Kanser prognostik de\u011ferlendirmelerinde, sa\u011fkal\u0131m analizi bilimsel ara\u015ft\u0131rmalar\u0131n temel ta\u015flar\u0131ndan biri olmaya devam ediyor. Son d\u00f6nemde BMC Cancer dergisinde yay\u0131mlanan \u00f6nemli bir \u00e7al\u0131\u015fma, g\u00f6rece s\u0131n\u0131rl\u0131 kullan\u0131lan h\u0131zland\u0131r\u0131lm\u0131\u015f ya\u015fam s\u00fcresi (Accelerated Failure Time – AFT) frailty modellerini, ileri d\u00fczey d\u00fczenleme teknikleriyle (regularization) birle\u015ftirerek meme kanseri sa\u011fkal\u0131m tahminlerindeki belirsizlikleri anlamay\u0131 ve model performans\u0131n\u0131 art\u0131rmay\u0131 ama\u00e7lad\u0131. Bu \u00e7al\u0131\u015fma, hem sim\u00fcle edilmi\u015f hem de ger\u00e7ek hasta verilerini kullanarak, farkl\u0131 frailty modellerinin kar\u015f\u0131la\u015ft\u0131r\u0131lmas\u0131n\u0131 detayl\u0131 bi\u00e7imde ele alarak meme kanserinde sa\u011fkal\u0131m\u0131 etkileyen anahtar fakt\u00f6rleri daha do\u011fru bir \u015fekilde belirlemeyi ba\u015fard\u0131.<\/p>\n
Sa\u011fkal\u0131m analizinde uzun zamand\u0131r kullan\u0131lan frailty modelleri, bireyler aras\u0131ndaki g\u00f6zlemlenemeyen heterojenli\u011fi modellemek i\u00e7in geli\u015ftirilmi\u015ftir. \u00d6zellikle klinik ve genetik verilerin boyutunun giderek artt\u0131\u011f\u0131 g\u00fcn\u00fcm\u00fczde, hangi frailty modelinin daha etkin ve yoruma a\u00e7\u0131k oldu\u011funu se\u00e7mek \u00f6nemli bir zorluk te\u015fkil etmektedir. Bahsi ge\u00e7en ara\u015ft\u0131rmada, Weibull, Log-logistic, Gamma, Gompertz, Log-normal, Generalized Gamma ve Extreme Value olmak \u00fczere yedi farkl\u0131 AFT frailty modeli birbirleriyle kar\u015f\u0131la\u015ft\u0131r\u0131ld\u0131. Bu modeller, LASSO, Ridge ve Elastic Net gibi s\u0131kl\u0131kla kullan\u0131lan d\u00fczenleme teknikleri ile desteklenerek, a\u015f\u0131r\u0131 uyum (overfitting) sorunlar\u0131n\u0131n \u00f6n\u00fcne ge\u00e7ildi.<\/p>\n
AFT modellerinin en \u00f6nemli avantajlar\u0131ndan biri, beklenen olay (\u00f6rne\u011fin \u00f6l\u00fcm ya da hastal\u0131\u011f\u0131n n\u00fcksetmesi) zaman\u0131n\u0131 do\u011frudan etkileyen fakt\u00f6rlerin zaman \u00fczerindeki etkisini a\u00e7\u0131k\u00e7a yorumlanabilir \u015fekilde sunmalar\u0131d\u0131r. Frailty terimleri ise modelin i\u00e7ine dahil edilen rastgele etkilerle, \u00f6l\u00e7\u00fclemeyen ancak hastan\u0131n riskini etkileyen temel nedenlerin temsilini sa\u011flar. Ara\u015ft\u0131rmac\u0131lar, modellerin ba\u015far\u0131s\u0131n\u0131 kar\u015f\u0131la\u015ft\u0131rmada; Akaike Bilgi Kriteri (AIC), Bayesyen Bilgi Kriteri (BIC) ve tahmin hatalar\u0131n\u0131 \u00f6l\u00e7en Ortalama Mutlak Hata (MAE) ile Ortalama Kare Hata (MSE) gibi \u00e7ok y\u00f6nl\u00fc kriterleri kulland\u0131.<\/p>\n
