5D Library › 下丘脑垂体肾上腺轴耦合振荡、皮质醇觉醒反应扁平化与倦怠恢复动力学 · English

第五章 五天上,两天下

一个循环,不是一次恢复

一周的工作是一次周期性的驱动,而一个一阶系统在周期性驱动下不会取得进展;它会落进一个循环。那个循环坐在哪里是一个有答案的问题,而答案并不令人鼓舞。Geurts 与 Sonnentag 把恢复阐述为连接急性应激反应与慢性损害的那个机制:要紧的不是每日反应的大小,而是它在下一次之前是否被撤销 (Geurts and Sonnentag 2006)。Sonnenschein 等人发现,以睡眠衡量的恢复受损,会独立于抑郁情绪而使倦怠的改善变得复杂,这是同一个要点的临床形式 (Sonnenschein et al. 2007)。

证据

已发表。 Geurts 与 Sonnentag 论证恢复不足是把急性应激反应与慢性健康损害连起来的机制 (Geurts and Sonnentag 2006)。Sonnenschein 等人发现睡眠恢复受损会独立于抑郁情绪而使倦怠的改善变得复杂 (Sonnenschein et al. 2007)。

推导

# Chapter 5. p09: Why two days off does not undo five days on. Receptor
# density is driven by the cortisol it sees, and cortisol follows the
# load. A working week puts the load on for five days and takes it off
# for two, over and over. Because the rise and the fall are both
# first-order with their own rates, the week settles into a cycle rather
# than making progress, and where that cycle sits is what this program
# computes. The comparison is a full rest of the same total length.
SYNTH, BASAL, EXTRA = 1.0, 1.0 / 28.0, 3.0 / 28.0
C_ON, C_OFF = 2.5, 1.0        # cortisol multiples with the load on, off
WEEKS = 40

def occupancy(c):
    return c / (1.0 + c)

def steady(c):
    return SYNTH / (BASAL + EXTRA * occupancy(c))

def rate(c):
    return BASAL + EXTRA * occupancy(c)

def step(r, c):
    return r + (steady(c) - r) * rate(c)

def run(pattern, days):
    r = steady(C_OFF)
    out = []
    for d in range(days):
        c = C_ON if pattern[d % len(pattern)] else C_OFF
        r = step(r, c)
        out.append(r)
    return out

if __name__ == "__main__":
    full = steady(C_OFF)
    worn = steady(C_ON)
    print("   receptor density rested {:.4f}, fully loaded {:.4f}"
          .format(full, worn))
    print("   the whole range a load can move is {:.4f}"
          .format(full - worn))
    five_two = (1, 1, 1, 1, 1, 0, 0)
    seven = (1,) * 7
    rest = (0,) * 7
    print("   week   five on two off   seven on   relative to rested")
    a = run(five_two, WEEKS * 7)
    b = run(seven, WEEKS * 7)
    for w in (1, 2, 4, 8, 16, 24, 40):
        print("   {:<7d}{:<18.4f}{:<11.4f}{:.4f}"
              .format(w, a[w * 7 - 1], b[w * 7 - 1],
                      a[w * 7 - 1] / full))
    print("   after {} weeks the five-on-two-off cycle sits at {:.4f},"
          .format(WEEKS, a[-1]))
    print("   which is {:.4f} of rested and {:.4f} of the way from the"
          .format(a[-1] / full, (a[-1] - worn) / (full - worn)))
    print("   fully loaded value back towards rested")
    print("   seven days on sits at {:.4f}, which is {:.4f} of rested"
          .format(b[-1], b[-1] / full))
    print("   so the two days off are worth {:.4f} of the gap"
          .format((a[-1] - b[-1]) / (full - b[-1])))
    swing = max(a[-7:]) - min(a[-7:])
    print("   the weekly swing in the settled cycle is only {:.4f},"
          .format(swing))
    print("   which is {:.4f} of the whole range"
          .format(swing / (full - worn)))
    print("   from the settled cycle, days of full rest to pass 0.95"
          " of rested:")
    r = a[-1]
    d = 0
    while r / full < 0.95 and d < 2000:
        r = step(r, C_OFF)
        d += 1
    print("   {}".format(d))
    print("   and to pass 0.99: ", end="")
    while r / full < 0.99 and d < 2000:
        r = step(r, C_OFF)
        d += 1
    print("{}".format(d))
#    receptor density rested 11.2000, fully loaded 8.9091
#    the whole range a load can move is 2.2909
#    week   five on two off   seven on   relative to rested
#    1      10.3476           9.9046     0.9239
#    2      9.9578            9.3417     0.8891
#    4      9.6980            8.9908     0.8659
#    8      9.6323            8.9120     0.8600
#    16     9.6293            8.9091     0.8598
#    24     9.6293            8.9091     0.8598
#    40     9.6293            8.9091     0.8598
#    after 40 weeks the five-on-two-off cycle sits at 9.6293,
#    which is 0.8598 of rested and 0.3144 of the way from the
#    fully loaded value back towards rested
#    seven days on sits at 8.9091, which is 0.7955 of rested
#    so the two days off are worth 0.3144 of the gap
#    the weekly swing in the settled cycle is only 0.3231,
#    which is 0.1410 of the whole range
#    from the settled cycle, days of full rest to pass 0.95 of rested:
#    12
#    and to pass 0.99: 29

