Population & Epidemiological Evidence

COVID lockdown natural experiment, electrification boundary, mobile phone paradox, 5G testicular evidence, ambient assessment, and the Kaiser Permanente series.

01COVID-19 lockdown validates three-channel prediction

Multiple studies report improved semen parameters during strict lockdown periods. A Chinese cohort (PubMed 41036143i) found sperm concentration and motility increased during home confinement. Zhangi et al. 2025 observed that semen quality declined again when restrictions were lifted, consistent with a reversible environmental component.

The three-channel model predicted this outcome: sperm quality improved while mental health declined because each frequency channel affects different tissues. The intermediate frequency (IF) channel (300 Hz – 1 MHz), which affects cell division through the same frequency-cell size relationship as FDA-approved TTFields cancer therapy, dropped dramatically during lockdown because office environments with hundreds of LED fixtures, HVAC variable frequency drives, and power electronics were eliminated. At environmental intensities (0.01–3 V/m), the IF mechanism operates via Ion Forced Oscillation (IFO-VGIC, threshold 10⁻⁵ V/m — Panagopoulos 2025 i), not via dielectrophoresis which requires TTFields-level intensity (100–300 V/m). A typical office floor has 200–500 LED fixtures versus 15–20 at home — a 10–25× difference in IF sources. The RF channel (> 1 MHz), which affects circadian and neuropsychological pathways, increased 30–50% because screen time and device usage rose. Two different frequencies, two different mechanisms, two different tissues, two different directions — exactly what the three-channel model predicts.

The sentinel species result confirms the outdoor component: COLOSS data shows bee colony winter loss increased by 2.27 percentage points during COVID (24/35 countries worsened, p = 0.043). BBS birds also declined 2.8–3.0% in 2020–22. Bees and birds remained in outdoor environments where ambient RF from cell towers continued uninterrupted, while human sperm quality benefited from reduced indoor IF exposure.

Citation	Year	Note
Chinese lockdown cohort (PubMed 41036143)i	2024	Sperm quality improvement during confinement
Zhang et al.i	2025	Quality decline after restriction lifting (reverse lockdown effect)
COLOSS winter loss paneli	2020–22	Counter-result: bees worsened (+2.27 pp, outdoor RF unchanged)
Optune TTFields (FDA PMA)i	2015	IF fields (100–300 kHz) disrupt cell division — same frequency as LED drivers

Three-Channel Model: COVID Lockdown Resolution

The three frequency channels changed in different directions during lockdown — resolving the apparent paradox

ELF (<300 Hz)
IF (300 Hz–1 MHz)
RF (>1 MHz)

IF↓ → less mitotic disruption → sperm↑

RF↑ → circadian + neuro disruption → mental health↓

Two different frequencies, two different mechanisms, two different tissues, two different directions — no paradox.

02The electrification boundary

The IFO-VGIC activation threshold (10⁻⁵ V/m, Panagopoulos 2025 i) is so low that every household appliance exceeds it at operating distance. This means the electrification boundary — whether a household has electricity or not — IS the biological exposure boundary. A person without electricity lives permanently below the IFO threshold. A person with electricity lives above it 24/7.

This is a binary threshold, not a gradient. Education, income, and contraception access are all gradients — more is gradually different from less. Only EMF exposure has a physical threshold that maps to an infrastructure boundary. This structural difference produces a testable prediction: adjusting national electricity consumption by the fraction of the population with access should improve TFR prediction. It does: correlation improves from r = −0.864 to r = −0.885 (54 countries, LOOCV full-model RMSE 0.522).

In partially electrified countries, the unelectrified population lives in the electromagnetic environment that evolution calibrated biological sensors for. Their TFR should be near the biological maximum (~6.5). National TFR is a mixture of the electrified (lower TFR) and unelectrified (higher TFR) populations. Correcting for this, Nigeria's electrified population (55%) has an estimated TFR of 4.03, Ethiopia's (51%) has 1.89, and Uganda's (42%) has 1.86 — dramatically lower than their national averages.

Citation	Year	Note
Panagopoulos et al. (Frontiers in Public Health)i	2025	IFO-VGIC threshold 10⁻⁵ V/m for polarized fields
Belmin et al. (Nature Sustainability)i	2022	155 DHS surveys: electricity access negatively affects fertility
DeLong et al. (PLOS ONE)i	2010	Energy consumption negatively related to population growth
BERM v17.1 formula discovery i	2026	54 countries, LOOCV full-model RMSE 0.522, R² 0.851

03The mobile phone paradox

Mobile phone subscriptions per 100 people — intuitively the most direct EMF proxy — is the WEAKEST single predictor of TFR across 54 countries (RMSE 1.053, worse than GDP alone at 0.719). Residential electricity consumption is the BEST (univariate RMSE 0.533 in exponential form).

This is paradoxical only if EMF exposure comes from phones. BERM's three-channel model resolves it: a mobile phone is one device in one channel (RF), used intermittently. Residential electricity measures the entire electromagnetic environment — every light, every appliance, every wire in every wall, the 50/60 Hz field that permeates the home 24 hours a day.

If the mechanism were 'information access' (TV, internet, phone), the mobile phone should be the BEST proxy because it IS the information device. The fact that it is the WORST proxy, while electricity consumption (which measures physical infrastructure, not information) is the BEST, is structurally consistent with EMF exposure rather than information access as the mechanism.

