5D Library › 双过程睡眠架构、慢波胶质淋巴清除与体温节律相位控制 · English

第五章 手暖,核心凉:睡眠的温度闸门

为什么是温度

核心体温在傍晚下降,在后半夜到达最低点。入睡往往发生在这条曲线的下降段上。这个下降主要不是靠产热减少造成的,而是靠散热增加;而身体散热的主要阀门,是流向手脚的血流。远端皮肤的小血管扩张时,来自核心的温血到达手脚皮肤,远端皮肤变暖,核心变凉。

Kräuchi 等人直接测量了这一点,发现远端与近端皮肤温度之差——远端-近端温度梯度(DPG)——是预测人们入睡快慢的最佳生理指标,比核心温度本身更好 (Kräuchi et al. 1999, 2000)。

一个有三个热库的模型

第九段程序把身体写成三个相互耦合的热库:核心、近端皮肤(躯干)与远端皮肤(手脚)。控制量是从核心到远端皮肤的传导率 kdk_d,远端血管扩张时它上升。数值是示意性的,但在生理量级上。

推导

方程。 设热容为 CC,传导率为 kk 与 hh,产热为 M(t)M(t),环境温度为 TaT_a: CcdTcdt=M−kp(Tc−Tp)−kd(t)(Tc−Td),C_c \frac{dT_c}{dt} = M – k_p(T_c – T_p) – k_d(t)(T_c – T_d), CpdTpdt=kp(Tc−Tp)−hp(Tp−Ta),C_p \frac{dT_p}{dt} = k_p(T_c – T_p) – h_p(T_p – T_a), CddTddt=kd(t)(Tc−Td)−hd(Td−Ta),C_d \frac{dT_d}{dt} = k_d(t)(T_c – T_d) – h_d(T_d – T_a), 且 DPG=Td−Tp\text{DPG} = T_d – T_p。

推导

# Chapter 5. Core, proximal skin and distal skin (hands and feet) as three
# coupled heat stores. Blood flow to the distal skin is the main control: when
# the hand and foot vessels dilate, the conductance k_d from core to distal
# skin rises, the distal skin warms, and the core loses heat. Heat production
# M falls in the evening. All values are illustrative but of physiological
# order (heat capacities in kJ/K, conductances in W/K, ambient 24 C under
# bedding). DPG = distal minus proximal skin temperature. One-minute steps;
# three days are run and the last is reported.
import math

CC, CP, CD = 160e3, 40e3, 8e3
KP, HP, HD, TA = 20.0, 6.0, 3.5, 24.0

def M(h):                                   # heat production, W
    return 88.0 + 3.0 * math.cos(2 * math.pi * (h - 15.0) / 24)

def kd(h, dilate_at=21.0, ramp=2.0):        # core -> distal conductance, W/K
    hh = h % 24
    low, high = 4.3, 10.0
    if dilate_at <= hh < dilate_at + ramp:
        return low + (high - low) * (hh - dilate_at) / ramp
    if hh >= dilate_at + ramp or hh < 6.5:
        return high
    if 6.5 <= hh < 7.5:
        return high - (high - low) * (hh - 6.5)
    return low

def run(dilate_at=21.0):
    Tc, Tp, Td = 37.0, 34.0, 30.0
    out = {}
    for m in range(3 * 24 * 60):
        h = m / 60.0
        k = kd(h, dilate_at)
        qp, qd = KP * (Tc - Tp), k * (Tc - Td)
        Tc += 60 * (M(h) - qp - qd) / CC
        Tp += 60 * (qp - HP * (Tp - TA)) / CP
        Td += 60 * (qd - HD * (Td - TA)) / CD
        if m >= 2 * 24 * 60:
            out[round(h - 48, 4)] = (Tc, Tp, Td)
    return out

def at(o, h):
    return o[round(h, 4)]

if __name__ == "__main__":
    base = run()
    print("A. A day with distal vasodilation beginning at 21:00")
    print("   time    core    proximal   distal    DPG")
    for h in (12, 18, 21, 22, 23, 24, 27, 29, 31):
        Tc, Tp, Td = at(base, h % 24)
        print(
            "   {:02d}:00  {:6.2f}   {:6.2f}    {:6.2f}   {:+5.2f}".format(
                h % 24, Tc, Tp, Td, Td - Tp
            )
        )
    cs = [v[0] for v in base.values()]
    print(
        "   core range over the day: {:.2f} C (max {:.2f}, min {:.2f})"
        .format(max(cs) - min(cs), max(cs), min(cs))
    )

