6257rv7 发表于 2024-10-3 06:49:58

预测模型校准曲线 | Calibration curve (上篇)

1论文实例<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2016年<span style="color: black;">发布</span>在&nbsp;J Clin Oncol&nbsp;(SCI影响因子26分)的<span style="color: black;">科研</span>对大肠癌<span style="color: black;">病人</span>术前运用放射组学<span style="color: black;">办法</span>,对淋巴结转移<span style="color: black;">状况</span><span style="color: black;">创立</span>预测模型与模型验证。<span style="color: black;">Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer.</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaPlFY48j3p2rv0emNnAPnJNlDJVKQmUZiaPW5n90sXX7n4K4NCvnSiaEw/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">做了预测模型<span style="color: black;">校准曲线 (Calibration curve)</span>,两个图分别是建模队列和验证队列。图的横坐标是预测概率:用预测模型对事件<span style="color: black;">出现</span>的可能性(Probability)进行预测,0到1<span style="color: black;">暗示</span>发生事件可能性是0到100%。纵坐标是<span style="color: black;">实质</span>概率:<span style="color: black;">病人</span><span style="color: black;">实质</span>的事件<span style="color: black;">出现</span>率。红色的线是拟合线,<span style="color: black;">暗示</span>预测值对应的<span style="color: black;">实质</span>值。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">结果<span style="color: black;">诠释</span>:</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">倘若</span>预测值=<span style="color: black;">实质</span>值,则红线与参考线(蓝线)完全重合;</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">倘若</span>预测值><span style="color: black;">实质</span>值,即<span style="color: black;">高估了<span style="color: black;">危害</span></span>,则红线在蓝线下面;</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">倘若</span>预测值<<span style="color: black;">实质</span>值,即<span style="color: black;">低估了<span style="color: black;">危害</span></span>,则红线在蓝线上面;</p><span style="color: black;">倘若</span>能把点估计的波动范围展示出来证据级别高<span style="color: black;">有些</span>。2论文实例<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2008年<span style="color: black;">发布</span>在&nbsp;J Clin Oncol&nbsp;的预测结肠癌复发的<span style="color: black;">科研</span>。<span style="color: black;">Individualized prediction of colon cancer recurrence using a nomogram.</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaPmycZXEcfoYbfBa5F0Ar1j8KAfHdHp6JQZAusRJRr6BFXCISABmJEQ/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">图3是校准曲线</span>,AB两图分别预测60个月和120个月<span style="color: black;">结果</span>事件。<span style="color: black;">一样</span>横纵坐标分别是预测概率和<span style="color: black;">实质</span>概率。与论文实例1<span style="color: black;">区别</span>的是:</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">1、坐标范围不是0-1,而是<span style="color: black;">按照</span><span style="color: black;">实质</span><span style="color: black;">状况</span>的取值范围绘制的。<span style="color: black;">这般</span>做是符合做图规范的。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2、展示了点估计的波动范围。证据级别高<span style="color: black;">有些</span>。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">3、本<span style="color: black;">科研</span>把<span style="color: black;"><span style="color: black;">区别</span>时间<span style="color: black;">出现</span>的<span style="color: black;">结果</span>画在两个图上</span>,下面这篇论文则呈<span style="color: black;">此刻</span>一张图上。</p>3论文实例<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2011年<span style="color: black;">发布</span>在&nbsp;Lancet Oncol(SCI影响因子36分)的<span style="color: black;">科研</span>,对慢性乙型肝炎<span style="color: black;">发展</span>为肝细胞癌<span style="color: black;">创立</span>预测模型进行<span style="color: black;">危害</span><span style="color: black;">评定</span>和验证。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2Za2rRdJvl9HDd9tm2HwtU9NmOu4dqHib4jJPOAVu0Sw7teBnic97sYEhgg/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">Calibration chart展示了三条线:Y<span style="color: black;">出现</span>时间分别是3、5和10年</span>(红、蓝和绿线)。结果<span style="color: black;">诠释</span>:</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">1、<span style="color: black;">当<span style="color: black;">危害</span>较低时</span>(<span style="color: black;">少于</span>10%),三条线均在参考线的上面,即<span style="color: black;">低<span style="color: black;">估</span></span><span style="color: black;">了<span style="color: black;">危害</span></span>;</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2、<span style="color: black;">当<span style="color: black;">危害</span>较高时</span>(大于10%),蓝线和绿线与参考线很接近(基本重合),即<span style="color: black;">未高估或低估<span style="color: black;">危害</span></span>;</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">3、横纵坐标轴的刻度间距不是等距分布的,而是等比例(10倍)分布的。