关于「一万步的出身:步数—风险样条与年龄平台」,《最大摄氧量、二区与五禽戏》给出的定量结论是什么?

结论:第二章追溯一万步的出身,用先降后平的样条把步数平台放在约 6500 步,并拒绝「没有平台」的纯线性说法。

数据口径:书中参数凡未注明出处者均为示意;全部计算由书中程序生成;本书不是医疗建议,不宣称任何活动对任何疾病有作用,自我试验须先咨询执业医师。 第六章为五禽戏估出示意的代谢当量,一轮约 43.75 MET 分钟,并排出周剂量。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 6500

knotHR/1000 stepsHR at knotchi2delta
50000.6580.5453.661.70
60000.7760.5372.570.61
65000.8030.5241.960.00
70000.8260.5172.940.99
75000.8440.5124.562.60
80000.8560.5015.413.45
90000.8640.4515.083.12
100000.8730.4177.555.59
110000.8810.39011.309.34
120000.8810.34311.309.34
130000.8810.30311.309.34
best knot 6500; knots not rejected: 5000 to 9000
查看完整验算程序
# Chapter 2. p22: Where is the plateau? A linear-then-flat spline for the
# log hazard, log HR(s) = b * (min(s, K) - 3553), fitted by inverse-variance
# weighted least squares to the three published quartile contrasts of
# Paluch et al. (2022). For each knot K we report the best slope b and the
# weighted residual chi-square; knots within 3.84 of the minimum are not
# rejected at the 5 % level. Three points cannot pin a knot down: that is
# the falsifiable content of "the curve levels off at 6000-8000 steps".
import math
from statistics import NormalDist

Z = NormalDist().inv_cdf(0.975)
S0 = 3553
PTS = [(5801, 0.60, 0.51, 0.71), (7842, 0.55, 0.49, 0.62),
       (10901, 0.47, 0.39, 0.57)]


def fit(knot):
    xs, ys, ws = [], [], []
    for s, hr, lo, hi in PTS:
        se = (math.log(hi) - math.log(lo)) / (2 * Z)
        xs.append(min(s, knot) - S0)
        ys.append(math.log(hr))
        ws.append(1 / se ** 2)
    sxy = math.fsum(w * x * y for x, y, w in zip(xs, ys, ws))
    sxx = math.fsum(w * x * x for x, w in zip(xs, ws))
    b = sxy / sxx
    chi = math.fsum(w * (y - b * x) ** 2 for x, y, w in zip(xs, ys, ws))
    return b, chi


if __name__ == "__main__":
    knots = list(range(5000, 13001, 1000)) + [6500, 7500]
    knots.sort()
    res = [(k, fit(k)) for k in knots]
    best = min(res, key=lambda r: r[1][1])
    print(" knot   HR/1000 steps   HR at knot   chi2   delta")
    for k, (b, chi) in res:
        print("%6d   %10.3f   %10.3f   %6.2f   %5.2f"
              % (k, math.exp(1000 * b), math.exp(b * (k - S0)),
                 chi, chi - best[1][1]))
    ok = [k for k, (b, chi) in res if chi - best[1][1] <= 3.84]
    print("best knot %d; knots not rejected: %d to %d"
          % (best[0], min(ok), max(ok)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《最大摄氧量死亡率剂量响应、二区有氧乳酸阈与五禽戏间歇代谢当量母典》第 2 章「一万步的出身:步数—风险样条与年龄平台」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-187 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026187,
  author    = {Fan, Yafu},
  title     = {The Mortality Dose-Response of VO2max, Zone-2 Lactate Thresholds and the Metabolic Equivalents of the Five-Animal Frolics},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论