Humans Are Not Dosimeters

A Genotype- and State-Dependent Bioelectric Response Framework for Electromagnetic Hypersensitivity

Manuscript draft: Perspective / Hypothesis and Testable Research Framework

Author(s): RF Safe   |   Version: 1.0

Abstract

Electromagnetic hypersensitivity (EHS) is commonly evaluated through acute provocation studies that ask whether self-identified sensitive individuals can consciously detect radiofrequency electromagnetic-field (RF-EMF) exposure or develop immediate symptoms during short active-versus-sham challenges. Those studies are important, and the largely negative results should be taken seriously. However, this paradigm answers a narrow question: whether humans can function as conscious EMF detectors under acute laboratory conditions. It does not exhaust the biologically more relevant question of whether defined EMF waveforms can produce delayed, state-dependent, genotype-dependent changes in objective physiological timing systems in a susceptible subgroup. Human physiology is not a dosimeter; it is a dynamic, genetically diverse, state-dependent control system.

This paper proposes a falsifiable bioelectric response framework for EHS. The framework treats EHS not as a single proven disease entity and not as a mere exposure-detection claim, but as a heterogeneous clinical-research domain in which a definable subset may show reproducible active-minus-sham changes in sleep EEG, autonomic regulation, calcium signaling, mitochondrial redox dynamics, or symptom-physiology coupling after defined RF/ELF exposure. The human proof-of-principle anchor is the 2025 randomized, double-blind, sham-controlled NeuroImage study showing that 3.6 GHz 5G RF-EMF modulated NREM sleep-spindle center frequency in a CACNA1C rs7304986 genotype-dependent manner. That study does not prove EHS, conscious EMF detection, harm, or clinical causation. It shows something narrower and more important for experimental design: CACNA1C genotype x defined RF exposure can produce an objective sleep-EEG difference.

We integrate this evidence with voltage-gated calcium-channel structure, S4 voltage-sensor biology, calcium oscillation frequency decoding, mitochondrial-redox coupling, the 2026 Cell report of a CYB5B-linked EMF-inducible gene switch, and channel-dependent amplitude-modulated RF-EMF effects in hepatocellular carcinoma models. These lines of evidence are not presented as proof of EHS. They are presented as mechanistic plausibility anchors for a rigorous research program. We outline a staged protocol combining 30-90 days of naturalistic dosimetry and physiology, whole-genome sequencing, blinded sleep-EEG challenge, objective autonomic monitoring, active-minus-sham symptom-coupling analysis, and patient-derived cellular calcium/redox assays with pharmacologic, knockdown/rescue, and CRISPR allele-editing perturbations.

The central question becomes: under what waveform, genotype, physiological state, and recovery window does a person show reproducible loss of biological timing fidelity? This reframing preserves skepticism, respects patient experience, and gives the field a way to identify or reject a biologically definable EHS endophenotype without reducing patients to meters.

Keywords: electromagnetic hypersensitivity; EHS; RF-EMF; CACNA1C; rs7304986; calcium channels; voltage sensors; CYB5B; mitochondrial redox; sleep EEG; sleep spindles; bioelectric response; endophenotype

1. Introduction: the field has been asking a narrower question than it thinks

For more than two decades, the standard experimental debate around electromagnetic hypersensitivity has centered on acute provocation: expose the participant to an active or sham electromagnetic field, ask whether they feel symptoms or can identify the exposure state, and use detection accuracy or immediate symptoms as the decisive outcome. This is a legitimate design for one claim: that a person can consciously perceive EMF exposure in real time, or that short-term exposure reliably produces immediate symptoms under the tested conditions. It is not a sufficient design for every biologically plausible form of sensitivity.

The central problem is that the paradigm treats the person as a meter. In that frame, the participant either detects the field or fails to detect it. Yet a human organism is not a dosimeter. It is a dynamic control system whose responses depend on genotype, sleep pressure, circadian phase, inflammatory load, metabolic state, autonomic tone, prior exposure history, and recovery capacity. A biological response can be delayed, threshold-dependent, cofactor-dependent, learned, conditioned, or detectable only through objective physiology rather than conscious perception.

This paper does not claim that EHS has already been proven as a uniform EMF-caused disorder. It does not claim that all self-identified EHS symptoms are caused by EMF exposure. It does not reject nocebo or conditioning as real physiological contributors. Instead, it argues that the prevailing acute perception-provocation paradigm is too narrow to test the strongest biological hypothesis. The better question is not whether a participant can feel Wi-Fi within five minutes. The better question is: under what physiological state, genetic background, exposure pattern, and recovery window does a person show a measurable change in biological timing fidelity?

A maximum-impact EHS research program should therefore move from exposure detection to endophenotyping: defined waveform plus susceptible receiver biology plus objective bioelectric, calcium, redox, autonomic, or sleep response plus symptom coupling in some people. This is a falsifiable hypothesis. If no objective active-minus-sham physiology, no genotype or molecular moderation, no symptom-physiology coupling, and no cellular response are found under rigorous state-modeled conditions, the proposed endophenotype should be rejected. But if such patterns are found, the field will have moved beyond the false binary of ‘all psychological’ versus ‘all EMF-caused’ and into precision physiology.