这是<span style="color: black;">按照</span>本<span style="color: black;">科研</span>数据分析<span style="color: black;">状况</span><span style="color: black;">选取</span>的,符合做图规范。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">4、本例<span style="color: black;">无</span><span style="color: black;">表示</span><span style="color: black;">每一个</span>点的波动范围。目的是<span style="color: black;">表示</span>三条线的分布,<span style="color: black;">倘若</span>要<span style="color: black;">表示</span>波动范围,最好<span style="color: black;">掰开</span>做三个图(如论文实例2)</p>4论文实例<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">有的<span style="color: black;">科研</span>点会密<span style="color: black;">有些</span>,<span style="color: black;">这儿</span>有10条线。波动范围大,与样本量小<span style="color: black;">相关</span>。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaJibt2XBJfOVvBNjGFVaYjQ4P7vjRn5g4XutUibSBobqbyGlOHSUicZCbw/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">Nomogram for Preoperative Estimation of Microvascular Invasion Risk in Hepatitis B Virus–Related Hepatocellular Carcinoma Within the Milan Criteria. JAMA Surgery, 2015.&nbsp; SCI IF=8.4</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">五花八门的Calibration curve</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">该<span style="color: black;">怎样</span><span style="color: black;">选取</span>呢?</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">JAMA<span style="color: black;">发布</span>指南出大招</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2017年在JAMA上<span style="color: black;">发布</span>的临床预测模型的区分和校准指南。Discrimination and Calibration of Clinical Prediction Models: Users Guides to the Medical Literature. JAMA, 2017.</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">指南中给的Calibration curve就更<span style="color: black;">美丽</span><span style="color: black;">有些</span>,<span style="color: black;">用曲线形式展示</span>。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2Za92tCXoZHlHibqJvL0UflBXqjtwGxP0ic2ia5h2DnTSu8ibrWl9l7rKQmeA/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">图中横坐标是预测模型得出的风险评分,纵坐标是<span style="color: black;">实质</span>值死亡率。黑色的点和线是<span style="color: black;">实质</span>观察到的1年死亡率和95%CI,<span style="color: black;">黄色的曲线是模型预测的1年的死亡率。</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_gif/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaRNhL2jo7ujtku2p2wu9IJnHwjTZVibuT5qWnAfOzWmxy5cI1ib2eGQtg/640?wx_fmt=gif&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">指南中写到用肉眼看(visual)是最佳的<span style="color: black;">评估</span>校准曲线的方式(is the best way to evaluate calibration)</span>,<span style="color: black;">亦</span><span style="color: black;">说到</span>有统计学<span style="color: black;">办法</span><span style="color: black;">能够</span>计算预测值和观测值的统计学差异(eg, the Hosmer-Lemeshow test),然而指南中并不<span style="color: black;">举荐</span>依靠p值<span style="color: black;">评估</span>校准曲线,并且用了三段来阐述不<span style="color: black;">意见</span><span style="color: black;">按照</span>统计学检验计算p值的<span style="color: black;">原由</span>,<span style="color: black;">这里</span>就不<span style="color: black;">仔细</span>讲述了。</p>5论文实例<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">这<span style="color: black;">指的是</span>南中<span style="color: black;">举荐</span>的参考文献的做法,用曲线拟合的方式呈现校准曲线,<span style="color: black;">并用肉眼结合指标的临床<span style="color: black;">道理</span>判断临床实用价值。</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaVrM8Fiae1tbs0iaAKykdFvJAA8HibYPdYS8UuLQZs7FicLT43jIwLsK1sw/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">可见当<span style="color: black;">实质</span>死亡率大于30%时,预测模型会低估死亡<span style="color: black;">危害</span></span>。临床价值在于:某些<span style="color: black;">病人</span>预测模型得出死亡<span style="color: black;">危害</span>是30%,很可能<span style="color: black;">选取</span><span style="color: black;">药品</span>治疗并推迟心脏移植治疗时间,然而其<span style="color: black;">实质</span>死亡<span style="color: black;">危害</span>可能接近50%。