    print(
        "\nB. When does the core start to fall? Vasodilation early, on time,"
        " or late"
    )
    print(
        "   dilation starts   core at 23:00   core change     core 0.20 C"
        " below"
    )
    print(
        "                                     23:00-24:00     its 21:00"
        " value at"
    )
    for d in (20.0, 21.0, 22.5):
        o = run(d)
        c21 = at(o, 21.0)[0]
        t_drop = next(
            h for h in sorted(o) if h >= 21.0 and o[h][0] <= c21 - 0.20
        )
        print(
            "   {:02d}:{:02d}             {:6.2f}          {:+6.3f}         "
            " {:02d}:{:02d}".format(
                int(d),
                int(round((d % 1) * 60)),
                at(o, 23.0)[0],
                at(o, 23 + 59 / 60)[0] - at(o, 23.0)[0],
                int(t_drop),
                int(round((t_drop % 1) * 60)) % 60,
            )
        )
    print(
        "\nReading: the core falls about 1 C overnight in this model, and the"
        " fall"
    )
    print(
        "begins when the hands and feet open up: distal vasodilation is the"
        " valve."
    )
    print(
        "Delaying it by 90 minutes delays the start of the core's decline by"
        " over an"
    )
    print("hour, with the core still near its evening peak at 23:00.")
# A. A day with distal vasodilation beginning at 21:00
#    time    core    proximal   distal    DPG
#    12:00   36.91    33.90     31.10   -2.80
#    18:00   37.41    34.30     31.39   -2.91
#    21:00   37.47    34.36     31.43   -2.94
#    22:00   37.36    34.32     32.72   -1.60
#    23:00   37.18    34.20     33.67   -0.53
#    00:00   37.02    34.07     33.67   -0.40
#    03:00   36.62    33.75     33.36   -0.38
#    05:00   36.43    33.59     33.22   -0.37
#    07:00   36.34    33.49     32.59   -0.90
#    core range over the day: 1.15 C (max 37.47, min 36.33)
#
# B. When does the core start to fall? Vasodilation early, on time, or late
#    dilation starts   core at 23:00   core change     core 0.20 C below
#                                      23:00-24:00     its 21:00 value at
#    20:00              37.05          -0.139          22:14
#    21:00              37.18          -0.159          22:32
#    22:30              37.41          -0.169          23:47
#
# Reading: the core falls about 1 C overnight in this model, and the fall
# begins when the hands and feet open up: distal vasodilation is the valve.
# Delaying it by 90 minutes delays the start of the core's decline by over an
# hour, with the core still near its evening peak at 23:00.

A 段重现了人体曲线的形状。21:00 时 DPG 为 −2.94 °C:手脚比躯干凉得多。远端血管一打开,DPG 到 23:00 升到 −0.53,而在 37.47 °C 达到峰值的核心,整夜下降到 36.33 °C——范围 1.15 °C,与人体昼夜体温的范围同一量级。其中大部分不需要产热的变化;阀门就做到了。

B 段问阀门早开或晚开会怎样。20:00 打开,核心在 22:14 降到比 21:00 低 0.20 °C;21:00 打开,在 22:32;晚 90 分钟、22:30 才打开,推迟到 23:47,而 23:00 时核心仍为 37.41 °C,接近傍晚的峰值。在这个模型里,「脚冷」不是比喻:它是一个关着的阀门,核心无法开始下降。

左:远端血管扩张分别在 21:00 与 22:30 开始时,一天中的核心体温。右:21:00 情形下的远端-近端皮肤温度梯度,随手脚打开而升向零。

先加热,为了降温

睡前洗个热水澡或淋浴,看上去是让核心降温的反方向。第十段程序把它放进同一个模型:向皮肤输送 20 分钟的热量,远端血管随之扩张。

推导

# Chapter 5. Passive body heating before bed, in the three-store model of p09.
# A warm bath is represented as 200 W delivered to the proximal skin for 20
# minutes, with the distal vessels fully dilated during it and for an hour
# after (the body's response to being heated). Lights out at 23:00. Compared:
# no bath, a bath ending 90 minutes before lights out, and one ending 15
# minutes before. Same parameters as p09.
import math

CC, CP, CD = 160e3, 40e3, 8e3
KP, HP, HD, TA = 20.0, 6.0, 3.5, 24.0

def M(h):
    return 88.0 + 3.0 * math.cos(2 * math.pi * (h - 15.0) / 24)

def kd(h, dilate_at=21.0, ramp=2.0):
    hh = h % 24
    low, high = 4.3, 10.0
    if dilate_at <= hh < dilate_at + ramp:
        return low + (high - low) * (hh - dilate_at) / ramp
    if hh >= dilate_at + ramp or hh < 6.5:
        return high
    if 6.5 <= hh < 7.5:
        return high - (high - low) * (hh - 6.5)
    return low

def run(bath_end=None):
    Tc, Tp, Td = 37.0, 34.0, 30.0
    out = {}
    for m in range(3 * 24 * 60):
        h = m / 60.0
        hh = h % 24
        k = kd(h)
        heat = 0.0
        if bath_end is not None and h >= 48:
            if bath_end - 20 / 60 <= hh < bath_end:
                heat = 200.0
            if bath_end - 20 / 60 <= hh < bath_end + 1.0:
                k = 10.0
        qp, qd = KP * (Tc - Tp), k * (Tc - Td)
        Tc += 60 * (M(h) - qp - qd) / CC
        Tp += 60 * (qp - HP * (Tp - TA) + heat) / CP
        Td += 60 * (qd - HD * (Td - TA)) / CD
        if m >= 2 * 24 * 60:
            out[round(h - 48, 4)] = (Tc, Tp, Td)
    return out

