5D Library › Finger–Brain Cortical Magnification, Distal Microneedle Afference and Computational Neurology of Post-Stroke Cognitive Network Reconstruction · 中文
Chapter 8. Practice and a Pre-registered Test
Note
The checklist and trial protocol in this chapter are for research and for reading the literature; they are not medical advice. The checklist cannot tell you whether to use press needles; it can only help you ask the right questions when you read a claim of efficacy. The trial protocol is a draft design for independent research institutions; it must be reviewed by ethics committees, registered on a clinical trial registry and carried out by qualified clinical teams, and no one should use it to needle stroke patients on their own.
Part 1. A four-step checklist
Chapters 1 to 6 can be condensed into a four-step checklist for reading any study claiming that a peripheral stimulus improves cognition after stroke.
- Comparison. Is there a sham intervention that looks and is handled like the real one? Does the main conclusion come from a between-group or a within-group before–after comparison? What are the p value and effect size of the main comparison (Chapter 6)?
- Outcome. Is the primary outcome a single outcome fixed in advance? Does it have a ceiling? Is the analysis adjusted for baseline? Are the numbers in the tables consistent, and do counts and percentages agree (Chapter 6)?
- Mechanism. Does the mechanistic evidence come from humans, animals or models? Is the “network” a physiological network or a co-occurrence network of prescriptions or papers? Is a “proof” inside a model being taken as an empirical conclusion (Chapters 1–3)?
- EEG. How many people? Is there a control? Are eyes open and eyes closed kept apart? Were measures and bands fixed in advance? Are group results shown, or single participants (Chapters 4 and 5)?
Applied to Lu Shunkang’s thesis, the checklist gives the table below. It is not an overall assessment of the thesis but a statement of how far its evidence reaches on these four points.
| Step | In the thesis | What it can support |
|---|---|---|
| Comparison | Open-label add-on design, no sham; MoCA response rates p = 0.287, post-treatment scores P = 0.005 (unadjusted for baseline) | The between-group results reported are mutually inconsistent; with open labels and inconsistent tables the extra effect of press needles cannot be determined; within-group improvement cannot be attributed to press needles |
| Outcome | Main measure MoCA analysed as “response rate”; formula in the wrong direction; counts in the response table inconsistent; most “medians” cannot be medians | Between-group differences in secondary measures need rechecking against raw data before interpretation |
| Mechanism | The network is a prescription co-occurrence network; the molecular mechanism comes from a study of brain-injured rats | Describes prescription habits and possible research directions; not evidence of a mechanism of efficacy |
| EEG | Six people, before–after, no control, one participant’s spectrum, no statistics | Feasibility information for a pilot; “alpha restoration” untested |
Part 2. A pre-registered three-arm randomised controlled trial
Question and design choices
In patients with post-stroke mild cognitive impairment, on top of usual care, does rubbing genuine press needles at the twelve well points of both hands three times a day, compared with sham press needles that look and are rubbed exactly the same but do not penetrate the skin, give a higher MoCA total at week 4, by at least the smallest effect of interest?
The author’s framework proposes a three-arm design: genuine press needles, validated blunt non-penetrating sham press needles, and standard care, with both the genuine and the sham groups pressing three times a day to separate the effect of “the act of pressing itself”. The book keeps the three arms and makes four design choices. First, a single primary comparison: genuine against sham press needles. This is the only comparison that can answer whether the needle itself does anything; Chapter 2 showed that rubbing supplies the most reliable afference in the protocol, and it is the same in both groups. Second, a single primary outcome: the MoCA total at week 4, analysed by analysis of covariance with baseline MoCA and centre as covariates. The MoCA is the thesis’s own main scale, its responsiveness has been studied after stroke (Wu et al. 2019), and Chapter 6 shows that analysis of covariance keeps most of its power despite the ceiling, with a Tobit model as sensitivity analysis; computerised timed tests have no ceiling but large practice effects, version and device differences and no accepted clinically important difference after stroke, so they are secondary. Third, multiple comparisons by fixed-sequence testing: the primary comparison is tested at two-sided α = 0.05; only if it is significant is genuine against standard care tested at the same α; sham against standard care is only estimated descriptively (it estimates the combined effect of rubbing, attention and expectation). Fourth, independent conduct: the protocol, analysis code and decision rule are registered publicly before the first participant is enrolled, and the author takes no part in recruitment, intervention, assessment or analysis.