<span style="color: black;">因此呢</span>,<span style="color: black;">运用</span>该模型会<span style="color: black;">引起</span><span style="color: black;">有些</span><span style="color: black;">病人</span>不适当地推迟移植。原文如下:</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaTHwruQwVBLrhf3mfV23H8to1nLvHics418MdliaJOqs83MgWqhJicWQ3A/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">Predicting survival in heart failure: validation of the MAGGIC heart failure risk score in 51 043 patients from the Swedish Heart Failure Registry. European Journal of Heart Failure, 2014. SCI IF=10.6</span></p>6论文实例<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2017年&nbsp;Eur Urol杂志(SCI影响因子17.5分)<span style="color: black;">发布</span>临床<span style="color: black;">科研</span>基于国家癌症数据库,分析局部治疗对转移性前列腺癌<span style="color: black;">病人</span>总<span style="color: black;">存活</span>率的影响。<span style="color: black;">亦</span>采用了曲线拟合的形式做校准曲线,并且做了分层分析。<span style="color: black;">The Impact of Local Treatment on Overall Survival in Patients with Metastatic Prostate Cancer on Diagnosis: A National Cancer Data Base Analysis.</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaCr9lF1jkNlCsHk9htGFbKE4TxNqAroe4ypnhbbWvhGZicMf3tD6eGbg/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">核心结果<span style="color: black;">诠释</span>:横坐标是预测的死亡概率,纵坐标是<span style="color: black;">实质</span>存活概率。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">当预测的死亡概率=30%时,红线在蓝线上面,<span style="color: black;">显示</span>局部治疗(LT)比非局部治疗NLT<span style="color: black;">实质</span>存活率高;</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">当预测的死亡概率≥72%时,红线在蓝线下面,<span style="color: black;">显示</span>LT比NLT<span style="color: black;">实质</span>存活率低;</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">因此呢</span>,得出基线<span style="color: black;">状况</span>好的<span style="color: black;">病人</span>,LT疗效好的结论。<span style="color: black;">临床用途:用基线指标给<span style="color: black;">病人</span>做预测,<span style="color: black;">倘若</span>预测死亡概率<72%,则<span style="color: black;">举荐</span>用局部治疗<span style="color: black;">方法</span>。</span></p><img src="https://mmbiz.qpic.cn/mmbiz_png/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaHEslyQaBUpFHXssv5OJsNLd7PgLzml9ypIFjwTichUQM345iaeXZvHeA/640?wx_fmt=png&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"><span style="color: black;">操作实例</span>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">例如:用多个临床(如年龄、性别和BMI)指标<span style="color: black;">创立</span><span style="color: black;">结果</span>指标Y的预测模型绘制校准曲线。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaWhnbTcflD9fvhCE8ZFSID29qo5P93eUbCp7PCZE2Gso4icG9As3BzsQ/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">首要</span>看数据结构:</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">需要有结果变量(<span style="color: black;">是不是</span><span style="color: black;">出现</span>事件),这个事件<span style="color: black;">能够</span>是死亡(是/否),<span style="color: black;">亦</span><span style="color: black;">能够</span>是发病(是/否)。<span style="color: black;">一般</span><span style="color: black;">无</span><span style="color: black;">出现</span>事件编码为0,<span style="color: black;">出现</span>事件编码为1。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">把原始数据整理到一张表中,<span style="color: black;">每一个</span><span style="color: black;">科研</span>对象一行,全数字编码,缺失的数据<span style="color: black;">能够</span>空着。</p><img src="https://mmbiz.qpic.cn/mmbiz_png/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaHEslyQaBUpFHXssv5OJsNLd7PgLzml9ypIFjwTichUQM345iaeXZvHeA/640?wx_fmt=png&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"><span style="color: black;">易侕软件操作</span>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">第1</span>步:数据操作的<span style="color: black;">计算残差与预测值模块</span>
    </p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">,给出结果变量、自变量(单个或多个指标,<span style="color: black;">能够</span>是连续或<span style="color: black;">归类</span>变量),点击<span style="color: black;">保留</span>。得出后缀是PRED的新变量,范围是0-1,<span style="color: black;">显示</span>预测的事件<span style="color: black;">危害</span>。</p>