2. Terminology: why this paper uses EHS

This paper uses the term electromagnetic hypersensitivity (EHS) intentionally. EHS is the term used by many affected individuals, is still used in scientific and public-health documents, and directly names the exposure class under investigation. The alternative term idiopathic environmental intolerance attributed to electromagnetic fields was introduced to avoid implying a known etiology. That caution has value in clinical classification. However, in a mechanistic research paper whose purpose is to specify testable biological pathways, an idiopathic label can obscure the experimental question.

As used here, EHS is not treated as a proven single disease entity, a diagnostic label, or a claim that all symptoms in all self-identified individuals are caused by EMF. EHS is used as a descriptive research term for individuals who report sensitivity to electromagnetic-field exposure and for the broader hypothesis that a subset may show objective, reproducible, waveform-dependent physiological responses. The term is therefore operational, not dogmatic. It identifies the population and the proposed sensitivity domain while leaving causation open to blinded testing.

This distinction matters. A descriptive term can be used while still applying rigorous skepticism. The study design proposed below explicitly allows several outcomes: objective response with symptom coupling; objective response without symptom coupling; symptoms without objective exposure-linked response; expectation- or conditioning-mediated autonomic response; environmental confounding; and no measurable effect. The purpose is not to force EHS into one explanation, but to separate mechanisms.

3. What the negative provocation literature does and does not show

The existing evidence cannot be ignored. WHO states that EHS symptoms are real and can be disabling, but also that EHS has no clear diagnostic criteria, no established scientific basis linking symptoms to EMF exposure, and is not known to represent a single medical problem [1,2]. A 2024 systematic review of human experimental RF-EMF studies included 41 studies and 2,874 participants and found no or only small non-significant effects of exposure on self-reported symptoms. The same review reported that study volunteers, including those classified as IEI-EMF/EHS, could not perceive RF-EMF exposure status better than chance [3].

Those results strongly challenge one claim: that self-identified EHS individuals can reliably detect RF-EMF exposure or develop acute symptoms under the short, controlled exposure conditions used in the included experiments. But the same review also acknowledged major limitations for the broader biological question: laboratory exposures differed substantially from real-life conditions in duration, frequency, distance, and position; most participants were young and healthy; outcomes were largely self-reported symptoms; available data did not allow assessment beyond acute exposure or in elderly or chronically ill populations; and a real effect in a susceptible subgroup could be masked by mixing sensitive and insensitive subjects [3].

Thus the most accurate interpretation is not ‘EHS is impossible.’ It is: acute symptom and perception-provocation studies have not demonstrated reliable conscious detection or immediate symptom causation under their tested conditions. That conclusion remains compatible with a more sophisticated hypothesis: some individuals may have delayed, threshold-based, cofactor-dependent physiological responses that are poorly captured by acute detection designs.

The analogy is not that EHS is food allergy. The analogy is methodological. In cofactor-dependent food anaphylaxis, exercise, NSAIDs, alcohol, and sleep deprivation can lower reaction thresholds or increase severity; the same food may be tolerated on one occasion and cause a serious reaction under another state [12]. A negative challenge without the relevant cofactors does not always exclude a cofactor-dependent response. Likewise, an acute EMF challenge in a resilient physiological state does not necessarily rule out a delayed or state-dependent response in a susceptible state. The appropriate scientific response is not to assume causation; it is to design a challenge that actually tests the proposed biology.

Table 1. Claims, evidence status, and disciplined wording

Claim	Evidence status	Publication-safe wording
EHS individuals can consciously detect RF-EMF in real time	Not supported by most controlled provocation literature	Do not make this the central claim; treat guessing accuracy as a control variable.
EHS symptoms are real and can be disabling	Recognized by WHO regardless of cause	Symptoms should be respected and clinically evaluated.
EHS is a single established disease diagnosis	Not established	Treat EHS as a heterogeneous research domain until subtypes are identified.
CACNA1C genotype can moderate sleep-EEG response to 3.6 GHz RF-EMF	Human proof-of-principle from a small randomized blinded study	Claim narrowly as genotype x waveform x objective EEG response.
CYB5B is an EHS marker	Not established	Use CYB5B as a mechanistic candidate for cellular assays, not as a diagnostic marker.
Defined AM RF-EMF effects can be calcium-channel dependent in some biological contexts	Supported in HCC model/device literature	Claim only as plausibility for channel-dependent EMF biology under defined conditions.
A genotype-stratified EHS endophenotype may exist	Hypothesis	Test prospectively with objective endpoints and blinded active/sham exposure.

 

4. The correct experimental question

The wrong question is: can this person consciously detect Wi-Fi and get a headache within five minutes? That is a meter test. It may be useful for identifying conscious perception or immediate nocebo-driven symptom reporting, but it is not a biologically complete test of sensitivity.