Protocol
| Element | Specification |
|---|---|
| Conduct and registration | Run by a multicentre clinical research team unconnected with the author; approved by the ethics committee of each centre; registered with the Chinese Clinical Trial Registry (ChiCTR) or an equivalent platform; written and reported following SPIRIT/CONSORT and the STRICTA guidelines for acupuncture trials (Schulz et al. 2010; MacPherson et al. 2010); no recruitment from the author’s community, students or customers |
| Participants | Age 40–80; 2 weeks to 6 months after a first ischaemic stroke; MoCA 18–25 (one point added for ≤ 12 years of education, as in the original and Beijing versions); able to press with the fingers unaided; excluded: dementia, severe aphasia, bleeding disorders or current anticoagulant use (as judged by the centre physician), broken or infected skin on the hands, allergy to the needles or tape, well-point acupuncture or press needles in the past 3 months |
| Randomisation | Central web-based system, 1:1:1, stratified by centre and baseline MoCA (18–21, 22–25), with variable, undisclosed block sizes; analyses adjust for the stratification factors (Kahan and Morris 2012) |
| Interventions | Genuine: Φ0.18 mm × 0.9 mm press needles at the twelve well points of both hands; sham: a validated placebo device with the same appearance, packaging and tape but no needle element; both groups rub at a fixed time of day 3 times a day, 2–3 minutes per point, with devices replaced every 24 hours by a research nurse; 5 days a week for 4 weeks. Standard-care group: usual care only. Usual drugs and rehabilitation in all groups follow each centre’s practice, unrestricted but recorded item by item |
| Blinding | Participants (genuine and sham groups), outcome assessors and statisticians are blinded; research nurses applying the devices cannot be fully blinded, are kept apart from assessors and follow a script with participants; participants in the standard-care group are not blinded |
| Pilot | Before the main trial, blinding of the sham press needle is tested in 30 stroke patients (Bang index (Bang et al. 2004)); if the index in the genuine group exceeds 0.2, the device is modified and the pilot repeated |
| Primary outcome | MoCA total at week 4 (an alternate form different from baseline, to reduce practice effects); analysis of covariance with baseline MoCA and centre as covariates; primary comparison: genuine against sham press needles |
| Sensitivity analyses | Tobit model treating 30 as censored; per-protocol analysis; multiple imputation under missing at random, plus scenario analyses departing from it |
| Secondary outcomes | MoCA at week 8; MMSE (Folstein et al. 1975); trail-making, digit-span, Stroop colour-word and symbol–digit tests; activities of daily living; adverse events; Bang blinding index; eyes-closed frontal alpha relative power at weeks 2 and 4 (the plan of Chapter 5). All descriptive or exploratory |
| Mediation analysis | Pre-specified secondary analysis: whether eyes-closed frontal 8–13 Hz relative power at week 2 mediates the difference between genuine and sham press needles in MoCA at week 4 |
Assumptions needed for mediation analysis
Randomisation ensures unbiased estimates of the effects “allocation → outcome” and “allocation → mediator”, but not that the relation “mediator → outcome” is unconfounded. A causal interpretation of a mediated effect needs at least four assumptions (Imai et al. 2010; VanderWeele 2015): no unmeasured confounding between mediator and outcome (vigilance, sleep or depression, for example, may affect both alpha power and the MoCA); no mediator–outcome confounder that is itself affected by allocation; the mediator precedes the outcome in time (hence alpha at week 2 and MoCA at week 4 in this protocol); and any interaction between allocation and mediator in their effect on the outcome is handled in the model. These assumptions cannot be fully tested from the data, so conclusions of the mediation analysis can only be stated “under these assumptions”, with a sensitivity analysis for unmeasured confounding. Chapter 4 showed that alpha power depends strongly on eyes open or closed, arousal and drowsiness; even a “significant” mediated effect cannot by itself prove that “alpha restoration is the route by which press needles act”.
Smallest effect of interest and sample size
The smallest effect of interest is set at 1.5 MoCA points. The reasons: the thesis reports no standard deviations; the earlier study it cites reported post-treatment MoCA standard deviations between 2.88 and 4.83 (section 2.2.1.2 of the thesis), and this protocol takes 3.0 as a planning assumption; 1.5 points is half this standard deviation, the commonly used “half a standard deviation” threshold (Norman et al. 2003); a smaller difference would not justify changing practice for a dozen minutes of daily rubbing plus the cost of needles. The correlation between baseline and week 4 is taken as 0.5. All of these are planning assumptions to be checked in the pilot.
