交互作用图 (Interaction Plot)¶
用途：展示治疗与协变量（或两个因子）之间的交互效应，用于亚组分析和交互效应探索。
交互式图表¶
不平行的线提示存在交互效应。Drug 在低 Biomarker 组效果更明显。
生成代码¶
library(plotly)
# 模拟交互数据
interact_data <- expand.grid(
Treatment = c("Drug", "Placebo"),
Biomarker = c("Low", "Medium", "High")
)
interact_data$Endpoint <- c(12, 10, 8, 10, 10, 10) # 交互模式
fig <- plot_ly(interact_data, x = ~Biomarker, y = ~Endpoint,
color = ~Treatment, type = "scatter",
mode = "lines+markers",
line = list(width = 3),
marker = list(size = 10)) |>
layout(
title = "Treatment × Biomarker Interaction",
yaxis = list(title = "Mean Endpoint"),
xaxis = list(title = "Biomarker Level")
)
fig
import plotly.express as px
import pandas as pd
interact_data = pd.DataFrame({
"Treatment": ["Drug", "Drug", "Drug", "Placebo", "Placebo", "Placebo"],
"Biomarker": ["Low", "Medium", "High"] * 2,
"Endpoint": [12, 10, 8, 10, 10, 10]
})
fig = px.line(interact_data, x="Biomarker", y="Endpoint",
color="Treatment", markers=True,
title="Treatment × Biomarker Interaction")
fig.show()
解读¶
- 平行线 → 无交互
- 交叉线 → 存在交互，提示 treatment effect 依赖于协变量水平
AI Prompt 模板¶
生成交互作用图。数据包含 treatment (Drug/Placebo)、
biomarker_level (Low/Medium/High) 和 mean_endpoint。
x 轴为 biomarker_level，y 轴为 mean_endpoint，
两条线分别代表 Drug 和 Placebo，显示点估计值。