关于「实证闭环:比价与托管台账、试算器与预注册检验」,《跨境齿科与健康服务的总成本与托管》给出的定量结论是什么?
结论:第八章给出比价与托管台账、蒸馏小工具「跨境服务总成本与托管放款试算器」与预注册检验方案:总成本检验需 71 份配对记录,信号检验每组需 181 份方案。书中以标注框区分已发表的证据、可运行的 Python 计算(20 段程序)、原典与作者的假说。
数据口径:书中价格、概率与成本均为示意参数,不对应任何真实机构、国家或价格,也不点名评价任何诊所、医院或中介;全部计算由书中程序生成。 本书是非个人化的医疗、财务与法律信息,不是建议,不写任何功效或成功率承诺;方案须由本人与持牌牙医或医师当面决定,资金托管须由持牌第三方办理。
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: 181, 71
| z(0.975) | 1.9600, z(0.80) = 0.8416 |
| H1 paired, delta 500, sd 1,500: n | 71 |
| H2 rework 0.06 vs 0.15: n | 181 per group |
| simulated power at that size, 2,000 reps | 0.818 |
| H3 zero failures in n stages, upper bound < 0.0364: n | 81 |
| bound at that n | 0.0363 |
查看完整验算程序
# Chapter 8. p82: sample sizes for the pre-registered one-year checks.
# H1 paired: realised minus ledger-predicted first-year extra cost,
# detect a mean shift of 500 U when the s.d. is 1,500 U.
# H2 two groups: rework within a year, clinics with >= 3 separating
# signals (0.06) against <= 1 signal (0.15); normal approximation,
# then a seeded simulation of the power at that size.
# H3 provider failure per stage: how many stages, all observed without
# a failure, push the exact one-sided 95 % upper bound below the
# escrow break-even 0.0364 of p53.
# alpha = 0.05 two-sided, power 0.80. Effect sizes illustrative.
import math
import random
from statistics import NormalDist
Z = NormalDist()
ZA, ZB = Z.inv_cdf(0.975), Z.inv_cdf(0.80)
n1 = math.ceil(((ZA + ZB) * 1500.0 / 500.0) ** 2)
print("z(0.975) = {:.4f}, z(0.80) = {:.4f}".format(ZA, ZB))
print("H1 paired, delta 500, sd 1,500: n = {}".format(n1))
p1, p2 = 0.06, 0.15
pb = 0.5 * (p1 + p2)
num = ZA * math.sqrt(2 * pb * (1 - pb)) + ZB * math.sqrt(
p1 * (1 - p1) + p2 * (1 - p2))
n2 = math.ceil((num / (p2 - p1)) ** 2)
print("H2 rework 0.06 vs 0.15: n = {} per group".format(n2))
rng = random.Random(20261120)
REPS, hits = 2000, 0
for _ in range(REPS):
a = 0
b = 0
for _ in range(n2):
a += rng.random() < p1
b += rng.random() < p2
pa, pbb = a / n2, b / n2
pp = (a + b) / (2 * n2)
se = math.sqrt(2 * pp * (1 - pp) / n2)
if se > 0 and abs(pa - pbb) / se > ZA:
hits += 1
print(" simulated power at that size, {:,} reps: {:.3f}".format(
REPS, hits / REPS))
n3 = 1
while 1.0 - 0.05 ** (1.0 / n3) >= 0.0364:
n3 += 1
print("H3 zero failures in n stages, upper bound < 0.0364: n = {}".format(
n3))
print(" bound at that n: {:.4f}".format(1.0 - 0.05 ** (1.0 / n3)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《跨境高端健康与齿科服务定价、多阶段品质信号甄别与客户信任托管母典》第 8 章「实证闭环:比价与托管台账、试算器与预注册检验」(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-338 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026338,
author = {Fan, Yafu},
title = {Cross-Border Dental and Wellness Service Pricing, Multi-Stage Quality Signalling and Client Escrow Mechanics},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}