在这些单位里,一个负荷所能移动受体密度的整个范围是 2.2909:从静息的 11.2000 到满负荷的 8.9091。五天上班两天休息大约在第八周稳定在 9.6293,也就是静息值的 0.8598,以及从满负荷值往回走了 0.3144。七天都上班稳定在 8.9091,也就是静息值的 0.7955。所以那两天休息值缺口的 0.3144,不多不少,无论重复多少周。

稳定循环内部的摆幅是解释主观体验的那一部分。它只有 0.3231,是整个范围的 0.1410。处在这个循环里的人,周日傍晚会感觉好一点,而跨越数月则完全不会感觉更好,因为每周的变动只有范围的七分之一,而循环的位置并不移动。

从稳定循环出发,十二天彻底休息把密度带过静息值的 0.95,二十九天带过 0.99。这两个数字就是那个周末所要竞争的对象。

五上二下与七上两种情形下逐周的受体密度,图中标出稳定循环(p09)。

那些替代方案值多少

人们实际会选的方案是一张短清单,而它们可以互相定价。

推导

# Chapter 5. p10: Which way of unloading actually works. Starting from
# the settled working-week cycle of p09, five candidate plans are run
# and the days each takes to bring receptor density back to a given
# share of its rested value are reported. The plans are the ones people
# actually choose: keep going and hope, take the weekends, cut the load
# by a third every day, take a fortnight off, take a month off. The
# model has no opinion about which is affordable; it only prices them.
SYNTH, BASAL, EXTRA = 1.0, 1.0 / 28.0, 3.0 / 28.0
C_ON, C_OFF = 2.5, 1.0
CUT = 0.65                 # the load left after cutting it by a third
TARGETS = (0.90, 0.95, 0.99)
HORIZON = 400

def occupancy(c):
    return c / (1.0 + c)

def steady(c):
    return SYNTH / (BASAL + EXTRA * occupancy(c))

def step(r, c):
    return r + (steady(c) - r) * (BASAL + EXTRA * occupancy(c))

def settled():
    r = steady(C_OFF)
    pattern = (1, 1, 1, 1, 1, 0, 0)
    for d in range(399):     # ends on a Sunday, as in p09
        r = step(r, C_ON if pattern[d % 7] else C_OFF)
    return r

def load_of(plan, day):
    if plan == "carry on, seven days":
        return C_ON
    if plan == "weekends off":
        return C_OFF if (day % 7) >= 5 else C_ON
    if plan == "a third less, every day":
        return C_OFF + CUT * (C_ON - C_OFF)
    if plan == "a fortnight off":
        return C_OFF if day < 14 else (C_OFF if (day % 7) >= 5 else C_ON)
    return C_OFF if day < 30 else (C_OFF if (day % 7) >= 5 else C_ON)

PLANS = ("carry on, seven days", "weekends off", "a third less, every day",
         "a fortnight off", "a month off then weekends")

