita miyezi itatu ndi woyang'anira wamkulu.
Gawo la kugonana ndi lachitukuko lachiwiri ndi lachitatu la nsagwada za nsagwada iliyonse mu maphunziro a maphunziro adayesedwa ndi woyang'anira wamkulu wophunzitsidwa ndi mitundu yosiyanasiyana ya DM, ndipo msinkhu weniweniwo unakhazikitsidwa ngati mtengo wamtengo wapatali.Mitundu ya SLP ndi MLP, yomwe imagwiritsidwa ntchito kwambiri pophunzira makina, idayesedwa motsutsana ndi ma algorithms obwerera.Mtundu wa DM umaphatikiza ntchito zofananira pogwiritsa ntchito magawo a chitukuko cha mano anayi ndikuphatikiza izi kuti muyerekeze zaka.SLP ndiye neural network yosavuta kwambiri ndipo ilibe zigawo zobisika.SLP imagwira ntchito potengera kufalikira pakati pa node.Mtundu wa SLP mu regression ndi masamu ofanana ndi ma regression angapo mizere.Mosiyana ndi mtundu wa SLP, mtundu wa MLP uli ndi magawo angapo obisika omwe ali ndi ntchito zoyambitsa zosagwirizana.Zoyesera zathu zidagwiritsa ntchito wosanjikiza wobisika wokhala ndi ma node 20 okha obisika omwe ali ndi ntchito zoyambitsa zosagwirizana.Gwiritsani ntchito kutsika kwa gradient ngati njira yokhathamiritsa ndi MAE ndi RMSE ngati ntchito yotayika pophunzitsa makina athu ophunzirira makina.Njira yabwino kwambiri yopumula idagwiritsidwa ntchito pazoyesa zamkati ndi zakunja ndipo zaka za mano zidayesedwa.
Algorithm yamagulu idapangidwa yomwe imagwiritsa ntchito kukhwima kwa mano anayi pamaphunziro omwe akhazikitsidwa kuti adziwike ngati chitsanzo chili ndi zaka 18 kapena ayi.Kuti tipange chitsanzocho, tidapeza njira zisanu ndi ziwiri zophunzirira makina oyimira6,43: (1) LR, (2) KNN, (3) SVM, (4) DT, (5) RF, (6) XGBoost, ndi (7) MLP. .LR ndi imodzi mwama algorithms44 omwe amagwiritsidwa ntchito kwambiri.Ndi njira yophunzirira yoyang'aniridwa yomwe imagwiritsa ntchito kubwezeredwa kuneneratu kuthekera kwa data yomwe ili m'gulu linalake kuchokera ku 0 mpaka 1 ndikuyika datayo m'gulu lomwe lingatheke kutengera kuthekera kumeneku;amagwiritsidwa ntchito makamaka pamagulu a binary.KNN ndi imodzi mwa njira zosavuta zophunzirira makina45.Ikapatsidwa deta yatsopano yolowera, imapeza k data pafupi ndi zomwe zilipo ndikuziika m'kalasi ndi maulendo apamwamba kwambiri.Timayika 3 pa chiwerengero cha oyandikana nawo omwe amaganiziridwa (k).SVM ndi algorithm yomwe imakulitsa mtunda pakati pa magulu awiri pogwiritsa ntchito kernel ntchito kukulitsa malo ozungulira kukhala malo osakhala a mzere otchedwa fields46.Pachitsanzo ichi, timagwiritsa ntchito bias = 1, mphamvu = 1, ndi gamma = 1 ngati ma hyperparameter a polynomial kernel.DT yakhala ikugwiritsidwa ntchito m'magawo osiyanasiyana ngati