Calculation
# Chapter 8. Sample size, genuine vs sham press needles (both pressed 3 times a day).
# Primary outcome: MoCA at week 4 by ANCOVA (baseline, centre). Planning assumptions:
# SD 3.0, baseline-outcome correlation 0.5, SESOI 1.5 points, two-sided 0.05, power 90 %.
# n per arm = 2 (z_0.975 + z_0.90)^2 s^2 (1 - r^2) / d^2 + z_0.975^2 / 4, then checked by
# simulation with integer scores, the 18-25 inclusion window and the ceiling at 30.
import math, random, statistics
Z = statistics.NormalDist().inv_cdf
SD, RHO, D = 3.0, 0.5, 1.5
def n_formula(sd=SD, rho=RHO, d=D, power=0.90):
za, zb = Z(0.975), Z(power)
return 2 * (za + zb) ** 2 * sd ** 2 * (1 - rho ** 2) / d ** 2 + za ** 2 / 4
def one_trial(rng, n, delta, mean_post=25.0):
rows = []
for arm in (1, 0):
for _ in range(n):
while True:
b = rng.gauss(21.5, 3.0); bo = round(b)
if 18 <= bo <= 25:
break
e = rng.gauss(0, SD * math.sqrt(1 - RHO ** 2))
y = mean_post + delta * arm + RHO * (b - 21.5) + e
rows.append((arm, bo, min(30, round(y))))
n2 = len(rows); mx = math.fsum(r[1] for r in rows) / n2
# ANCOVA via residualising on baseline within the pooled sample
X = [(1.0, r[0], r[1] - mx) for r in rows]; Y = [r[2] for r in rows]
S = [[math.fsum(x[i] * x[j] for x in X) for j in range(3)] for i in range(3)]
v = [math.fsum(x[i] * y for x, y in zip(X, Y)) for i in range(3)]
inv = invert3(S)
beta = [math.fsum(inv[i][j] * v[j] for j in range(3)) for i in range(3)]
res = [y - math.fsum(bb * xx for bb, xx in zip(beta, x)) for x, y in zip(X, Y)]
s2 = math.fsum(r * r for r in res) / (n2 - 3)
z, df = Z(0.975), n2 - 3
tcrit = z + (z ** 3 + z) / (4 * df) # t quantile, Cornish-Fisher
return abs(beta[1] / math.sqrt(s2 * inv[1][1])) > tcrit, \
sum(1 for r in rows if r[0] == 1 and r[2] == 30) / n
def invert3(S):
a, b, c = S[0]; d, e, f = S[1]; g, h, i = S[2]
det = a * (e * i - f * h) - b * (d * i - f * g) + c * (d * h - e * g)
return [[(e * i - f * h) / det, (c * h - b * i) / det, (b * f - c * e) / det],
[(f * g - d * i) / det, (a * i - c * g) / det, (c * d - a * f) / det],
[(d * h - e * g) / det, (b * g - a * h) / det, (a * e - b * d) / det]]
if __name__ == "__main__":
n0 = n_formula()
print(f"normal theory: n = {n0:.3f} -> {math.ceil(n0)} per arm"
f" (residual SD {SD * math.sqrt(1 - RHO ** 2):.2f})")
rng = random.Random(2026); chosen = None
for n in (math.ceil(n0), 68, 72, 76):
res = [one_trial(rng, n, D) for _ in range(2000)]
pw = sum(r[0] for r in res) / len(res); se = math.sqrt(pw * (1 - pw) / 2000)
top = math.fsum(r[1] for r in res) / len(res)
print(f" simulated power, n = {n}/arm: {pw:.3f} (s.e. {se:.3f});"
f" treated arm at 30: {top:.0%}")
if chosen is None and pw - 2 * se >= 0.90:
chosen = n
t1 = sum(one_trial(rng, chosen, 0.0)[0] for _ in range(2000)) / 2000
print(f" smallest n with power - 2 s.e. >= 0.90: {chosen}; type I error there"
f" {t1:.3f}")
plan = math.ceil(chosen / 0.85)
print(f"planned: {chosen} analysable per arm; 15 % loss -> {plan} randomised"
f" per arm,\n {3 * plan} in three arms (1:1:1)")
print("sensitivity, analysable per arm (normal theory):")
print(" SESOI \\ rho 0.3 0.5 0.7")
for d in (1.0, 1.5, 2.0):
print(f" {d:4.1f} points " + "".join(
f"{math.ceil(n_formula(rho=r, d=d)):7d}" for r in (0.3, 0.5, 0.7)))
# normal theory: n = 64.005 -> 65 per arm (residual SD 2.60)
# simulated power, n = 65/arm: 0.894 (s.e. 0.007); treated arm at 30: 14%
# simulated power, n = 68/arm: 0.898 (s.e. 0.007); treated arm at 30: 14%
# simulated power, n = 72/arm: 0.919 (s.e. 0.006); treated arm at 30: 14%
# simulated power, n = 76/arm: 0.930 (s.e. 0.006); treated arm at 30: 14%
# smallest n with power - 2 s.e. >= 0.90: 72; type I error there 0.057
# planned: 72 analysable per arm; 15 % loss -> 85 randomised per arm,
# 255 in three arms (1:1:1)
# sensitivity, analysable per arm (normal theory):
# SESOI \ rho 0.3 0.5 0.7
# 1.0 points 174 143 98
# 1.5 points 78 65 44
# 2.0 points 44 37 26The normal-theory formula gives 64.005 per group, rounded up to 65 analysable. The program then simulates the thesis’s inclusion range (baseline 18–25), integer scores and the ceiling of 30: power is 0.894 with 65 per group, 0.898 with 68 and 0.919 with 72; by the rule “power minus two standard errors at least 0.90”, 72 analysable per group are taken, where the type I error rate without an effect is 0.057 (simulation standard error about 0.005). Allowing 15 % for loss to follow-up and non-assessable outcomes, 85 are randomised per group, 255 in three groups. The sensitivity table shows that with a smallest effect of 1.0 point, 143 per group would be needed; with a baseline–outcome correlation of only 0.3, an effect of 1.5 points needs 78.