if __name__ == "__main__":
    full = steady(C_OFF)
    start = settled()
    print("   starting density {:.4f}, which is {:.4f} of rested"
          .format(start, start / full))
    print("   plan                          " + "".join(
        "{:>9.2f}".format(t) for t in TARGETS))
    for plan in PLANS:
        r = start
        hits = {t: None for t in TARGETS}
        for day in range(HORIZON):
            r = step(r, load_of(plan, day))
            for t in TARGETS:
                if hits[t] is None and r / full >= t:
                    hits[t] = day + 1
        print("   {:<30}".format(plan) + "".join(
            "{:>9}".format(hits[t] if hits[t] is not None else "never")
            for t in TARGETS))
    print("   where a plan says never, it means not within {} days"
          .format(HORIZON))
    print("   density reached after {} days under each plan:"
          .format(HORIZON))
    for plan in PLANS:
        r = start
        for day in range(HORIZON):
            r = step(r, load_of(plan, day))
        print("   {:<30}{:.4f}   ({:.4f} of rested)"
              .format(plan, r, r / full))
    r = start
    for day in range(30):
        r = step(r, C_OFF)
    print("   a month of complete rest alone reaches {:.4f} of rested"
          .format(r / full))
    print("   the two plans that get there are the ones that remove the")
    print("   load rather than reduce it, which is the finding a reader")
    print("   is least likely to want")
#    starting density 9.6293, which is 0.8598 of rested
#    plan                               0.90     0.95     0.99
#    carry on, seven days              never    never    never
#    weekends off                      never    never    never
#    a third less, every day           never    never    never
#    a fortnight off                       4       12    never
#    a month off then weekends             4       12       29
#    where a plan says never, it means not within 400 days
#    density reached after 400 days under each plan:
#    carry on, seven days          8.9091   (0.7955 of rested)
#    weekends off                  9.5484   (0.8525 of rested)
#    a third less, every day       9.3596   (0.8357 of rested)
#    a fortnight off               9.5484   (0.8525 of rested)
#    a month off then weekends     9.5484   (0.8525 of rested)
#    a month of complete rest alone reaches 0.9915 of rested
#    the two plans that get there are the ones that remove the
#    load rather than reduce it, which is the finding a reader
#    is least likely to want

从稳定循环在周日结束时的静息值 0.8598 出发(与 p09 同一个点),无论是继续撑着、取用周末,还是每天把负荷减三分之一,在四百天之内连 0.90 都到不了。两周休假在 4 天达到 0.90、12 天达到 0.95,而一个月休假在 29 天达到 0.99。行得通的两个方案,正是彻底移除负荷的那两个。

有一行值得再看一眼。每天把负荷减三分之一到达静息值的 0.8357,低于周末休息所到的 0.8525(两者都在第 400 天、一个周一结束时读取,那时每周的摆动正处于低点)。连续地减小负荷比在七天里移除两天更差,因为处置速率是皮质醇的饱和函数,所以待在低水平上的时间比均摊的等量削减更值钱。这与通常给一个无法请假的人的建议正好相反,而它是本章唯一一个有立即实用价值的结果:如果负荷无法移除,就应当成块地移除,而不是摊薄。

注

模型看不见的东西。 这些方案只按受体密度定价。两周的假期同时改变睡眠、酒精、运动、日照以及造成这份负荷的那件事;而每天少三分之一的工作,也许是一个人唯一的选项。算术说的是哪个方案把某一个变量推得更快;它没有说一个人应当选哪个方案,而它对这份假期是否可得、是否负担得起一无所知。

假说 · 尚未验证

作者的诠释(尚未验证): 因为处置速率会饱和,成块的彻底休息胜过均摊的等量负荷削减;所以每六周四天的休息,应当优于交付同等总缓解量的永久性减少一成工时。什么会证明它错: 在一项以受体或敏感度为终点、比较总缓解量相当的若干排程的试验里,如果均摊的削减与成块休息持平或更优,那么这个饱和论证就不能迁移。

小结

五天上班两天休息稳定在静息值的 0.8598,收回缺口的 0.3144 然后停住,而每周摆幅只有范围的 0.1410。从那里出发,十二天彻底休息越过 0.95,二十九天越过 0.99。每天把负荷减三分之一只到 0.8357,低于周末休息所到之处,因为处置速率会饱和,而成块的休息比均摊的摊薄更值钱。第六章转向唯一一个可以在数天之内移动的变量。

参考文献

Geurts, Sabine A. E., and Sabine Sonnentag. 2006. “Recovery as an Explanatory Mechanism in the Relation between Acute Stress Reactions and Chronic Health Impairment.” Scandinavian Journal of Work, Environment and Health 32 (6): 482 to 492.
Sonnenschein, Melanie, Marjolijn J. Sorbi, Lorenz J. P. van Doornen, Wilmar B. Schaufeli, and Cora J. M. Maas. 2007. “Evidence That Impaired Sleep Recovery May Complicate Burnout Improvement Independently of Depressive Mood.” Journal of Psychosomatic Research 62 (4): 487 to 494.

Fan, Y. (2026). HPA-Axis Coupled Oscillations, Cortisol Awakening Response Flattening and Burnout Recovery Kinetics: Glucocorticoid Receptor Downregulation, Ultradian Pulsatility and Allostatic Load Reset [下丘脑垂体肾上腺轴耦合振荡、皮质醇觉醒反应扁平化与倦怠恢复动力学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090252 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).