if __name__ == "__main__":
    print(
        "   bath         peak core (time)  core at 23:00  change        DPG"
        " at 23:00"
    )
    print("                                                 23:00-23:30")
    for label, end in (
        ("none", None),
        ("ends 21:30", 21.5),
        ("ends 22:45", 22.75),
    ):
        o = run(end)
        evening = {h: v for h, v in o.items() if 20.0 <= h <= 23.5}
        hp = max(evening, key=lambda h: evening[h][0])
        c23 = o[23.0]
        print(
            "   {:<12} {:6.2f} ({:02d}:{:02d})     {:6.2f}       {:+6.3f}    "
            "    {:+5.2f}".format(
                label,
                evening[hp][0],
                int(hp),
                int(round((hp % 1) * 60)) % 60,
                c23[0],
                o[23.5][0] - c23[0],
                c23[2] - c23[1],
            )
        )
    print("\nReading: a bath warms the core briefly, then"
          " the dilated hands and feet")
    print("dump the heat; ending it 90 minutes before"
          " lights-out leaves the core")
    print("falling faster at 23:00 than with no bath."
          " Ending it 15 minutes before")
    print("puts lights-out on the rising or peak part of the curve.")
#    bath         peak core (time)  core at 23:00  change        DPG at 23:00
#                                                  23:00-23:30
#    none          37.47 (20:43)      37.18       -0.084        -0.53
#    ends 21:30    38.03 (22:10)      37.90       -0.124        -0.62
#    ends 22:45    37.92 (23:25)      37.82       +0.099        -2.53
#
# Reading: a bath warms the core briefly, then the dilated hands and feet
# dump the heat; ending it 90 minutes before lights-out leaves the core
# falling faster at 23:00 than with no bath. Ending it 15 minutes before
# puts lights-out on the rising or peak part of the curve.

21:30 结束的热水澡使核心在 22:10 升到 38.03 °C;到 23:00 核心仍比不洗澡时更暖(37.90 对 37.18),但它下降得更快(之后半小时 −0.124 对 −0.084 °C),因为扩张的手脚正在把多余的热量排出去。22:45 才结束的热水澡则让熄灯落在曲线的上升段(23:00 之后半小时 +0.099 °C),DPG 为 −2.53,因为躯干皮肤还很热。

所以模型的预测关乎时机:如果加热有助于入睡,它是通过让傍晚的下降更陡而起作用的,而且只有在熄灯前足够早结束时才做得到。

证据

已知的情况。 一项关于睡前温水淋浴或泡澡研究的系统综述与荟萃分析发现,在就寝前 1-2 小时进行104-109 °F(约 40-43 °C)的水中被动加热,与更短的入睡潜伏期和更好的主观睡眠质量相关 (Haghayegh et al. 2019)。这里引用的是所综述研究的条件,为了准确,不是推荐;有心脏病、低血压或怀孕的人,泡热水澡前应先问医生。

假说 · 尚未验证

作者的假说(尚未验证): 在个体内部,熄灯后 30 分钟内核心体温的下降速率,比熄灯时的核心体温本身更能预测入睡潜伏期;被动加热只有在提高了这一速率时才缩短入睡潜伏期。什么结果会证明它错:熄灯时的核心体温预测入睡潜伏期同样好或更好,或者加热后的改善并不伴随更陡的下降。

小结

傍晚核心体温的下降主要是靠打开远端血管、而不是靠减少产热造成的,远端-近端皮肤温度梯度追踪着这个阀门;它是快速入睡的最佳生理预测指标。在三热库模型中,阀门打开后核心整夜下降约 1 °C;晚 90 分钟打开,下降的开始被推迟一个多小时。睡前加热可以让傍晚的下降更陡,但只有在熄灯前早早结束才行;临睡前才结束,会让熄灯落在曲线的上升段。

参考文献

Haghayegh, Shahab, Sepideh Khoshnevis, Michael H. Smolensky, Kenneth R. Diller, and Richard J. Castriotta. 2019. “Before-Bedtime Passive Body Heating by Warm Shower or Bath to Improve Sleep: A Systematic Review and Meta-Analysis.” Sleep Medicine Reviews 46: 124–35.
Kräuchi, Kurt, Christian Cajochen, Esther Werth, and Anna Wirz-Justice. 1999. “Warm Feet Promote the Rapid Onset of Sleep.” Nature 401 (6748): 36–37.
Kräuchi, Kurt, Christian Cajochen, Esther Werth, and Anna Wirz-Justice. 2000. “Functional Link between Distal Vasodilation and Sleep-Onset Latency?” American Journal of Physiology – Regulatory, Integrative and Comparative Physiology 278 (3): R741–48.

Fan, Y. (2026). Two-Process Sleep Architecture, Slow-Wave Glymphatic Clearance and Thermoregulatory Phase Control: Adenosine Homeostasis, Flip-Flop Switch Bistability, Core-Distal Heat Flux and Sleep-Stage Markov Chains [双过程睡眠架构、慢波胶质淋巴清除与体温节律相位控制] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090162 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).