As shown in Chapter 6 of Book 33, “90 % power” means that when the true effect equals exactly 1.5 points, the probability of a conclusion of the kind “the genuine needle is better” (lower confidence limit above 0) is about 90 %; in that situation the strongest conclusion, “better by at least 1.5 points”, has only about a 2.5 % chance.
Decision rule
Calculation
# Decision rule, fixed before unblinding. Primary comparison: genuine versus sham press
# needle. D = adjusted difference in week-4 MoCA (ANCOVA: baseline MoCA, centre), with
# its 95 % CI. SESOI 1.5 points. Gates, checked first: more than 20 % of randomised
# participants without a primary outcome; allocation revealed to outcome assessors
# before database lock. Pre-specified qualifier: a Bang blinding index above 0.2 in
# the genuine arm (participants guessing correctly beyond chance). A serious-adverse-
# event excess in the genuine arm is reported first, whatever the efficacy result.
SESOI = 1.5
def verdict(d, lo, hi, missing=0.08, bang=0.05, revealed=False, sae_excess=False):
pre = "SAFETY SIGNAL REPORTED FIRST; " if sae_excess else ""
for bad, why in ((missing > 0.20, f"{missing:.0%} without primary outcome"
" (limit 20 %)"),
(revealed, "allocation revealed to assessors before lock")):
if bad:
return f"{pre}D = {d:+.2f}: NOT INTERPRETABLE - {why}"
if lo >= SESOI: v = "benefit of at least the SESOI"
elif lo > 0: v = "benefit, but not shown to reach the SESOI"
elif hi < 0: v = "sham better than genuine"
elif lo >= -SESOI and hi <= SESOI: v = "no difference worth acting on"
else: v = "unclear: the interval is too wide to decide"
if bang > 0.2:
v += f"\n qualifier: Bang index {bang:.2f}, participants guessed their"
v += "\n allocation; expectation may explain part of D"
return f"{pre}D = {d:+.2f} ({lo:+.2f} to {hi:+.2f}): {v}"
for case in ((2.4, 1.6, 3.2), (1.1, 0.2, 2.0), (0.3, -0.6, 1.2), (-0.9, -1.8, -0.1),
(0.5, -1.7, 2.7), (2.0, 1.0, 3.0, 0.08, 0.35), (1.0, 0.1, 1.9, 0.26),
(0.2, -0.7, 1.1, 0.05, 0.0, False, True)):
print(verdict(*case))
print("secondary outcomes (MMSE, timed tests, frontal 10-Hz alpha power, the comparisons")
print(" with usual care) are reported with intervals; they never change the verdict")
# D = +2.40 (+1.60 to +3.20): benefit of at least the SESOI
# D = +1.10 (+0.20 to +2.00): benefit, but not shown to reach the SESOI
# D = +0.30 (-0.60 to +1.20): no difference worth acting on
# D = -0.90 (-1.80 to -0.10): sham better than genuine
# D = +0.50 (-1.70 to +2.70): unclear: the interval is too wide to decide
# D = +2.00 (+1.00 to +3.00): benefit, but not shown to reach the SESOI
# qualifier: Bang index 0.35, participants guessed their
# allocation; expectation may explain part of D
# D = +1.00: NOT INTERPRETABLE - 26% without primary outcome (limit 20 %)
# SAFETY SIGNAL REPORTED FIRST; D = +0.20 (-0.70 to +1.10): no difference worth acting on
# secondary outcomes (MMSE, timed tests, frontal 10-Hz alpha power, the comparisons
# with usual care) are reported with intervals; they never change the verdictThe example results are invented. If any precondition fails (more than 20 % of randomised participants without a primary outcome, or allocation revealed to assessors before the results are locked), the report says “not interpretable” whatever the primary outcome. If serious adverse events are more frequent in the genuine group than in the comparison group, the safety signal is reported first. Only if the lower confidence limit is at least 1.5 points does the report say “benefit of at least the smallest effect of interest”; a lower limit above 0 but below 1.5 points gives “benefit, but not shown to reach the smallest effect of interest”; an upper limit below 0 gives “sham better”; an interval within ±1.5 points gives “no difference worth acting on”; anything else gives “unclear”. A Bang index above 0.2 adds a qualifier: participants guessed their allocation, and expectation may explain part of the difference.