The better question is: under what waveform, genotype, physiological state, and recovery window does this person show a reproducible active-minus-sham change in objective biological timing systems, and do symptoms track that change? This reframes EHS from an exposure-detection claim into an endophenotype problem.

The proposed endophenotype is defined as a reproducible objective physiological response to a defined EMF waveform, moderated by genetic or molecular receiver state, with symptoms coupling to that response within a prespecified lag window in some individuals. This definition requires objective evidence and allows non-responders to remain non-responders. It also permits the possibility that symptoms in some people arise predominantly from expectation, conditioning, environmental confounding, or unrelated illness. The framework is therefore not a blanket validation of all EHS claims; it is a route to sorting them.

Core equation

Defined EMF waveform + susceptible receiver biology + vulnerable physiological state + measurable bioelectric/calcium/redox response + symptom coupling in some people = candidate EHS bioelectric response endophenotype.

 

5. Human proof-of-principle: CACNA1C-stratified 5G sleep EEG

The 2025 NeuroImage study by Sousouri and colleagues provides the strongest human anchor for this framework. It was not a subjective ‘I felt bad near Wi-Fi’ endpoint. It was a randomized, double-blind, sham-controlled crossover study of 34 participants genotyped for CACNA1C rs7304986: 15 T/C carriers and 19 matched T/T carriers. Participants underwent standardized left-hemisphere exposure to 3.6 GHz 5G RF-EMF, 700 MHz RF-EMF, or sham for 30 minutes before sleep. Sleep spindle activity was measured using high-density EEG and the FOOOF algorithm [4].

The key result was an exposure x genotype interaction. T/C carriers reported longer sleep latency than T/T carriers, and only 3.6 GHz exposure in T/C carriers induced faster NREM sleep-spindle center frequency in central, parietal, and occipital cortex compared with sham [4]. The study concluded that 3.6 GHz 5G RF-EMF modulated NREM spindle center frequency in a CACNA1C genotype-dependent manner and implicated L-type voltage-gated calcium channels in physiological RF-EMF response [4].

This result should be neither overstated nor minimized. It does not prove EHS. It does not show that people can consciously detect RF-EMF. It does not show harm. It does not establish clinical causation. It shows something narrower and scientifically valuable: a defined RF exposure can alter an objective sleep-EEG rhythm in a genotype-dependent manner. That is enough to justify a genotype-stratified research program.

The immediate implication is methodological. If a 34-person study can detect a genotype-dependent RF-EMF effect on a native EEG rhythm, then pooled acute symptom-provocation studies may be too blunt to detect response subgroups. The next EHS study should not ask only whether participants guess exposure. It should ask whether genotype-stratified participants show active-minus-sham changes in sleep-spindle center frequency, sleep latency, autonomic tone, and symptom coupling.

6. Observational support: CACNA1C rs2302729 and sleep complaints

A related observational study by Eicher and colleagues examined 2,040 participants who completed validated questionnaires on EMF sensitivity, sleep quality, daytime sleepiness, and diurnal preference, and provided saliva samples for genotyping CACNA1C variants. Participants who self-identified as electro-hypersensitive or who attributed symptoms to electromagnetic pollution reported prolonged sleep latency and reduced sleep quality compared with non-EHS participants. The T allele of CACNA1C rs2302729 was associated with both reduced subjective sleep quality and self-reported EMF sensitivity. Habitual mobile-phone use was not associated with self-rated sleep latency or sleep quality [5].

This study is hypothesis-generating rather than causal. It does not show that RF-EMF exposure caused poor sleep. It does not show that EHS mediates CACNA1C-related sleep impairment. But together with the 2025 sleep-EEG study, it makes CACNA1C the first serious human candidate gene for genotype-stratified EHS research. The correct inference is not ‘CACNA1C is the EHS gene.’ The correct inference is that calcium-channel regulatory genetics may alter sleep-related bioelectric response phenotypes and should be prospectively tested.

7. Receiver biology: voltage-gated calcium channels, S4 voltage sensors, and regulatory variants

Voltage-gated calcium channels are biologically coherent receiver-state candidates because they convert membrane voltage dynamics into calcium entry and downstream signaling. The pore-forming alpha1 subunit of voltage-gated calcium channels is organized into four homologous domains, each with six transmembrane segments, S1-S6. The S4 segment contains gating charges that sense changes in the electric field and initiate conformational changes that open the pore; the S5-S6 pore loop determines conductance and selectivity [6].

This does not mean that every relevant variant must be an S4 coding mutation. The CACNA1C rs7304986 variant in the 2025 NeuroImage study is intronic. It should not be described as an S4 amino-acid mutation. Its relevance may lie in expression, splicing, chromatin state, transcript regulation, developmental context, or network-level calcium-channel contribution to sleep rhythms. Therefore, the genetic design should include both coding variants in S4/S4-S5/pore regions and noncoding regulatory variants that alter channel behavior.

This is why a simple SNP panel is too weak for a definitive study. Whole-genome sequencing, transcriptomic context, chromatin annotation, allele-specific expression, and functional cellular assays are needed. The target is not an ‘EHS gene.’ The target is a receiver-state architecture: the inherited and regulatory configuration of channels, mitochondria, circadian machinery, and autonomic control systems that determines how a defined waveform is transduced or ignored.