Ara\u015ft\u0131rman\u0131n dikkat \u00e7ekici bulgusu, Extreme Value Frailty AFT modelinin, farkl\u0131 \u00f6rneklem oranlar\u0131nda (y\u00fczde 25, 50 ve 75) tutarl\u0131 \u015fekilde di\u011fer modellere k\u0131yasla daha ba\u015far\u0131l\u0131 performans sergilemesi oldu. AIC ve BIC de\u011ferlerinde elde edilen en d\u00fc\u015f\u00fck skorlara bu modelle ula\u015f\u0131l\u0131rken, MAE ve MSE gibi tahmin do\u011frulu\u011funu \u00f6l\u00e7en kriterlerde de belirgin \u00fcst\u00fcnl\u00fck g\u00f6zlendi. Bu, Extreme Value modelinin meme kanseri sa\u011fkal\u0131m\u0131n\u0131 tahmin etmede hem istikrarl\u0131 hem de klinik a\u00e7\u0131dan anlaml\u0131 sonu\u00e7lar verdi\u011fini ortaya koydu.<\/p>\n
Model yorumlanabilirli\u011fi, klinik sahada kullan\u0131m\u0131 do\u011frudan etkileyen \u00f6nemli bir unsur olarak \u00f6n plana \u00e7\u0131kt\u0131. D\u00fczenleme teknikleri aras\u0131nda \u00f6zellikle LASSO (En K\u00fc\u00e7\u00fck Mutlak B\u00fcy\u00fckl\u00fck S\u0131n\u0131rland\u0131rmas\u0131) metodu, \u00f6nemsiz de\u011fi\u015fkenleri s\u0131f\u0131ra indirerek modelin yal\u0131nla\u015fmas\u0131n\u0131 sa\u011flad\u0131. Bu sayede ya\u015f, progesteron resept\u00f6r durumu (PR) ve hastaneye yat\u0131\u015f gibi klinik anlamda ay\u0131rt edici olmayan de\u011fi\u015fkenler modelden \u00e7\u0131kar\u0131ld\u0131. B\u00f6ylece sa\u011fkal\u0131m\u0131 anlamada daha kritik olan unsurlar \u00f6n plana \u00e7\u0131kt\u0131.<\/p>\n
LASSO y\u00f6ntemiyle d\u00fczenlenmi\u015f Extreme Value modeli, sa\u011fkal\u0131m s\u00fcrecinde belirleyici rol oynayan birka\u00e7 \u00f6nemli klinik de\u011fi\u015fkeni ortaya koydu: hastalar\u0131n rekabet eden risk durumlar\u0131, metastaz varl\u0131\u011f\u0131, kanser evresi ve lenf nodu tutulumu. \u00d6zellikle metastaz durumu, sa\u011fkal\u0131m s\u00fcresinde yakla\u015f\u0131k 2.5 kat daha uzun bir beklentiyle e\u015fle\u015fti. Erken evre te\u015fhisi konulan hastalar\u0131n sa\u011fkal\u0131m\u0131 %26, minimal lenf nodu tutulumu olanlar\u0131n ise %16 oran\u0131nda iyile\u015fti\u011fi belirlendi. Bu rakamlar, klinisyenlerin risk de\u011ferlendirmesinde bu de\u011fi\u015fkenlere \u00f6ncelik vermesini destekliyor.<\/p>\n
Molek\u00fcler belirte\u00e7ler ve t\u00fcm\u00f6r \u00f6zellikleri de modelde ba\u011f\u0131ms\u0131z olarak anlam ta\u015f\u0131d\u0131. HER2-negatif t\u00fcm\u00f6rlere sahip hastalarda sa\u011fkal\u0131m s\u00fcresi HER2-pozitiflere k\u0131yasla %20 daha uzundu. Agresif alt tiplerden biri olan Triple Negative hastalar\u0131n\u0131n yoklu\u011funda ise %15 uzama kaydedildi. T\u00fcm\u00f6r\u00fcn d\u00fc\u015f\u00fck gradeye sahip olmas\u0131 %11 ek sa\u011fkal\u0131m avantaj\u0131 getirirken, hastal\u0131k n\u00fcks\u00fc g\u00f6r\u00fclenlerin sa\u011fkal\u0131m\u0131 %19 oran\u0131nda azald\u0131. Bu de\u011ferler, tedavi planlar\u0131 olu\u015fturulurken t\u00fcm\u00f6r biyolojisinin \u00f6nemini bir kez daha vurgulad\u0131.<\/p>\n