    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2Zaics3x6JeahXR5icgwLpekSIP0ujVLA8ZNL62cgLy4YfTQ2fcPxXkAiaicg/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">第二步:数据操作菜单下的<span style="color: black;">“平滑曲线拟合</span>”模块,给出应变量(<span style="color: black;">结果</span>指标)、暴露变量(上一步得出的预测值)、曲线拟合分层因子(group),点击查看结果。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2Za33pzCDAn1SuEicM2icVYxibIKWMlAib9I6Zib243CToeS86SRJicCvM5DkOA/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">结果出来了,横坐标预测的事件<span style="color: black;">出现</span>率,纵坐标是<span style="color: black;">实质</span>事件<span style="color: black;">出现</span>率。红线是曲线拟合线,蓝线是95%CI。点的疏密程度<span style="color: black;">表率</span>样本量。图中绿色的线是参考线,即预测值=<span style="color: black;">实质</span>值的<span style="color: black;">状况</span>。可见红线在绿线的上面,<span style="color: black;">显示</span>预测值<span style="color: black;">危害</span>高于<span style="color: black;">实质</span><span style="color: black;">危害</span>。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaUtVuy48lFekGVdGk4CLQtAX5b66ujicdOE2p7ffibXHrgWn5GXNBydicA/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">对!操作就这么简单!</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_gif/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaSpHuDOe3x2836zk51MmvJQb8SPcfAugV18FkMkIkGNOMiblzeauZTHA/640?wx_fmt=gif&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1" style="width: 50%; margin-bottom: 20px;"></p><img src="https://mmbiz.qpic.cn/mmbiz_png/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2Za5EhpWtUXBic0d0wjoNnfaojIickH6GP92bdOZQd1P4cuTZMiclCK75dqQ/640?wx_fmt=png&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"><img src="https://mmbiz.qpic.cn/mmbiz_png/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaEaNrDPUnsEODrPJ3978alO1AF8UB2CeEsD1rQSibx0P2Aich1Jeiajwpw/640?wx_fmt=png&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;">
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">重视</span>:易侕软件输出的是曲线拟合+95%CI的图(png<span style="color: black;">照片</span>和pdf格式),以及用于绘制曲线的原始数据(Excel格式)。参考线(如上图绿线)是后期做图添加的,其实很简单,两点连起来<span style="color: black;">便是</span>一条线,本例中把0.2和0.6对应的横纵线一画,找到两个交点后连线就ok啦。</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2ZaEJnmGQN1VqaL8toOS00ib4FnDf8ESZ5x8uFRb22EobqW1QIzib9gIahQ/640?wx_fmt=jpeg&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;"></p><img src="https://mmbiz.qpic.cn/mmbiz_png/xrTQhwMJUTKCicVdwEfmYDxlbEsBIg2Zaic3r5a0icCTLx1vCvwgpdO83eOdCQROxiatkCqTbZjRJyfAuKbCcyQ2iag/640?wx_fmt=png&amp;tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;">往期<span style="color: black;">精彩</span>回顾<a style="color: black;">危险<span style="color: black;">原因</span><span style="color: black;">科研</span>需要<span style="color: black;">调节</span><span style="color: black;">那些</span>变量?</a><a style="color: black;">建模与验证 | 预测模</a>型国际规范(TRIPOD)的技术瓶颈破解办法<a style="color: black;">多个<span style="color: black;">结果</span>指标存在竞争关系怎么办?竞争<span style="color: black;">危害</span>模型(Fine &amp; Gray)</a>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">点击官网,免费下载安装易侕软件</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">http://www.empowerstats.com/empowerU/#</p><img src="https://mmbiz.qpic.cn/mmbiz_jpg/xrTQhwMJUTK7Eu9L7mVyAOpwRyqmMLiaWPI8ZNE1nbO3yTYH20VdRAVib8TBb56Noa4AGRw8uDCqwHwHnzUlu2HA/640?tp=webp&amp;wxfrom=5&amp;wx_lazy=1&amp;wx_co=1" style="width: 50%; margin-bottom: 20px;">扫一扫,惊喜<span style="color: black;">持续</span>!<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><strong style="color: blue;">近期培训班</strong></span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">第十六期临床<span style="color: black;">研究</span>设计、数据分析与实战培训</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">贵阳贵州省第二人民医院会议室</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">2018/08/6-10 (8月5日报到)</p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">临床<span style="color: black;">研究</span>工作者<span style="color: black;">发布</span>SCI的新途径</p>利用既往病例资料,将临床经验转化为证据

    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">报名方式见本链接:</span></p>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><a style="color: black;"><span style="color: black;">起始</span>报名啦 | 第十六期临床<span style="color: black;">研究</span>设计、数据分析与实战培训</a></span></p>




nykek5i 发表于 2024-10-7 03:57:52

软文发布平台 http://www.fok120.com/

b1gc8v 发表于 2024-10-20 15:29:33

感谢楼主分享,祝愿外链论坛越办越好!

4zhvml8 发表于 4 天前

楼主果然英明!不得不赞美你一下!
页: [1]
查看完整版本: 预测模型校准曲线 | Calibration curve (上篇)