8. Calcium timing, not just calcium amount

The strongest mechanistic version of this hypothesis is not that EMF simply increases intracellular calcium. The stronger hypothesis is that field structure can alter calcium timing. Calcium oscillations are information-bearing signals. Frequency modulation of calcium oscillations can differentiate biological responses in health and disease, and cellular frequency-decoding molecules are central to how cells control important programs [7]. Other reviews similarly describe cytosolic calcium oscillations as flexible signals whose frequency and amplitude can encode specific messages for downstream molecular events [8].

Therefore, the relevant readouts are not only peak calcium or average intracellular calcium concentration. They include oscillation frequency, amplitude, duty cycle, inter-spike interval variability, waveform regularity, localization, spectral entropy, phase locking, and wavelet coherence. A defined EMF waveform could, in principle, have no effect on average calcium but still alter the temporal structure of calcium signaling. Conversely, a symptom-free objective responder might show a measurable calcium or EEG timing shift without conscious perception.

This distinction is crucial for EHS research. If cells and neural circuits encode information through timing, a sensitivity phenotype may appear as loss of timing fidelity rather than as immediate sensation. The human analogue is the CACNA1C sleep-spindle finding: the endpoint is not a symptom declaration but a shift in a native oscillatory rhythm.

9. CYB5B, mitochondrial-redox coupling, and EMF-inducible gene control

The 2026 Cell paper on an electromagnetic-field-inducible in vivo gene switch is relevant as mechanistic plausibility, not as EHS proof. The study reports an EMF-inducible gene switch for remote spatiotemporal control of gene expression and identifies cytochrome b5 type B (Cyb5b) through a CRISPR-Cas9 screen as an essential mediator. The indexed Cell summary describes Cyb5b-mediated EMF-specific calcium oscillations as central to switch activation [9].

The proper inference is limited but important: a biological system can be engineered or configured so that a defined EMF program is transduced into patterned calcium oscillations and gene-expression output. CYB5B/Cyb5b is therefore a candidate transduction node worth studying in EHS-related cell assays, especially because mitochondrial-redox state and calcium dynamics are deeply coupled. However, CYB5B is not currently an EHS marker. No published evidence yet shows that naturally occurring CYB5B variants identify EHS individuals or that CYB5B mediates symptoms in humans.

The cellular part of the proposed study should therefore treat CYB5B as a mechanistic candidate. Patient-derived cells from objective responders and non-responders can be exposed under blinded active/sham conditions, and CYB5B can be knocked down, overexpressed, rescued, or allele-edited. If response signatures appear or disappear in the same genetic background after CYB5B perturbation, the field would move from association toward mechanism.

10. Channel-dependent AM RF biology: the CACNA1H/TheraBionic clue

A separate biomedical line of evidence supports the narrower claim that defined non-thermal amplitude-modulated RF-EMF biology can be channel-dependent under specific conditions. FDA describes TheraBionic P1 as a handheld RF electromagnetic-field generator whose antenna is placed in the mouth and emits specific amplitude-modulated frequencies for adults with advanced hepatocellular carcinoma after failed therapy. FDA lists calcium-channel blockers among conditions in which the device should not be used [10].

The related 2019 EBioMedicine study reported that intrabuccal administration of amplitude-modulated 27.12 MHz RF-EMF produced systemic athermal exposure at levels lower than those generated by cell phones held close to the body, and that anti-proliferative and cancer stem-cell effects in hepatocellular carcinoma cells were mediated by calcium influx through CaV3.2 T-type voltage-gated calcium channels encoded by CACNA1H [11].

This does not prove EHS. HCC cells are not neurons, sleep circuits, or EHS patients. The waveform, tissue context, disease state, and endpoint are different. But it does support one key premise: under defined conditions, non-thermal EMF biology can interact with voltage-gated calcium-channel pathways. CACNA1H should therefore be included as a mechanistic candidate, especially in cellular assays and genotype-stratified analyses.

11. Protein-scale magnetic/RF responsiveness: plausibility boundary, not proof

The 2026 Nature MagLOV study provides a plausibility boundary for protein-scale magnetic and RF-responsive biology. The authors reported engineered magneto-sensitive fluorescent proteins that exhibit optically detected magnetic resonance in living bacterial cells at room temperature, with effects explained through a radical-pair mechanism involving the protein backbone and bound flavin cofactor [13].

This evidence should be used carefully. It does not show that ordinary human proteins respond pathologically to telecom RF fields. It does not prove EHS. It does show that engineered proteins can be made magnetically/RF responsive in living cells and that radical-pair-like biology can be measurable under biological conditions. In a paper on EHS, this belongs in the plausibility section, not in the proof section. The primary human anchor remains CACNA1C-stratified sleep EEG.