Ara\u015ft\u0131rma ayn\u0131 zamanda klinik anlamda hasta alt gruplar\u0131n\u0131n risk baz\u0131nda s\u0131n\u0131fland\u0131r\u0131lmas\u0131n\u0131 sa\u011flayan bir ara\u00e7 i\u015flevi g\u00f6rd\u00fc. Hastalar d\u00fc\u015f\u00fck, orta ve y\u00fcksek risk s\u0131n\u0131flar\u0131na ayr\u0131larak, Kaplan-Meier e\u011frilerine uygun olarak farkl\u0131 sa\u011fkal\u0131m \u00f6r\u00fcnt\u00fcleri ortaya kondu. Metastaz, lenf nodu durumu, t\u00fcm\u00f6r grade\u2019i, HER2 durumu ve molek\u00fcler alt tipler gibi fakt\u00f6rler, bu alt gruplar\u0131n sa\u011fkal\u0131m farklar\u0131n\u0131 net bir bi\u00e7imde ay\u0131rd\u0131. Bu t\u00fcr risk stratifikasyonu, onkologlar\u0131n hastalara \u00f6zel takip ve tedavi \u015femas\u0131n\u0131 \u015fekillendirmesinde \u00f6nemli bir referans olu\u015fturuyor.<\/p>\n
Rekabet eden risklerin, \u00f6zellikle hastaneye yat\u0131\u015f gibi fakt\u00f6rlerle ili\u015fkilendirilmesi, hastalar\u0131n sadece kanserle de\u011fil, e\u015flik eden di\u011fer sa\u011fl\u0131k sorunlar\u0131yla da m\u00fccadele etmek zorunda oldu\u011funu ortaya koydu. Bu bulgu, entegre bir tedavi yakla\u015f\u0131m\u0131 gereklili\u011fine dikkat \u00e7ekerek, onkolojik bak\u0131m\u0131n yan\u0131nda ek sa\u011fl\u0131k problemlerine y\u00f6nelik y\u00f6netimin hayati \u00f6nem ta\u015f\u0131d\u0131\u011f\u0131n\u0131 g\u00f6sterdi.<\/p>\n
Y\u00fcksek boyutlu verilerde klasik modellerin a\u015f\u0131r\u0131 uyum ya\u015famas\u0131 ve genellenebilirli\u011finin azalmas\u0131 s\u0131k\u00e7a ya\u015fanan bir sorunken, ara\u015ft\u0131rmada LASSO ve di\u011fer d\u00fczenleme y\u00f6ntemlerinin bu problemlere etkin \u00e7\u00f6z\u00fcmler sundu\u011fu ortaya kondu. G\u00fcr\u00fclt\u00fc yaratan veya anlams\u0131z de\u011fi\u015fkenlerin elenmesiyle model daha stabil hale geldi. B\u00f6ylece Extreme Value Frailty AFT modelinde, hem tahmin do\u011frulu\u011fu hem de yorumlanabilirlik \u00f6nemli \u00f6l\u00e7\u00fcde art\u0131r\u0131ld\u0131.<\/p>\n
\u00d6rneklem b\u00fcy\u00fckl\u00fckleri farkl\u0131 oldu\u011funda da modelin performans\u0131 incelendi. Sadece verinin %25\u2019lik k\u0131sm\u0131 kullan\u0131ld\u0131\u011f\u0131nda bile Extreme Value modeli, 100.41\u2019lik AIC puan\u0131yla en yak\u0131n rakibi Log-logistic modelini geride b\u0131rakt\u0131. Bu dayan\u0131kl\u0131l\u0131k, ger\u00e7ek klinik verilerde kar\u015f\u0131la\u015f\u0131lan s\u0131n\u0131rl\u0131 veri sorunlar\u0131na ra\u011fmen modelin g\u00fcvenle kullan\u0131labilece\u011fini i\u015faret ediyor.<\/p>\n
Teorik olarak, Extreme Value da\u011f\u0131l\u0131m\u0131 y\u00fcksek kuyruklu ve u\u00e7 g\u00f6zlemleri modelleme konusundaki esnekli\u011fi ile \u00f6ne \u00e7\u0131k\u0131yor. Di\u011fer da\u011f\u0131l\u0131mlar\u0131n yetersiz kald\u0131\u011f\u0131 durumlarda, bu model heterojen hasta yan\u0131tlar\u0131n\u0131n daha iyi tan\u0131mlanmas\u0131n\u0131 sa\u011fl\u0131yor. Bu \u00f6zellik, klinik alanda b\u00fcy\u00fck \u00f6nem ta\u015f\u0131yan de\u011fi\u015fken hasta tepkilerinin daha do\u011fru tahmin edilmesini m\u00fcmk\u00fcn k\u0131l\u0131yor.<\/p>\n
