个体轨迹图 (Spaghetti Plot)¶

用途：展示纵向数据中每个受试者随时间的变化轨迹，叠加组均值趋势线。

交互式图表¶

浅色半透明线 = 个体轨迹，粗实线 = 组均值趋势。悬停查看具体数值。

生成代码¶

RPython

library(plotly)

# 模拟数据
set.seed(123)
n_subj <- 30
visits <- c(0, 4, 8, 12, 16, 20, 24)

spag_data <- do.call(rbind, lapply(1:n_subj, function(i) {
  trt <- ifelse(i <= 15, "Drug", "Placebo")
  bl <- rnorm(1, 50, 10)
  slope <- ifelse(trt == "Drug", -2, -0.5)
  data.frame(
    Subject = paste0("S", i),
    Treatment = trt,
    Visit = visits,
    Value = bl + slope * visits + rnorm(length(visits), 0, 3)
  )
}))

# 个体轨迹
fig <- plot_ly(spag_data, x = ~Visit, y = ~Value,
               color = ~Treatment,
               split = ~Subject,
               type = "scatter", mode = "lines+markers",
               opacity = 0.4, line = list(width = 0.5),
               showlegend = FALSE)

# 叠加组均值
group_mean <- aggregate(Value ~ Treatment + Visit, spag_data, mean)
fig <- fig |> add_trace(data = group_mean,
                        x = ~Visit, y = ~Value,
                        color = ~Treatment,
                        type = "scatter", mode = "lines+markers",
                        opacity = 1, line = list(width = 3),
                        name = ~Treatment)
fig

import plotly.express as px
import pandas as pd
import numpy as np

np.random.seed(123)
n_subj = 30
visits = [0, 4, 8, 12, 16, 20, 24]

rows = []
for i in range(n_subj):
    trt = "Drug" if i < 15 else "Placebo"
    bl = np.random.normal(50, 10)
    slope = -2 if trt == "Drug" else -0.5
    for v in visits:
        rows.append({
            "Subject": f"S{i+1}", "Treatment": trt,
            "Visit": v, "Value": bl + slope*v + np.random.normal(0, 3)
        })
spag_data = pd.DataFrame(rows)

fig = px.line(spag_data, x="Visit", y="Value", 
              color="Treatment", line_group="Subject",
              opacity=0.3,
              title="Spaghetti Plot — Individual Trajectories")

# 叠加组均值
means = spag_data.groupby(["Treatment", "Visit"])["Value"].mean().reset_index()
fig.add_scatter(x=means[means.Treatment=="Drug"].Visit,
                y=means[means.Treatment=="Drug"].Value,
                mode="lines+markers", name="Drug (Mean)",
                line=dict(width=4, color="blue"))
fig.show()

参考方法¶