Methodology · Efficiency · BASELINE

Method effectiveness as a measurable process

Why we don't measure results by group and session averages—and how the NIHILO platform turns subjective practice into repeatable, testable design.

ex NIHILO · Konstantin Shel

“The program works” is not the impression of the participant and not a beautiful number at the end of the report. This is a question of measurement: what exactly has moved, relative to what, in how many people, at what speed and whether this is explained by chance. Below is the methodological the framework on which the platform places efficiency, without the promises made can't stand it.

Where does the illusion of effect come from?

Most “evidence” of practices is based on two pitfalls that look like convincing, but methodologically insignificant:

Group average. Averaging masks: it also hides those who have become worse, and those for whom nothing has changed, and gives “significance” in a sample where half no dynamics participants. This is not evidence for a statement about the effectiveness of the method.
One before and after session. A shift in one measurement does not differentiate the effect practices from the novelty effect, regression to the mean after a bad period, or simply background variability of the day.

Both traps are the result of one mistake: attempts to measure individual change group and instant tools. The platform solves this differently.

Design: N-of-1, not RCT

We use single-case experimental design (N-of-1, within-subject): each participant is his own control. Instead of comparing "group A versus group B" we compare a person with him before and after intervention on a continuous longitudinal basis. This is not “weaker than RCT” - for personalized practice it is the only honest one design:

RCT assumes sample homogeneity and an average effect, which in practice there is no self-regulation - everyone has their own base level, their own speed and their own response profile;
N-of-1 gives an estimate of the effect for a specific person, not for "average" patient" who does not exist;
the longitudinal format decouples causality: the shift is tied to a specific to an event on the time axis, and not to the “program in general.”

What does it give

Efficiency ceases to be a probability and becomes trajectory: we see not “it worked / didn’t work”, but “this person’s Relaxation Index has reached a plateau 9 days after the start, stayed for 3 weeks and did not return to the original after the pause.”

Multidimensional triangulation: three independent channels

Efficiency is not measured by one device - a monomethodological approach in itself source of bias. The platform triangulates three independent streams, and only their results are considered coincidence in one time window:

Three channels

Brain (EEG NEIRY). Normalized rhythms and indices: alpha (increases with immersion), beta (falls - internal commentator subsides), theta (drowsiness marker - when deepening is silent), SMR (body contact), Relaxation Index (Alpha/Beta, grows 2–3 times upon entry), Concentration Index (falls - attention ceases to be a directed beam). Channel shows depth and relaxation, but not the address of attention.
Body (fitness tracker, HRV according to Baevsky). RMSSD, voltage index, respiratory sinus arrhythmia, perfusion index. Vegetative background on which it rests state: when contact is established with the organ, the indicators reach the upper level and are held (peak-and-plateau, not peak-and-back).
Phenomenology (audio-self-report). Description of what Where gone attention and what was lived. The only channel that directly distinguishes between "dissolved" in the void" (MPE) and "kept in contact with the body" (V-MPE) - something that the device does not see.

Three streams are combined into a composite index MNPI (Multidimensional Neuroplasticity / self-regulation Progress Index). Phenomenology enters into it as separate subindex PI (Phenomenological Index) - structured LLM-assessment of self-reports on a fixed 5-criteria scale; brain and body - through subindices RI (regulation), DI (dysregulation), SI (stress). The index does not “average” the channels, but requires their consistency.

Shift significance: Reliable Change Index

The shift itself does not mean anything - you need to separate the effect from the “weather”. For this platform uses Reliable Change Index (Jacobson & Truax, 1991): a shift is considered significant only if it exceeds human variability (his personal spread between days), normalized to the reliability of the measurement.

RC = (X₂ − X₁) / S_diff, where S_diff = √(2·SE²), SE = SD·√(1−r_xx)

Where X₁ is the base value, X₂ is the value after, SD is the standard deviation of the base period, r_xx — metric reliability. Threshold |RC| ≥ 1.96 (p < 0.05 means, What change goes beyond the random fluctuation of that particular person. We don't count "growth percentages" - they are meaningless without variance. We We consider whether the shift exceeds personal variation.