njira yogawanitsa deta yonse mumagulu angapo poyimira malamulo a chisankho mu mtengo47.Chitsanzocho chimakonzedwa ndi chiwerengero chochepa cha zolemba pa node ya 2 ndipo amagwiritsa ntchito ndondomeko ya Gini monga muyeso wa khalidwe.RF ndi njira yophatikizira yomwe imaphatikiza ma DT angapo kuti apititse patsogolo magwiridwe antchito pogwiritsa ntchito njira yophatikizira ya bootstrap yomwe imapanga gulu lofooka lachitsanzo chilichonse pojambula mosasintha zitsanzo za kukula kofanana kangapo kuchokera pa dataset48 yoyambirira.Tidagwiritsa ntchito mitengo 100, kuya kwamitengo 10, kukula kwa node 1, ndi Gini admixture index monga njira zolekanitsa ma node.Kugawika kwa deta yatsopano kumatsimikiziridwa ndi mavoti ambiri.XGBoost ndi algorithm yomwe imaphatikiza njira zolimbikitsira pogwiritsa ntchito njira yomwe imatenga ngati zophunzitsira zolakwika pakati pa zomwe zidanenedweratu za mtundu wakale ndikuwonjezera cholakwikacho pogwiritsa ntchito ma gradients49.Ndi algorithm yomwe imagwiritsidwa ntchito kwambiri chifukwa cha magwiridwe ake abwino komanso magwiridwe antchito, komanso kudalirika kwakukulu ngati ntchito yowongolera mopitilira muyeso.Mtunduwu uli ndi mawilo 400 othandizira.MLP ndi neural network momwe ma perceptrons amodzi kapena angapo amapanga zigawo zingapo ndi chimodzi kapena zingapo zobisika pakati pa zolowetsa ndi zotulutsa38.Pogwiritsa ntchito izi, mutha kuchita gulu losakhala laling'ono pomwe muwonjeza wosanjikiza wolowera ndikupeza phindu, mtengo wonenedweratu umafananizidwa ndi mtengo weniweni ndipo cholakwikacho chimafalitsidwanso.Tinapanga wosanjikiza wobisika wokhala ndi ma neuroni 20 obisika pagawo lililonse.Mtundu uliwonse womwe tidapanga unkagwiritsidwa ntchito pamaseti amkati ndi akunja kuyesa magwiridwe antchito powerengera chidwi, kutsimikizika, PPV, NPV, ndi AUROC.Sensitivity imatanthauzidwa ngati chiŵerengero cha zitsanzo zomwe zikuyerekezeredwa kukhala zaka 18 kapena kupitirira ku chitsanzo chomwe chikuyembekezeka kukhala ndi zaka 18 kapena kupitirira.Kutsimikizika ndi kuchuluka kwa zitsanzo zosakwana zaka 18 ndi zomwe zikuyerekezeredwa kukhala zosakwana zaka 18.
Magawo a mano omwe amawunikidwa pamaphunzirowa adasinthidwa kukhala magawo owerengera kuti awerengere.Multivariate linear ndi logistic regression adapangidwa kuti apange zitsanzo zolosera za kugonana kulikonse ndikupeza njira zochepetsera zomwe zingagwiritsidwe ntchito kuyerekezera zaka.Tidagwiritsa ntchito mafomuwa kuyerekeza zaka za mano pamaseti amkati ndi kunja kwa mayeso.Gulu 4 likuwonetsa njira zobwerera m'mbuyo ndi zamagulu zomwe zimagwiritsidwa ntchito mu phunziroli.