12. Operationalizing ‘low-fidelity biology’

‘Low-fidelity biology’ is useful as a metaphor, but a scientific paper must convert it into variables. Here, biological timing fidelity means the stability, coherence, and reproducibility of oscillatory physiological systems. Timing fidelity can be measured in EEG, autonomic physiology, calcium imaging, mitochondrial redox dynamics, and symptom-physiology coupling.

The proposed measurable variables include spectral entropy, phase-locking value, wavelet coherence, inter-event interval variability, rhythm regularity, active-minus-sham signal-to-noise ratio, sleep-spindle center frequency, HRV coherence, calcium oscillation duty cycle, mitochondrial redox oscillation stability, and cross-system coupling. In this framework, a defined EMF exposure is not treated as a generic toxin. It is tested as a possible timing perturbation.

The most conservative prediction is not that all EHS subjects will show global biological disruption. The prediction is that a subset, defined by genotype and physiological state, will show a reproducible active-minus-sham timing shift in one or more objective systems. Symptoms may couple to that shift in some but not all responders. This allows a four-part taxonomy: objective responder with symptom coupling; objective responder without symptom coupling; symptom responder without objective exposure-linked physiology; and non-responder.

Table 2. Objective endpoint domains

Domain	Primary measures	Why it matters	Interpretation rule
Sleep EEG	NREM spindle center frequency, spindle density, microarousals, NREM architecture	Direct human anchor from CACNA1C/5G study	Primary active-minus-sham endpoint.
Autonomic physiology	HRV, skin conductance, pupillometry, blood pressure, respiration	Captures conditioned and exposure-linked autonomic response	Analyze against exposure, guessing, expectation, and state.
Symptoms	Repeated ratings during exposure and 0-72 h after	Needed for clinical relevance	Not sufficient alone; test coupling to objective physiology.
Cellular calcium	Frequency, amplitude, duty cycle, entropy, phase locking	Tests timing-code hypothesis	Compare active vs sham in blinded assays.
Mitochondrial redox	Membrane potential, NADH/FAD ratio, ROS, ATP, OCR	Links calcium timing to metabolic state	Use genotype and responder status as moderators.
Mechanistic perturbation	Knockdown, rescue, blocker, CRISPR allele swap	Moves from association to causation	Response should appear, disappear, or change direction predictably.

 

13. Proposed staged study: Genotype-Stratified Bioelectric Response Testing in EHS

The proposed study combines naturalistic monitoring, genotype stratification, blinded provocation, objective physiology, cellular assays, and mechanistic perturbation. Its goal is not to diagnose EHS by a single SNP. Its goal is to determine whether self-identified EHS individuals contain a biologically definable responder subgroup.

Primary hypothesis: Certain CACNA1C, CACNA1H, CACNA1F, CYB5B, and voltage-sensor/channel-regulatory variants predict objective active-minus-sham physiological responses to defined RF/ELF EMF exposure under blinded conditions.

Secondary hypothesis: In self-identified EHS subjects, symptoms are more likely to track objective bioelectric, autonomic, calcium, redox, or sleep changes in genetically or molecularly susceptible subgroups than in non-susceptible subgroups.

The study should be conducted in two stages: discovery and locked replication. The discovery stage identifies candidate responder signatures and exposure-response lags. The replication stage tests prespecified endpoints and decision rules in an independent cohort. Without replication, the study risks becoming a high-dimensional fishing exercise.

Table 3. Staged protocol

Phase	Duration / method	Measures	Purpose
1. Naturalistic monitoring	30-90 days	Personal RF/ELF dosimetry, sleep, HRV, symptoms, diet, stress, illness, caffeine, alcohol, medications, light, noise, temperature	Define individual state, exposure patterns, and lag windows before lab challenge.
2. Trigger modeling	Distributed-lag models	0-2 h, 2-8 h, 8-24 h, 24-72 h windows	Avoid assuming immediate symptom onset; identify candidate cascade timing.
3. Genotype stratification	Whole-genome sequencing	CACNA1C, CACNA1H, CACNA1F, CYB5B, auxiliary channel subunits, S4/S4-S5/pore variants, redox and circadian genes	Define receiver-state candidates, including noncoding regulatory variation.
4. Blinded human challenge	Randomized double-blind sham-controlled crossover	3.6 GHz pre-sleep exposure, sham, optional 700 MHz; high-density EEG and autonomic physiology	Test objective active-minus-sham response and genotype moderation.
5. Symptom coupling	During exposure and 0-72 h follow-up	Symptoms, guessing accuracy, expectation/threat belief, objective physiology	Determine whether symptoms track objective physiology better than guessing or expectation.
6. Cellular validation	Blinded active/sham assays	Calcium waveforms, redox, ROS, ATP, OCR, transcriptomics	Test whether patient-derived cells show parallel molecular response.
7. Mechanistic perturbation	Pharmacology, knockdown/rescue, CRISPR	CYB5B, CACNA1C, CACNA1H, CACNA1F	Determine whether candidate pathways are necessary or sufficient for response.