Convergent Validation: MNPI Does Not Replace Psychometrics

A composite index in itself is not proof - it is a working construct. Therefore MNPI tested for convergent validity (Campbell & Fiske, multimethod matrix) against a battery of 12 already validated psychometric instruments: PHQ-9, GAD-7, PSS-10, ISI, ASRS-v1.1, MAAS, MEQ, RRS, ATQ-30, MCTQ, FAS, CFQ-7. Each of the 4 MNPI sub-indices is mapped to each of the 12 instruments (4 × 12 matrix) with pre-specified effect size criteria and correction for multiple comparisons.

Honest frame

MNPI doesn't pretend on the status of a neurobiological indicator, and the connection between its longitudinal shift (ΔMNPI) and neuroplasticity is working hypothesis, which the program checks, not an established fact. In case of discrepancy psychometric instruments remain the reference standard. We do not customize data under the index - we check whether the index is holding.

Controlling threats to validity

Efficiency is assessed honestly only when the threats that affect it are taken into account. imitate. The platform incorporates them into the design:

The effect of novelty. The base period (2–3 weeks of measurements BEFORE the start) catches a surge of motivation before he gets to the comparison base.
Regression to the mean. People come in decline - and return to their to the average it looks like progress. The base period and RCIndex decouple this: the shift is considered only relative to the spread, not relative to the worst point.
Background life. Timeline of interventions (start, intensive, illness, vacation, sleep change) is placed on the same axes as the data - to distinguish the effect of practice from external event.
Attrition. Who stopped walking is part of the result, not data loss (analogous to intent-to-treat): a program where half has fallen off still works worse than it looks from the rest.

Longitudinal windows and speed as a metric

MNPI is constructed as a longitudinal index: each day is a point on the same time axis. The analysis is carried out on three practically meaningful horizons - 7 days, 40 days and season - as cycles of real changes in self-regulation, and not formal ones statistical windows. By the difference between today's and baseline values (ΔMNPI) we see not only What the method does to the brain and body, but also at what speed: how many weeks did it take for the Relaxation Index to reach a plateau, when concentration collapsed, how quickly the night vegetative flow was restored. The interval between intervention and MNPI shift is measured speed perestroika, which distinguishes a working method from random enlightenment.

Honest publication of zero results

The absence of an effect is also a result, and for confidence in the method it is more important than another beautiful numbers. The platform records and publishes what didn't move: “for N participants out of M, the shift did not exceed the personal spread” - this is not a failure of the program, but a correct conclusion on which an honest iteration of the method is based.

Platform boundaries - and why this is not a weakness

Expertise includes knowing the limits of your tools. NEIRY wearable EEG has frontal location of electrodes: no central leads, no ECG for individual beats. That's why direct interoception - response of the cortex to heartbeats (HEP, heartbeat-evoked potentials) and rhythms over the sensorimotor cortex - requires laboratory equipment that the hoop does not have. We do not hide it: the device confirms depth and stability states, and the address of attention (body vs emptiness) closes phenomenology, not a raw signal. This is not a hole in the method - it is correct triangulation where one channel physically cannot respond.

Bottom line: what we consider proven

The effectiveness of the method on the NIHILO platform is considered established when all conditions at the same time:

Base period fixed before intervention (Point A).
The shift is significant according to the Reliable Change Index regarding personal dispersion, rather than the group average.
Three channels matched in one time window (brain + body + phenomenology).
Speed measured: The interval between intervention and ΔMNPI is fixed.
Retention checked after a break without a leader.
Convergent Validation MNPI is consistent with psychometric battery (or the discrepancy is clearly recorded).

We do not consider one of the conditions, in isolation from the rest, as evidence. Exactly this one the framework—not the belief of the participant—makes the result repeatable and testable.

The effectiveness of the method is visible not in one curve, but in the coincidence three independent channels - brain, body and phenomenology - combined in MNPI and verified on the significance of the shift and convergent validity. One channel in isolation from others, we do not We consider it evidence.

Participants’ data is anonymized, only aggregated indicators are published. Graphs and tables are collected from the original downloads automatically.

ex NIHILO · research program on measurable conditions consciousness · BASELINE - hub