Omurga niteli\u011findeki orman grafikleri (forest plot) sayesinde, hastalar\u0131n sa\u011fkal\u0131m\u0131n\u0131 etkileyen temel de\u011fi\u015fkenlerin risk oranlar\u0131 ve g\u00fcven aral\u0131klar\u0131 g\u00f6rsel olarak desteklendi. Metastaz ve lenf nodu tutulumun en belirgin prognostik fakt\u00f6rler oldu\u011fu grafiklerde a\u00e7\u0131k\u00e7a ortaya kondu. Bu veriler, klinisyenlerin bilgi temelli karar verme s\u00fcre\u00e7lerini g\u00fc\u00e7lendiren \u00f6nemli referanslar sa\u011flad\u0131.<\/p>\n
Kaplan-Meier e\u011frileri ise molek\u00fcler alt tiplerdeki sa\u011fkal\u0131m farkl\u0131l\u0131klar\u0131n\u0131 \u00e7arp\u0131c\u0131 \u015fekilde ortaya koydu. Triple Negative ve HER2 pozitif hastalar\u0131n en k\u00f6t\u00fc sa\u011fkal\u0131m sonu\u00e7lar\u0131na sahip olduklar\u0131 net bi\u00e7imde g\u00f6zlemlendi. Bu durum, daha hedefe y\u00f6nelik ve alt tipe \u00f6zg\u00fc tedavi yakla\u015f\u0131mlar\u0131n\u0131n geli\u015ftirilmesinin gereklili\u011fini peki\u015ftirdi.<\/p>\n
Sonu\u00e7 olarak, Bosson-Amedenu ve ekibinin bu \u00f6nemli \u00e7al\u0131\u015fmas\u0131, geli\u015fmi\u015f frailty AFT modellerinin ve d\u00fczenleme tekniklerinin meme kanseri sa\u011fkal\u0131m\u0131 konusunda dokuyu \u00e7\u00f6zen yeni bir ara\u00e7 oldu\u011funu g\u00f6sterdi. \u00d6zellikle LASSO ile desteklenen Extreme Value modelinin, hem istatistiksel hem de klinik anlamda kayda de\u011fer kazan\u0131mlar sa\u011flad\u0131\u011f\u0131 ortaya \u00e7\u0131kt\u0131. Bu metodoloji, onkologlar, epidemiyologlar ve biyostatistik\u00e7iler i\u00e7in kanser prognozu ve bireyselle\u015ftirilmi\u015f tedavi planlamalar\u0131nda \u00f6nemli bir destek sistemi olu\u015fturuyor.<\/p>\n
\u00c7al\u0131\u015fma, meme kanserinin \u00f6tesinde, heterojen yap\u0131ya sahip ba\u015fka hastal\u0131klar\u0131n sa\u011fkal\u0131m analizlerinde de benzer yakla\u015f\u0131mlar\u0131n kullan\u0131labilece\u011fini g\u00f6stermesi a\u00e7\u0131s\u0131ndan geni\u015f bir perspektif sunuyor. B\u00f6ylece modern t\u0131ptaki ki\u015fiselle\u015ftirilmi\u015f tedavi anlay\u0131\u015f\u0131, bu t\u00fcr sofistike modellerle daha da sa\u011flam temeller \u00fczerine oturacak. Artan veri hacimleri ve bilgi i\u015flem kapasiteleriyle bu y\u00f6nelim, \u00f6n\u00fcm\u00fczdeki y\u0131llarda sa\u011fl\u0131k ara\u015ft\u0131rmalar\u0131nda kritik bir rol oynayacak gibi g\u00f6r\u00fcn\u00fcyor.<\/p>\n
—<\/p>\n
Ara\u015ft\u0131rma Konusu<\/strong>: Makale Ba\u015fl\u0131\u011f\u0131<\/strong>: Haberin Yay\u0131n Tarihi<\/strong>: Web References<\/strong>: Doi Referans<\/strong>: Resim Credits<\/strong>: Anahtar Kelimeler<\/strong>: Kanser prognostik de\u011ferlendirmelerinde, sa\u011fkal\u0131m analizi bilimsel ara\u015ft\u0131rmalar\u0131n temel ta\u015flar\u0131ndan biri olmaya devam ediyor. Son d\u00f6nemde BMC Cancer dergisinde yay\u0131mlanan \u00f6nemli bir \u00e7al\u0131\u015fma, g\u00f6rece s\u0131n\u0131rl\u0131 kullan\u0131lan h\u0131zland\u0131r\u0131lm\u0131\u015f ya\u015fam s\u00fcresi (Accelerated Failure Time – AFT) frailty modellerini, ileri d\u00fczey d\u00fczenleme teknikleriyle (regularization) birle\u015ftirerek