关于「实践与预注册检验」,《仪式神经学与空间切换》给出的定量结论是什么?

结论:第八章给出四步核查表与一项由独立机构实施的预注册三臂随机对照试验方案(TSST 应激后,唯一主要比较为空间切换 + 仪式 + 慢呼吸对同室慢呼吸,唯一主要结局为第 15–20 分钟自然呼吸下的 ln RMSSD,协方差分析,最小关注效应 0.15,90% 把握度下每臂 145 人、共招募 495 人,判定规则写成代码)。书中以标注框区分已发表的证据、可运行的 Python 计算(16 段程序)、原典与作者的模型和假说;书中数据除注明的文献数据外均为合成数据。

数据口径:仪式能不能降低焦虑?走过一道门能不能「清空」压力?慢呼吸能不能经迷走神经修复内分泌与免疫?本书是教学与研究用的数学模型与文献评述,不是医疗建议,书中的练习都不是疾病的治疗方案,任何症状都应先由医生评估。 书中以标注框区分已发表的证据、可运行的 Python 计算(16 段程序)、原典与作者的模型和假说;书中数据除注明的文献数据外均为合成数据。

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: 0.15, 145, 495, 90

normal theory144.1 -> 145 per arm; t-corrected 144.8 -> 145 per arm
without the baseline covariate (rho0) it would be 283 per arm
simulation, 3,000 trials of 145 per arm: power 0.907 (mean estimate 0.150);
type I error at delta0: 0.057 (Monte Carlo s.e. about 0.004)
allowing 10% loss of the primary outcome: recruit 165 per arm, 495 in three arms (11:1)
sensitivity: completers per arm for 90 % power
SESOI (ln units) sd 0.45 sd 0.55 sd 0.65
rho .6.7 .8 rho .6 .7 .8 rho .6 .7 .8
0.10 (x1.105)273 218 154 407 325 229 569 453 320
0.15 (x1.162)122 97 69 181 145 102 253 202 143
0.20 (x1.221)69 55 39 102 82 58 143 114 80
查看完整验算程序
# Sample size for the Chapter 8 trial. Primary comparison: arm 1 (spatial transition +
# fixed ritual sequence + paced breathing) vs arm 2 (paced breathing in the same room).
# Primary outcome: ln RMSSD in the 15-20 min window after the TSST, by ANCOVA with
# baseline ln RMSSD as covariate. Two-sided alpha 0.05, power 0.90. Planning values (to be
# confirmed by the independent team from its own pilot or published data): sd of the
# outcome 0.55 ln units, baseline-outcome correlation 0.70, SESOI 0.15 ln units.
#   n per arm = 2 (z_{1-a/2} + z_{1-b})^2 sd^2 (1 - rho^2) / delta^2   (normal theory)
# then a t-based correction and a simulation check of the ANCOVA test.
import math, random, statistics

Z = statistics.NormalDist().inv_cdf
SD, RHO, DELTA, ATTR = 0.55, 0.70, 0.15, 0.10

def n_normal(sd, rho, delta, power=0.90):
    return 2 * (Z(0.975) + Z(power)) ** 2 * sd ** 2 * (1 - rho ** 2) / delta ** 2

def t975(df):                                   # Abramowitz-Stegun 26.7.5
    z = Z(0.975)
    return (z + (z ** 3 + z) / (4 * df)
            + (5 * z ** 5 + 16 * z ** 3 + 3 * z) / (96 * df ** 2))

def ancova_t(y, x, g):
    n = len(y)
    mx = math.fsum(x) / n; mg = math.fsum(g) / n; my = math.fsum(y) / n
    sxx = math.fsum((a - mx) ** 2 for a in x); sgg = math.fsum((b - mg) ** 2 for b in g)
    sxg = math.fsum((a - mx) * (b - mg) for a, b in zip(x, g))
    sxy = math.fsum((a - mx) * (c - my) for a, c in zip(x, y))
    sgy = math.fsum((b - mg) * (c - my) for b, c in zip(g, y))
    det = sxx * sgg - sxg ** 2
    bx = (sgg * sxy - sxg * sgy) / det; bg = (sxx * sgy - sxg * sxy) / det
    res = math.fsum((c - my - bx * (a - mx) - bg * (b - mg)) ** 2
                    for a, b, c in zip(x, g, y))
    s2 = res / (n - 3)
    return bg, bg / math.sqrt(s2 * sxx / det)

def sim_power(n, delta, reps, seed):
    rng = random.Random(seed); hits = 0; crit = t975(2 * n - 3); est = []
    for _ in range(reps):
        x, y, g = [], [], []
        for arm in (0, 1):
            for _ in range(n):
                b = rng.gauss(0, SD)
                e = RHO * b + math.sqrt(1 - RHO ** 2) * rng.gauss(0, SD)
                x.append(b); y.append(e + delta * arm); g.append(arm)
        bg, t = ancova_t(y, x, g); est.append(bg)
        hits += abs(t) > crit
    return hits / reps, math.fsum(est) / reps

n0 = n_normal(SD, RHO, DELTA)
n1 = n0
for _ in range(5):                              # replace z_{a/2} by t_{df} iteratively
    df = 2 * math.ceil(n1) - 3
    n1 = 2 * (t975(df) + Z(0.90)) ** 2 * SD ** 2 * (1 - RHO ** 2) / DELTA ** 2
n_arm = math.ceil(n1)
rec = math.ceil(n_arm / (1 - ATTR) / 5) * 5
print(f"normal theory: {n0:.1f} -> {math.ceil(n0)} per arm; t-corrected {n1:.1f} ->"
      f" {n_arm} per arm")
print(f"without the baseline covariate (rho = 0) it would be"
      f" {math.ceil(n_normal(SD, 0.0, DELTA))} per arm")
pw, mest = sim_power(n_arm, DELTA, 3000, 8)
p0, m0 = sim_power(n_arm, 0.0, 3000, 9)
print(f"simulation, 3,000 trials of {n_arm} per arm: power {pw:.3f}"
      f" (mean estimate {mest:.3f});")
print(f"   type I error at delta = 0: {p0:.3f} (Monte Carlo s.e. about"
      f" {math.sqrt(0.05 * 0.95 / 3000):.3f})")
print(f"allowing {ATTR:.0%} loss of the primary outcome: recruit {rec} per arm,"
      f" {3 * rec} in three arms (1:1:1)")
print("sensitivity: completers per arm for 90 % power")
print("   SESOI (ln units)   sd 0.45            sd 0.55            sd 0.65")
print("                    rho .6  .7  .8     rho .6  .7  .8     rho .6  .7  .8")
for d in (0.10, 0.15, 0.20):
    cells = []
    for sd in (0.45, 0.55, 0.65):
        cells.append(" ".join(f"{math.ceil(n_normal(sd, r, d)):3d}"
                              for r in (0.6, 0.7, 0.8)))
    print(f"   {d:5.2f} (x{math.exp(d):.3f})  " + "     ".join(cells))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《仪式神经学、空间边界切换与自我转移的计算生物医学》第 8 章「实践与预注册检验」(版本 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23839910

DOI: 10.5281/zenodo.23839910 · 阅读本章所在专著 · 5DL-B-051 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026051,
  author    = {Fan, Yafu},
  title     = {Neurobiology of Ritual, Spatial Boundary Transitions and Computational Biomedicine of Self-Transcendence},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  doi       = {10.5281/zenodo.23839910},
  url       = {https://doi.org/10.5281/zenodo.23839910}
}

本书其他章节结论