 

14. Human challenge protocol

The first human protocol should replicate and extend the CACNA1C sleep-EEG template. Participants should include self-identified EHS subjects and non-EHS controls, with deliberate genotype stratification. The primary exposure condition should be a calibrated 3.6 GHz RF-EMF exposure delivered for 30 minutes before sleep, with a physically indistinguishable sham condition. A 700 MHz condition can be included to mirror the NeuroImage study and test frequency specificity. Exposure must be double-blind, randomized, and counterbalanced, with careful control of heating, device light, acoustic cues, vibration, room temperature, humidity, experimenter leakage, and order effects.

The primary endpoint should be active-minus-sham change in NREM sleep-spindle center frequency, stratified by CACNA1C rs7304986 genotype. The key EHS-specific endpoint should be whether active-minus-sham physiological shifts correlate with active-minus-sham symptoms in prespecified lag windows. Guessing accuracy should be measured, but it should not be the primary endpoint. Guessing accuracy is a control variable, not the scientific center of the study.

Secondary endpoints should include sleep latency, sleep efficiency, NREM architecture, spindle density, microarousals, HRV, skin conductance, pupillometry, blood pressure, respiration, reaction time, cognitive performance, symptom ratings, exposure-detection accuracy, expectation ratings, and threat-belief scores. The analysis should explicitly test whether physiology tracks exposure, whether symptoms track physiology, and whether either is explained by expectation or guessing.

15. Cofactor and state modeling

The state-dependent model must be ethically operationalized. It is not acceptable to push participants into extreme physiological stress merely to elicit symptoms. The study should use mild, standardized, review-board-approved cofactors such as controlled prior sleep duration, standardized caffeine abstinence, standardized meal timing, consistent light exposure, menstrual/hormonal tracking, medication documentation, illness exclusion or stratification, and stable room conditions.

The naturalistic phase should determine whether a participant’s symptoms or objective physiology are associated with specific exposure patterns and lag windows. Laboratory challenge can then test the most plausible waveform and state combination without abandoning blinding or safety. This is the difference between personalized provocation and uncontrolled anecdote. The waveform is individualized only after predefined modeling, and active/sham assignment remains hidden.

16. Genomic and molecular stratification

The genetic arm should begin with whole-genome sequencing rather than a narrow SNP panel. Candidate-gene analysis should include CACNA1C rs7304986, CACNA1C rs2302729, CACNA1C coding and regulatory haplotypes, CACNA1H, CACNA1F, CYB5B, calcium-channel auxiliary subunits, broader voltage-gated channel S4/S4-S5/pore variants, mitochondrial-redox genes, circadian genes, and autonomic/migraine/sleep-related loci as covariates.

Noncoding variation is not secondary. It may be central. The CACNA1C sleep-EEG anchor involves an intronic variant, so regulatory architecture, splice effects, allele-specific expression, enhancer context, and chromatin accessibility are all plausible. Genotype should be treated as a receiver-state moderator, not a deterministic diagnosis.

17. Cellular and mechanistic sub-study

The cellular study should use PBMCs, fibroblasts, and, where resources allow, iPSC-derived neurons or glial-neuronal co-cultures from EHS objective responders, EHS non-responders, and non-EHS controls. Genotype-matched controls should be included where possible. Cells should be exposed under blinded active/sham conditions using the same waveform or a mechanistically linked waveform, with rigorous control of temperature, vibration, acoustic noise, media conditions, incubator effects, batch, passage number, and operator blinding.

Primary cellular endpoints should include calcium oscillation frequency, amplitude, duty cycle, inter-spike interval variability, calcium spectral entropy, phase locking, mitochondrial membrane potential, NADH/FAD redox ratio, ROS, ATP, oxygen consumption rate, transcriptomics, and NFAT/CREB/immediate-early-gene activation.

Mechanistic perturbations should include CYB5B knockdown and rescue, CACNA1C knockdown and rescue, CACNA1H knockdown and rescue, CACNA1F perturbation where biologically appropriate, in vitro calcium-channel blockers, mitochondrial redox perturbation, and CRISPR allele swaps. The gold-standard proof would be: same genetic background, candidate allele edit, active/sham exposure, and response appears, disappears, or changes in the predicted direction.

18. Statistical framework

The human EEG analysis should use mixed-effects models appropriate for crossover data, with participant as a random effect and exposure, genotype, EHS status, order, sex, age, sleep pressure, chronotype, caffeine, medication, baseline sleep quality, and relevant state variables as fixed effects. The primary interaction is exposure x CACNA1C rs7304986 genotype on NREM spindle center frequency. The EHS-specific interaction is exposure x objective physiology x symptom change, with lag windows prespecified.

A representative primary model is: spindle center frequency = exposure + genotype + EHS status + exposure x genotype + exposure x EHS status + exposure x genotype x EHS status + covariates + participant random effect. In practice, the model should analyze active-minus-sham contrasts and correct for order and night effects.

The cellular analysis should model active-minus-sham calcium/redox features as a function of genotype, objective-responder status, EHS status, cell type, passage number, batch, and exposure condition. High-dimensional outcomes such as transcriptomics and calcium-feature panels require preregistered multiple-comparison correction and locked validation. Discovery clustering should never be presented as proof until replicated.