meme kanseri sa\u011fkal\u0131m tahminlerindeki belirsizlikleri anlamay\u0131 ve model performans\u0131n\u0131 art\u0131rmay\u0131 ama\u00e7lad\u0131. Bu \u00e7al\u0131\u015fma, hem…<\/p>\n","protected":false},"author":1,"featured_media":2697,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_yoast_wpseo_title":"","_yoast_wpseo_metadesc":"","_yoast_wpseo_focuskw":"","rank_math_title":"","rank_math_description":"","rank_math_focus_keyword":"","_wpan_schema_json_ld":"","_wpan_ai_seo_metadata":"","_wpan_ai_seo_status":"","_wpan_ai_seo_policy":"","_wpan_ai_seo_faq_block":"","_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[28],"tags":[316,318,315,319,317],"tmauthors":[],"class_list":{"0":"post-2696","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-kanser","8":"tag-hizlandirilmis-yasam-suresi-frailty-modelleri","9":"tag-ileri-duzey-frailty-modelleri-karsilastirmasi","10":"tag-meme-kanseri-sagkalim-analizi","11":"tag-meme-kanseri-sagkalim-model-performans-degerlendirmesi","12":"tag-meme-kanseri-sagkalim-tahmininde-duzenleme-teknikleri"},"jetpack_featured_media_url":"https:\/\/haber360.com\/wp-content\/uploads\/2025\/04\/Meme-Kanseri-Ileri-Duzey-Sagkalim-Modeli-Karsilastirmasi-1744774172.jpg","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/posts\/2696","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/comments?post=2696"}],"version-history":[{"count":0,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/posts\/2696\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/media\/2697"}],"wp:attachment":[{"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/media?parent=2696"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/categories?post=2696"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/tags?post=2696"},{"taxonomy":"tmauthors","embeddable":true,"href":"https:\/\/haber360.com\/index.php\/wp-json\/wp\/v2\/tmauthors?post=2696"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\nMeme kanseri sa\u011fkal\u0131m tahminlerinde d\u00fczenleme teknikleri ile desteklenmi\u015f h\u0131zland\u0131r\u0131lm\u0131\u015f ya\u015fam s\u00fcresi (AFT) frailty modellerinin kar\u015f\u0131la\u015ft\u0131r\u0131lmas\u0131 ve prognostik belirleyicilerin de\u011ferlendirilmesi.<\/p>\n
\nEvaluating key predictors of breast cancer through survival: a comparison of AFT frailty models with LASSO, ridge, and elastic net regularization<\/p>\n
\n2025<\/p>\n
\nhttps:\/\/doi.org\/10.1186\/s12885-025-14040-z<\/p>\n
\nhttps:\/\/doi.org\/10.1186\/s12885-025-14040-z<\/p>\n
\nScienmag.com<\/p>\n
\nH\u0131zland\u0131r\u0131lm\u0131\u015f Ya\u015fam S\u00fcresi Modelleri, Meme Kanseri Sa\u011fkal\u0131m Analizi, Frailty Modelleri, LASSO Regularizasyonu, Y\u00fcksek Boyutlu Klinik Veri, Kanser Prognostik Belirleyicileri, \u0130statistiksel Model Kar\u015f\u0131la\u015ft\u0131rmas\u0131, Regularizasyon Teknikleri, Ki\u015fiselle\u015ftirilmi\u015f T\u0131p, Overfitting Sorunu, Risk Stratifizasyonu<\/p>\n","protected":false},"excerpt":{"rendered":"