Razi Journal of Medical Sciences
مجله علوم پزشکی رازی
RJMS
Medical Sciences
http://rjms.iums.ac.ir
39
journal39
2228-7043
2228-7051
en
jalali
1392
8
1
gregorian
2013
11
1
20
113
online
1
fulltext
fa
کاربرد مدل شبکه عصبی مصنوعی در پیشبینی پاسخهای آمیخته بیماری قلبی
Application of artificial neural network model in predicting the mixed response of atherosclerosis disease
آمار حیاتی
Biostatistics
پژوهشي
Research
<font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify line-height: normal unicode-bidi: embed direction: rtl text-justify: kashida text-kashida: 0%" dir="rtl"><b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">زمینه و هدف</span></b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">:
در مطالعات اپیدمیولوژی و پزشکی، گاهی پژوهشگر با مواردی مواجه میشود که لازم است
دو متغیر پاسخ را به صورت توام (همزمان) از روی تعدادی متغیر کمکی پیشبینی نماید.
زمانی که متغیر پاسخ آمیخته باشد، با توجه به محدودیتها و برقرار نبودن برخی پیش
فرضها، روشهای کلاسیک آماری برای مدلبندی و پیشبینی کارایی لازم را ندارند.
هدف این مطالعه بکارگیری مدل شبکه عصبی مصنوعی برای پیشبینی متغیر پاسخ آمیخته در
بیماری قلبی است.<o:p /></span></p><font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify unicode-bidi: embed direction: rtl text-justify: kashida text-kashida: 0%" dir="rtl"><b><span lang="AR-SA" style="line-height: 115% font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">روش کار</span></b><span lang="AR-SA" style="line-height: 115% font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">:
در پاییز و زمستان 1390، تعداد 276 بیمار قلبی که از بیمارستان شهید مدنی خرم آباد
ترخیص شده بودند به صورت کوهورت تاریخی مورد مطالعه قرار گرفتند. از این نمونه
برای پیشبینی توام کلسترول و سطح </span><span style="line-height: 115% font-family: "Times New Roman","serif" font-size: 9pt mso-bidi-font-family: "B Mitra" mso-bidi-font-size: 10.0pt" dir="ltr">LDL</span><span dir="rtl"></span><span lang="AR-SA" style="line-height: 115% font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt"><span dir="rtl"></span> استفاده شد. دادهها به تصادف به دو
گروه آموزش (175 نفر) و آزمون (91 نفر) تقسیم شدند. برای تحلیل دادهها از شبکه
عصبی مصنوعی با الگوریتم شیب توام مقیاس شده (</span><span style="line-height: 115% font-family: "Times New Roman","serif" font-size: 9pt mso-bidi-font-family: "B Mitra" mso-bidi-font-size: 10.0pt" dir="ltr">Scaled Conjugate
Gradient</span><span dir="rtl"></span><span lang="AR-SA" style="line-height: 115% font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt"><span dir="rtl"></span>)
و برای تعیین مناسبترین مدل از معیار صحت پیشبینی استفاده شد. برای تحلیل از نرم
افزار </span><span style="line-height: 115% font-family: "Times New Roman","serif" font-size: 9pt mso-bidi-font-family: "B Mitra" mso-bidi-font-size: 10.0pt" dir="ltr">MATLAB</span><span dir="rtl"></span><span style="line-height: 115% font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt"><span dir="rtl"></span>
<span lang="AR-SA"><font face="B Mitra" size="2">نسخه 11/7 استفاده شد.</font></span></span><span lang="FA" style="font-family: "B Mitra" mso-bidi-language: FA mso-ascii-font-family: Calibri mso-hansi-font-family: Calibri"><o:p /></span></p><font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify line-height: normal unicode-bidi: embed direction: rtl text-justify: kashida text-kashida: 0%" dir="rtl"><b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">یافتهها:</span></b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 13pt mso-bidi-language: FA"> </span><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">بالاترین
صحت پیشبینی برای مدل شبکه عصبی مصنوعی چهار لایه برای متغیر پاسخ آمیخته برابر 76/51
درصد به دست آمد.</span><span lang="FA" style="font-family: "B Mitra" font-size: 12pt mso-bidi-language: FA mso-ascii-font-family: Calibri mso-hansi-font-family: Calibri"><o:p /></span></p><font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify line-height: normal unicode-bidi: embed direction: rtl text-justify: kashida text-kashida: 0%" dir="rtl"><b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">نتیجه</span></b><b><span style="font-family: "Times New Roman","serif" font-size: 9pt mso-bidi-font-size: 10.0pt" dir="ltr"></span></b><b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">گیری</span></b><span lang="AR-SA" style="font-family: "B Mitra" font-size: 10pt mso-ascii-font-family: "Times New Roman" mso-hansi-font-family: "Times New Roman" mso-ansi-font-size: 9.0pt">:
مدل شبکه عصبی مصنوعی با دو لایه میانی برای پیشبینی متغیر پاسخ آمیخته مناسب است.<o:p /></span></p><font face="Times New Roman" size="3"> </font>
<font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify line-height: normal text-justify: inter-ideograph"><b><span style="font-family: "Times New Roman","serif" font-size: 12pt mso-bidi-language: FA">Background: </span></b><span style="font-family: "Times New Roman","serif" font-size: 12pt">In
epidemiological and medical studies, sometimes researchers are faced for
prediction of two response variables (simultaneously) based on a number of
independent variables. When the response variable is mixed, according to
established limits and absence of assumption, the classical statistical methods
are not enough efficient for classification and prediction goals. The purpose
of this study is using Artificial Neural Network (ANN) model to predict the
mixed response variable in heart disease.</span><span style="font-size: 13pt mso-bidi-font-family: "B Mitra""><o:p /></span></p><font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify line-height: normal text-justify: inter-ideograph tab-stops: right 7.3pt center 265.9pt"><b><span style="font-family: "Times New Roman","serif" font-size: 12pt">Methods</span></b><span style="font-family: "Times New Roman","serif" font-size: 12pt">: </span><span style="font-family: "Times New Roman","serif" font-size: 12pt mso-bidi-font-family: Arial mso-bidi-theme-font: minor-bidi">A total of 276 cardiac patients who were
discharged from Madani Hospital were studied as historical cohort, from October
2011 to March 2012. This sample was used to predict the cholesterol and also
LDL levels of patients. Data was randomly divided into two sets: training (175
cases) and testing (91 cases) sets. Data analysis was made by ANN model with
SCG algorithms in MATLAB software, version 7.11 and appropriateness of the
model was assessed by the accuracy prediction.</span><span lang="AR-SA" style="font-family: "Times New Roman","serif" font-size: 12pt" dir="rtl"><o:p /></span></p><font face="Times New Roman" size="3"> </font><h1 style="margin: 0in 0in 0pt text-align: justify line-height: normal text-justify: inter-ideograph"><span style="color: windowtext font-family: "Times New Roman","serif" font-size: 12pt mso-fareast-font-family: "Times New Roman" mso-fareast-theme-font: minor-fareast">Results:</span><span style="font-family: "Calibri","sans-serif" font-size: 12pt mso-bidi-font-family: "B Mitra""><font color="#365f91"> </font></span><span style="color: windowtext font-family: "Times New Roman","serif" font-size: 12pt font-weight: normal mso-fareast-font-family: "Times New Roman" mso-fareast-theme-font: minor-fareast">The highest accuracy of prediction of mixed response
variable was 51.76% for a four-layer ANN model</span><span style="color: windowtext font-family: "Times New Roman","serif" font-size: 12pt font-weight: normal">.<o:p /></span></h1><font face="Times New Roman" size="3"> </font><p class="MsoNormal" style="margin: 0in 0in 0pt text-align: justify line-height: normal text-justify: inter-ideograph tab-stops: right 7.3pt center 265.9pt"><b><span style="font-family: "Times New Roman","serif" font-size: 12pt">Conclusions: </span></b><span style="font-family: "Times New Roman","serif" font-size: 12pt">The ANN model is
suggested to predict the mixed response variable in medical studies.<o:p /></span></p><font face="Times New Roman" size="3"> </font>
شبکه عصبی مصنوعی، پاسخ آمیخته، بیماری قلبی، کلسترول، سطح LDL.
Artificial neural network, Mixed response, Cardiac disease, Cholesterol, LDL level.
20
28
http://rjms.iums.ac.ir/browse.php?a_code=A-10-1-1465&slc_lang=fa&sid=1
Mahin
Adeli
مهین
عادلی
3900319475328460024300
3900319475328460024300
No
University of Social Welfare and Rehabilitation Sciences
دانشگاه علوم بهزیستی و توانبخشی تهران
Akbar
Biglarian
اکبر
بیگلریان
3900319475328460024301
3900319475328460024301
Yes
University of Social Welfare and Rehabilitation Sciences
دانشگاه علوم بهزیستی و توانبخشی تهران
Enayatollah
Bakhshi
عنایتاله
بخشی
3900319475328460024302
3900319475328460024302
No
University of Social Welfare and Rehabilitation Sciences
دانشگاه علوم بهزیستی و توانبخشی تهران
Omid Ali
Adeli
امیدعلی
عادلی
3900319475328460024303
3900319475328460024303
No
Lorestan University of Medical Sciences
دانشگاه علوم پزشکی لرستان