Table 4. Falsifiable predictions

Prediction	Falsification / interpretation
CACNA1C rs7304986 T/C carriers show larger active-minus-sham NREM spindle center-frequency shift after 3.6 GHz exposure than matched T/T carriers.	If absent in a powered replication, the CACNA1C anchor weakens substantially.
Only a subset of self-identified EHS subjects meet objective-responder criteria.	If all or none respond, the subtype model must be revised.
Objective responders show stronger symptom-physiology coupling than non-responders.	If symptoms track only guessing/expectation, the exposure-linked clinical phenotype is not supported.
Active/sham guessing accuracy is not required for objective physiological response.	If objective effects only occur when exposure is guessed, expectation becomes the leading explanation.
Patient-derived cells from objective responders show active-minus-sham calcium/redox waveform differences.	If no cellular parallel is found, mechanism remains limited to organism-level physiology or becomes weaker.
Perturbing CACNA1C, CACNA1H, CACNA1F, or CYB5B pathways alters cellular response signatures.	If perturbation has no effect, candidate mechanism is unsupported.
If no objective physiology, no genotype moderation, no cellular response, and no symptom coupling are found under state-modeled conditions, the proposed endophenotype should be rejected.	This is the decisive falsifiability criterion.

 

19. Controls, artifacts, and blinding integrity

A study of this kind will be judged by its controls. The exposure system must control and document RF/ELF dosimetry, SAR or induced-field estimates as appropriate, device heating, coil or antenna heating, acoustic noise, vibration, airflow, light emission, ozone, room temperature, humidity, electrode artifacts, and experimenter leakage. Sham must be physically indistinguishable from active exposure. Participants and staff interacting with participants must remain blind to condition. Data analysts should ideally be blinded until the primary pipeline is locked.

The study must also measure expectation, threat belief, prior conditioning, exposure guessing, and environmental concern. These variables should not be treated as embarrassing confounders. They are physiological variables that can alter autonomic tone, sleep, migraine threshold, blood pressure, gut motility, and symptom perception. A good study does not deny nocebo or conditioning; it determines whether exposure-linked physiology remains after accounting for them.

20. Ethical and clinical posture

The framework should not be used as a clinical diagnostic test until replicated. Participants should receive medical evaluation for alternative causes of symptoms, including sleep disorders, migraine, dysautonomia, endocrine disease, medication effects, indoor-air problems, ergonomic stressors, psychiatric comorbidities, and other treatable conditions. Respecting EHS reports does not mean neglecting standard medicine.

The study should avoid inducing severe stress, sleep deprivation, or prolonged exposures. It should not encourage participants to abandon necessary medical care or isolate socially. The ethical posture is patient-respecting but evidence-demanding: symptoms are real, mechanisms may be mixed, and the purpose of research is to identify which mechanisms apply to which people.

21. Discussion: replacing the meter test with precision physiology

The central methodological error in much of the EHS debate is the assumption that failure of conscious exposure detection is equivalent to absence of biological response. That assumption would be inappropriate in many areas of medicine. Biological responses can be delayed, threshold-dependent, cofactor-dependent, and moderated by genotype. Therefore, the relevant question is not whether a participant can identify when a router is active. The relevant question is whether a defined waveform, delivered under controlled and biologically plausible conditions, produces a reproducible objective change in a susceptible physiological system, and whether that change predicts symptoms better than expectation, guessing, or nonspecific distress.

The 2025 CACNA1C sleep-EEG study is important precisely because it does not depend on conscious detection. It shows that an RF-EMF exposure can shift a native EEG rhythm in a genotype-dependent manner. The 2024 Eicher observational study does not prove causality, but it identifies CACNA1C as a plausible sleep/EHS candidate. Voltage-gated calcium-channel architecture makes S4-containing channels biologically coherent targets, while the intronic nature of rs7304986 reminds us that noncoding regulation may be as important as coding variation. Calcium oscillation literature shows that timing structure matters. CYB5B and channel-dependent AM RF literature provide additional mechanistic candidates, not proof.

This is a stronger position than either dismissal or overclaim. It accepts that acute provocation studies have not supported reliable EMF detection. It also recognizes that those studies may be poorly matched to a delayed, genotype- and state-dependent timing-perturbation hypothesis. The proposed study can therefore move the field from argument to measurement.

22. Limitations of the framework

Several limitations must be made explicit. First, the CACNA1C/5G sleep-EEG study was small and not an EHS study. It requires powered replication. Second, an EEG spindle-frequency shift is not harm. Clinical relevance depends on replication, magnitude, dose-response, and symptom coupling. Third, EHS is likely heterogeneous; the framework should expect multiple mechanisms and many non-responders. Fourth, CYB5B evidence comes from an engineered gene-switch system and cannot be assumed to apply directly to naturally occurring human EHS. Fifth, HCC AM RF-EMF data involve cancer cells and a medical-device context, not nervous-system sensitivity. Sixth, MagLOV shows engineered protein-scale responsiveness, not natural human pathology. Seventh, high-dimensional genomic, EEG, and cellular datasets are vulnerable to false discovery without preregistration and replication.