Kudalirika kwa intra- ndi interobserver kunawerengedwa pogwiritsa ntchito ziwerengero za Cohen's kappa.Kuti tiyese kulondola kwa DM ndi mitundu yobwerera m'mbuyo, tidawerengera MAE ndi RMSE pogwiritsa ntchito zaka zoyerekeza ndi zenizeni zamaseti oyesa amkati ndi akunja.Zolakwa izi zimagwiritsidwa ntchito poyesa kulondola kwa zolosera zachitsanzo.Zolakwika zing'onozing'ono, m'pamenenso zimakulitsa kulondola kwa zolosera24.Yerekezerani MAE ndi RMSE ya mayeso amkati ndi akunja omwe amawerengedwa pogwiritsa ntchito DM ndi kutsika kwachikhalidwe.Kugwira ntchito m'magulu a zaka 18 zochepetsera ziwerengero zachikhalidwe kunayesedwa pogwiritsa ntchito tebulo lamwadzidzi la 2 × 2.Kukhudzika kowerengeredwa, kutsimikizika, PPV, NPV, ndi AUROC ya mayesowo adayerekezedwa ndi milingo yoyezedwa yamtundu wa DM.Deta imawonetsedwa ngati ± kupatuka kokhazikika kapena nambala (%) kutengera mawonekedwe a data.Miyezo iwiri ya P <0.05 idawonedwa ngati yofunika kwambiri.Kusanthula konse kwachizoloŵezi kunachitidwa pogwiritsa ntchito SAS version 9.4 (SAS Institute, Cary, NC).Mtundu wa DM regression model unakhazikitsidwa ku Python pogwiritsa ntchito Keras50 2.2.4 backend ndi Tensorflow51 1.8.0 makamaka pa masamu.Mtundu wa DM classification unakhazikitsidwa mu Waikato Knowledge Analysis Environment ndi Konstanz Information Miner (KNIME) 4.6.152 nsanja yowunikira.
Olembawo amavomereza kuti deta yochirikiza mfundo za phunziroli imapezeka m'nkhani ndi zipangizo zowonjezera.Zolemba zomwe zapangidwa ndi/kapena zowunikidwa panthawi ya phunziroli zimapezeka kuchokera kwa wolemba yemwe akugwirizana nazo pa pempho loyenera.
Ritz-Timme, S. et al.Kuwunika kwa zaka: luso laukadaulo kuti likwaniritse zofunikira zenizeni zamachitidwe azamalamulo.mayiko.J. Mankhwala ovomerezeka.113, 129-136 (2000).
Schmeling, A., Reisinger, W., Geserik, G., ndi Olze, A. Mkhalidwe wapano wa kafukufuku wazaka za anthu omwe ali ndi moyo chifukwa cha milandu.Zazamalamulo.mankhwala.Matenda.1, 239-246 (2005).
Pan, J. et al.Njira yosinthidwa yowunika zaka zamano za ana azaka zapakati pa 5 mpaka 16 kummawa kwa China.zachipatala.Kafukufuku wapakamwa.25, 3463–3474 (2021).
Lee, SS etc. Chronology ya chitukuko chachiwiri ndi chachitatu molars mu Koreas ndi ntchito yake azamalamulo zaka kuwunika.mayiko.J. Mankhwala ovomerezeka.124, 659-665 (2010).
O, S., Kumagai, A., Kim, SY ndi Lee, SS Kulondola kwa kulingalira kwa zaka ndi kulingalira kwa zaka za 18 kutengera kukhwima kwa ma molars wachiwiri ndi wachitatu mu Korea ndi Japan.PLoS ONE 17, e0271247 (2022).
Kim, JY, et al.Kusanthula kwa data potengera makina opangira ma preoperative kumatha kulosera zotsatira za chithandizo cha opareshoni mwa odwala omwe ali ndi OSA.sayansi.Lipoti 11, 14911 (2021).
Han, M. et al.Kuyerekeza zaka zolondola pakuphunzira pamakina popanda kulowererapo kwa anthu?mayiko.J. Mankhwala ovomerezeka.136, 821–831 (2022).
Khan, S. ndi Shaheen, M. Kuchokera ku Data Mining kupita ku Data Mining.J. Information.sayansi.https://doi.org/10.1177/01655515211030872 (2021).
Khan, S. ndi Shaheen, M. WisRule: The First Cognitive Algorithm for Association Rule Mining.J. Information.sayansi.https://doi.org/10.1177/01655515221108695 (2022).
Shaheen M. ndi Abdullah U. Karm: Kukumba kwa data kwachikhalidwe kutengera malamulo okhudzana ndi mayanjano.werengera.Mat.pitilizani.68, 3305–3322 (2021).
Muhammad M., Rehman Z., Shaheen M., Khan M. ndi Habib M. Deep based based semantic kufanana kuzindikirika pogwiritsa ntchito deta ya mawu.dziwitsa.matekinoloje.kulamulira.https://doi.org/10.5755/j01.itc.49.4.27118 (2020).