Citation	Year	Note
BERM v17.1 cross-sectional analysis i	2026	Mobile RMSE 1.053 vs electricity univariate RMSE 0.533 (54 countries)
World Bank Development Indicators i	2024	Source for residential electricity, broadband, mobile data

04CoQ10 and a defined 3.5 GHz rat protocol

Bektas 2026 i studied 28 rats in four groups with a GSM-modulated 3.5 GHz signal, 2 h/day for 30 days. CoQ10 attenuated some hormonal, testicular and redox changes. This was not a 5G NR waveform. Early calcium responses and repair time constants were not measured, so the result does not isolate a downstream repair site or demonstrate complete reversal of established damage.

Meyer 2026i found null viability/DNA outcomes in HaCaT cells at 50 Hz, 200 µT, while Haidar 2025i found null basal-ROS/DNA outcomes in skin cells with a 5G-modulated 3.5 GHz protocol. Their tissues, endpoints and waveforms differ from Bektas. A shared carrier frequency does not make GSM modulation equivalent to 5G NR. Matched protocols are needed to separate tissue dependence from protocol dependence.

Important: Lab baseline bias is symmetric. It does not only explain negative results — it also means that positive results underestimate the true effect size. When a study finds that RF exposure increases ROS by 30% compared to sham controls, the actual increase relative to a truly unexposed baseline may be larger, because sham controls are themselves partially exposed. This systematic underestimation affects all in vitro EMF research, not selectively.

Citation	Year	Note
Bektas et al. (Bioelectromagnetics)i	2026	GSM-modulated 3.5 GHz; CoQ10 attenuated some rat endpoints. Not a 5G NR protocol; the intervention site and recovery time remain unmeasured.
Meyer et al. (Bioelectromagnetics)i	2026	50 Hz, 200 µT, HaCaT keratinocytes: no viability change, no DNA damage. Blinded sham-controlled.
Haidar et al. (Scientific Reports)i	2025	5G-modulated 3.5 GHz, fibroblasts + keratinocytes: no ROS, no DNA damage (CPD). High-quality null.

055G ambient exposure assessment (Deprez i 2025)

Deprez i et al. (2025, Bioelectromagnetics, bem.70019) present spectral measurements of 5G RF-EMF exposure levels in four European countries. They can supply a raw ambient coordinate z_ambient, but the candidate v17 calculation must first declare a dimensionless normalization x=N(z_ambient) before using total = natural + ambient + χ_geo(x) × personal. Choosing and calibrating N is the open L0→L2 step; the measurements do not directly determine χ_geo.

Citation	Year	Note
Deprez et al. (Bioelectromagnetics)i	2025	5G spectral exposure assessment, 4 European countries. Beamforming changes ambient/personal ratio.

The Kaiser Permanente Series

The most comprehensive epidemiological evidence for EMF reproductive and developmental effects comes from Kaiser Permanente Northern California, where Dr. De-Kun Li's research group conducted six prospective cohort studies between 2002 and 2020, funded by the National Institute of Environmental Health Sciences (NIEHS).

All participating women wore an EMDEX meter for 24 hours during pregnancy, measuring ACTUAL magnetic field exposure from all sources — not self-reported device use. This objective exposure measurement distinguishes the Kaiser series from most other EMF epidemiology.

Findings across four BERM cascades

Miscarriage: 2× risk above 1.6 mG (Li 2002 i, 2017i)
ADHD: elevated risk with higher prenatal MF (Li 2020i, JAMA Network Open)
Childhood obesity: OR 1.69 (Li 2012 i, Scientific Reports)
Asthma: elevated risk (Li 2011 i)
Sperm quality: 2× risk of abnormal motility (Li 2010 i)

The EMDEX meter does not separate ELF from IF frequencies. High total MF correlates with both power grid exposure (ELF) and appliance/lighting exposure (IF). The Kaiser series cannot identify which channel drives the effect — but it demonstrates that MEASURED magnetic field exposure predicts reproductive and developmental outcomes across multiple endpoints.

Epistemic level: [E] (prospective cohort, measured exposure, NIEHS-funded). NOTE: Li 2017i Editorial Expression of Concern due to data sharing inability (not retraction, not about results).

Japan's IH Research Program

Japan has the world's most extensive research program on intermediate-frequency health effects, driven by the high penetration of induction heating cookers (~50% of households). Three studies examined IH use and pregnancy outcomes.

Sato et al. (2023)i found a statistically significant association between stationary/tabletop IH cooker use and preterm birth (OR 1.27–1.44). Critically, the association was DOSE-DEPENDENT: high-emission cookers (9 mT at 30 cm) showed higher risk than low-emission built-in models (4 mT). This dose-response pattern suggests a biological mechanism rather than a confounder.

Tokinobu et al. (2021)i found the opposite: IH use was associated with REDUCED preterm birth risk. The contradiction likely arises from exposure classification: Tokinobu grouped all IH types together, while Sato separated high-emission from low-emission models. When high and low are averaged, opposing effects may cancel.

Epistemic level: [C] (human epidemiology, contradictory but dose-response signal). These findings are preliminary but represent the only epidemiological evidence for IF-EMF effects on human pregnancy from a non-ELF, non-RF frequency range.

Retrodictions: What the modulome explains

Six biological outcomes provide retrospective tests of the twelve-layer modulome. Their proposed convergence is a shared-cause hypothesis: each outcome needs an explicit exposure measure, comparison group and competing explanations. The cards distinguish observations, proposed mechanisms and testable predictions.

These outcomes are not established as statistically independent. Shared data, selection and confounding must be assessed before combining evidence. No joint probability is assigned here, and the proposed EMF pathway requires direct testing.