These limitations do not weaken the rationale for the study. They define the discipline needed to make it credible. The paper should not claim that existing safety guidelines have been overturned by this framework. It should claim that a more precise experimental question is now justified.

23. Conclusion

EHS research should stop treating people like meters. The failure of many self-identified EHS individuals to consciously detect RF-EMF exposure in short provocation studies is an important finding, but it is not the final biological word. Humans are state-dependent physiological systems. If a subgroup exists, it will likely be discovered through genotype-stratified objective physiology, not through asking people to guess whether a router is on.

The most rigorous next step is a genotype-stratified, double-blind, sham-controlled bioelectric response study anchored in CACNA1C sleep EEG, expanded through autonomic monitoring, naturalistic dosimetry, symptom-coupling analysis, patient-derived calcium/redox cellular assays, and mechanistic perturbation. The core test is simple: defined waveform plus susceptible receiver biology plus vulnerable state should produce reproducible active-minus-sham physiological change; in a clinically relevant subgroup, symptoms should couple to that change.

If the pattern appears and is mechanistically perturbable, EHS will contain a biologically definable endophenotype. If it does not, the proposed endophenotype should be rejected. Either outcome is progress. The field needs less argument about whether patients are good meters and more science about whether some human biological systems are susceptible receivers.

References

[1] World Health Organization. Electromagnetic hypersensitivity. Radiation and health. WHO states that EHS symptoms are real and can be disabling, but that EHS has no clear diagnostic criteria and no established scientific basis linking symptoms to EMF exposure. URL: https://www.who.int/teams/environment-climate-change-and-health/radiation-and-health/non-ionizing/hypersensitivity

[2] World Health Organization. Electromagnetic hypersensitivity: proceedings, International Workshop on Electromagnetic Field Hypersensitivity, Prague, Czech Republic, October 25-27, 2004. Published 2006. ISBN: 9789241594127. URL: https://www.who.int/publications/i/item/9789241594127

[3] Bosch-Capblanch X, Esu E, Oringanje CM, Dongus S, Jalilian H, Eyers J, Auer C, Meremikwu M, Roosli M. The effects of radiofrequency electromagnetic fields exposure on human self-reported symptoms: A systematic review of human experimental studies. Environment International. 2024;187:108612. doi:10.1016/j.envint.2024.108612.

[4] Sousouri G, Eicher C, D’Angelo RM, Billecocq M, Fussinger T, Studler M, Capstick M, Kuster N, Achermann P, Huber R, Landolt H-P. 5G radio-frequency-electromagnetic-field effects on the human sleep electroencephalogram: A randomized controlled study in CACNA1C genotyped volunteers. NeuroImage. 2025;121340. doi:10.1016/j.neuroimage.2025.121340.

[5] Eicher C, Marty B, Achermann P, Huber R, Landolt H-P. Reduced subjective sleep quality in people rating themselves as electro-hypersensitive: An observational study. Sleep Medicine. 2024;113:165-171. doi:10.1016/j.sleep.2023.11.029.

[6] BPS/IUPHAR Guide to Pharmacology. Voltage-gated calcium channels (CaV): introduction. Describes alpha1 subunit domain organization and S4 gating charges. URL: https://www.guidetopharmacology.org/GRAC/FamilyIntroductionForward?familyId=80

[7] Smedler E, Uhlen P. Frequency decoding of calcium oscillations. Biochimica et Biophysica Acta. 2014;1840(3):964-969. doi:10.1016/j.bbagen.2013.11.015.

[8] Uhlen P, Fritz N. Biochemistry of calcium oscillations. Biochemical and Biophysical Research Communications. 2010;396(1):28-32. doi:10.1016/j.bbrc.2010.02.117.

[9] Kim J et al. Electromagnetic field-inducible in vivo gene switch for remote spatiotemporal control of gene expression. Cell. 2026. PII: S0092-8674(26)00330-2. URL: https://www.cell.com/cell/abstract/S0092-8674(26)00330-2

[10] U.S. Food and Drug Administration. TheraBionic P1 – H220001. Approval date September 26, 2023; content current June 4, 2024. URL: https://www.fda.gov/medical-devices/recently-approved-devices/therabionic-p1-h220001

[11] Jimenez H, Wang M, Zimmerman JW, et al. Tumour-specific amplitude-modulated radiofrequency electromagnetic fields induce differentiation of hepatocellular carcinoma via targeting CaV3.2 T-type voltage-gated calcium channels and Ca2+ influx. EBioMedicine. 2019;44:209-224. doi:10.1016/j.ebiom.2019.05.034.

[12] Bartra J, et al. Cofactors in food anaphylaxis in adults. Journal of Allergy and Clinical Immunology: In Practice. 2023. Describes exercise, NSAIDs, alcohol, and sleep deprivation as frequent cofactors that can lower reaction thresholds or increase severity.

[13] Abrahams G, et al. Quantum spin resonance in engineered proteins for multimodal sensing. Nature. 2026. doi:10.1038/s41586-025-09971-3.