What this trial cannot show
First, it tests one device, one set of points, one rubbing frequency and a 4-week course, and cannot be generalised to bloodletting, electroacupuncture or other points. Second, the absence of penetration may change how participants rub, which a blinding test can only partly exclude. Third, it does not test the controllability, cortical magnification, stochastic-resonance or plasticity mechanisms of Chapters 1–3. Fourth, the frontal-alpha mediation analysis depends on assumptions that cannot be fully tested. Fifth, the conclusions do not extend to severe cognitive impairment, aphasia or acute-phase patients. Sixth, outcomes at 4 weeks and follow-up at 8 weeks cannot answer questions about long-term effects. Seventh, if the result is “no difference worth acting on”, that is the conclusion of the trial, and it cannot be replaced by secondary outcomes, subgroups or EEG measures.
Ethics and conflicts of interest
Press needles are medical devices; application, replacement and the handling of adverse events are the responsibility of qualified research nurses and physicians; an independent data and safety monitoring board is set up; participants may withdraw at any time without affecting their usual care. Whatever the result, the protocol, statistical analysis plan, analysis code and aggregate results will be published, with de-identified individual data under a data-sharing statement; the author takes no part in recruitment, intervention, assessment or analysis.
Disclosure
The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and therefore has a commercial interest in the subject of this book. The book promotes no course, card deck or service; the Chapter 8 trial is designed to be run and analysed by researchers independent of the author’s business and does not recruit from the author’s community, students or customers; all the code and the study protocol are published openly for anyone to check. Chapters 1 to 6 of this book draw extensively on the master’s thesis that Lu Shunkang made public in 2024; this book was written independently by the 5D Library author and has not been reviewed or endorsed by the thesis author, the thesis supervisors or Tianjin University of Traditional Chinese Medicine, and the author of this book is responsible for the re-analysis of the thesis data and for the consistency checks reported here. The author does not sell needles or medical services.
Chapter summary
The four-step checklist reads studies of peripheral stimulation for cognition through comparison, outcome, mechanism and EEG; applied to Lu Shunkang’s thesis, it shows that the between-group results reported are mutually inconsistent and the extra effect of press needles cannot be determined, that secondary results need rechecking against raw data, and that the mechanistic and EEG evidence only suggests directions for research. The pre-registered three-arm trial has genuine against sham press needles as its single primary comparison (both groups rub three times a day) and analysis of covariance of the MoCA total at week 4 as its single primary outcome, with a smallest effect of interest of 1.5 points, 72 analysable and 85 randomised per group, 255 in three groups, and a decision rule written as code; frontal alpha is only a pre-specified secondary mediation analysis, with its assumptions stated. The trial is run by independent institutions and published whatever the result.
References
Fan, Y. (2026). Finger–Brain Cortical Magnification, Distal Microneedle Afference and Computational Neurology of Post-Stroke Cognitive Network Reconstruction: From Complex-Network Pinning Control and Thalamocortical 10 Hz Alpha Restoration to Drift-Diffusion Models, Read Beside the Twelve Jing-Well Points of the Lingshu [指—脑皮层放大、远端微针传入与卒中后认知网络重建的计算神经学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23839915 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).