EndoTODAY | EndoATLAS | OPD

Parasite | Eso | Sto | Cancer | ESD

Boxim | DEX | Sono | Schedule

Home | Recent | Blog | Links

EndoTODAY ³»½Ã°æ ±³½Ç


[¿ë¾î Á¤¸®]

·ÎÁþ¸ðµ¨(logistic model) : RÀÇ odds(´ëÀÀºñ)ÀÎ R/(1-R)¿¡ log(»ó¿ë ¶Ç´Â ÀÚ¿¬)À» ¾º¿î log(R/(1-R))À» logitÀ̶ó°í ºÎ¸£°í log(R/(1-R))=A'+B'XÀÇ ¸ðµ¨À» Àû¿ëÇÑ´Ù. A', B'Àº ÃÖ¼ÒÀڽ¹ýÀÌ ¾Æ´Ñ maximum likelihood¹ýÀ¸·Î computer¿¡ ÀÇÇØ ±¸ÇÑ´Ù. ¿øÄ¢»ó ºñ¿¬¼Óº¯¼ö´Â regressionÀÌ ºÒ°¡´ÉÇѵ¥ ºñ¿¬¼Óº¯¼ö¿¡¼­µµ regressionÀÌ °¡´ÉÇϵµ·Ï ¸¸µç °ÍÀÌ logitÀÌ´Ù.

¸ð¼öÀû ºÐ¼®¹æ¹ý : ¸ðÁý´Ü¿¡ ´ëÇÑ ¾î¶°ÇÑ °¡Á¤À» ÀüÁ¦·ÎÇÑ Åë°èÀû ºÐ¼®¹æ¹ýÀ¸·Î t-test, Pearson's correlation coefficient, Pearson's chi-square test µîÀÌ ´ëÇ¥ÀûÀÌ´Ù. ¸ðÁý´ÜÀ» °¡Á¤Çϱ⠶§¹®¿¡ Ç¥º»ÀÇ Å©±â°¡ ¾î´À Á¤µµ Ä¿¾ß Çϴµ¥ ÀÇÇп¡¼­´Â 30°³ ÀÌ»óÀÇ Ç¥º»À» ºÐ¼®ÇÒ ¶§ ÀÌ¿ëµÈ´Ù.

Á¤º¸Æí°ß(information bias) : Á¤º¸¸¦ ¹Ù¸£°Ô ÃøÁ¤ÇÏÁö ¸øÇÏ¿´±â ¶§¹®¿¡ ¹ß»ýÇÑ bias¸¦ ¸»ÇÑ´Ù. ¿¹·Î ȯÀÚ´ëÁ¶±º¿¬±¸¿¡¼­ ȯÀÚ±ºÀÌ ´ëÁ¶±º¿¡ ºñÇÏ¿© °ú°ÅÀÇ À§Çè ³ëÃâ¿©ºÎ¸¦ ±â¾ïÇÏ´Â µ¥ ´õ ¿­½ÉÀ̱⠽±´Ù(recall bias).

F-ratio : Ç¥º»Æò±ÕÄ¡µéÀÌ °¡Áö´Â ºÐ»ê(variance between groups, Áý´Ü°£ ºÐ»ê, B)°ú ÃøÁ¤Ä¡µéÀÌ °¡Áö°í ÀÖ´Â ¿ø·¡ÀÇ ºÐ»ê(variance within groups, Áý´Ü³» ºÐ»ê, W)ÀÇ ºñ. Áï [F-ratio=(Áý´Ü°£ºÐ»ê/Áý´Ü³» ºÐ»ê)]

°á·ÐÀÇ Çؼ®: ÀϹÝÀûÀ¸·Î ¿¬±¸ÀÚµéÀº Åë°èÀûÀÎ À¯ÀǼº¿¡ Áö³ªÄ£ ºñÁßÀ» µÎ´Â °æÇâÀÌ ÀÖ´Ù. ±×·¯³ª Åë°èÀûÀÎ À¯ÀǼºÀÌ ÀÖ´Ù°í ÇÏ´õ¶óµµ ½ÇÁ¦ÀûÀÎ À¯ÀǼº(practical significance) ¶Ç´Â ÀÓ»óÀû À¯ÀǼº(clinical significance)ÀÌ ¾ø´Â ¹®Á¦µµ ¾ó¸¶µçÁö ÀÖÀ» ¼ö ÀÖ´Ù. µû¶ó¼­ Åë°èÀû À¯ÀǼº¿¡ ´ëÇÑ Áö³ªÄ£ ÁýÂø°ú È®´ëÇؼ­Àº ¿Ã¹Ù¸¥ °ÍÀÌ ¾Æ´Ï´Ù. ÀÌ¿¡ ´ëÇÑ ´ë¾ÈÀ¸·Î ½Å·Ú±¸°£(confidence interval)À» ÀÌ¿ëÇÑ ±¸°£ÃßÁ¤(interval estimation)ÀÌ ÃßõµÈ´Ù.

© ÀÏ¿ø³»½Ã°æ±³½Ç ¹Ù¸¥³»½Ã°æ¿¬±¸¼Ò ÀÌÁØÇà. EndoTODAY Endoscopy Learning Center. Lee Jun Haeng.