Tabish, M., Tanoli, Z., ndi Shahin, M. A dongosolo lozindikira zochitika mumavidiyo amasewera.multimedia.Kugwiritsa Ntchito Zida https://doi.org/10.1007/s11042-021-10519-6 (2021).
Halabi, SS et al.RSNA Machine Learning Challenge mu Pediatric Bone Age.Radiology 290, 498–503 (2019).
Li, Y. et al.Kuyerekeza zaka zakubadwa kuchokera ku pelvic X-ray pogwiritsa ntchito kuphunzira mozama.EURO.radiation.29, 2322–2329 (2019).
Guo, YC, et al.Magulu olondola azaka pogwiritsa ntchito njira zamanja komanso maukonde ozama a neural kuchokera pazithunzi zowonetsera.mayiko.J. Mankhwala ovomerezeka.135, 1589–1597 (2021).
Alabama Dalora et al.Kuyerekeza zaka za mafupa pogwiritsa ntchito njira zosiyanasiyana zophunzirira makina: kuwunika mwadongosolo mabuku ndi kusanthula meta.PLoS ONE 14, e0220242 (2019).
Du, H., Li, G., Cheng, K., ndi Yang, J. Kuyerekeza zaka kwa anthu aku Africa ku America ndi ku China kutengera kuchuluka kwa ma molars oyambira pogwiritsa ntchito cone-beam computed tomography.mayiko.J. Mankhwala ovomerezeka.136, 811–819 (2022).
Kim S., Lee YH, Noh YK, Park FK ndi Oh KS Kuzindikira magulu azaka za anthu amoyo pogwiritsa ntchito zithunzi zanzeru zopangira ma molars oyamba.sayansi.Lipoti 11, 1073 (2021).
Stern, D., Payer, C., Giuliani, N., ndi Urschler, M. Kuyerekeza kwa zaka zokha komanso magulu azaka zambiri kuchokera ku data ya multivariate MRI.IEEE J. Biomed.Zidziwitso Zaumoyo.23, 1392–1403 (2019).
Cheng, Q., Ge, Z., Du, H. ndi Li, G. Kuyerekeza zaka kutengera gawo la 3D zamkati la chipinda cha ma molars oyambira kuchokera ku cone beam computed tomography pophatikiza kuphunzira mozama ndi ma seti amlingo.mayiko.J. Mankhwala ovomerezeka.135, 365–373 (2021).
Wu, WT, et al.Kuyika kwa data mu data yayikulu yachipatala: nkhokwe wamba, masitepe, ndi mitundu ya njira.Dziko.mankhwala.gwero.8, 44 (2021).
Yang, J. et al.Chiyambi cha Zosungira Zamankhwala ndi Ma Technologies A Mining mu Big Data Era.J. Avid.Mankhwala oyambira.13, 57-69 (2020).
Shen, S. et al.Njira ya Camerer yoyezera zaka za mano pogwiritsa ntchito makina ophunzirira.BMC Oral Health 21, 641 (2021).
Galliburg A. et al.Kuyerekeza njira zosiyanasiyana zophunzirira makina zolosera zaka zamano pogwiritsa ntchito njira ya Demirdjian.mayiko.J. Mankhwala ovomerezeka.135, 665–675 (2021).
Demirdjian, A., Goldstein, H. ndi Tanner, JM Dongosolo latsopano lowunika zaka zamano.kufwenthera.biology.45, 211–227 (1973).
Landis, JR, ndi Koch, GG Measures of monitoring agreement on categorical data.Biometrics 33, 159-174 (1977).
Bhattacharjee S, Prakash D, Kim C, Kim HK and Choi HK.Kusanthula kwamalemba, morphological and statistical imaging of two-dimensional magnetic resonance imaging pogwiritsa ntchito njira zanzeru zopangira kusiyanitsa zotupa zoyambirira muubongo.Zambiri zaumoyo.gwero.https://doi.org/10.4258/hir.2022.28.1.46 (2022).
Nthawi yotumiza: Jan-04-2024