This was a 3-week, remote, randomized, double-blind, placebo-controlled trial conducted entirely through secure digital platforms. All procedures complied with the Declaration of Helsinki. Participants provided electronic informed consent.

The study was designed as a two-armed, double blinded sham-controlled study with a three-week treatment period with a measurement point at the beginning and at the end of the study. A study protocol was finalized before commencement of recruitment. Volunteers were recruited via an existing network of people interested in this type of treatment. After signing the online informed consent form, participants were randomly assigned to one of the two study groups:

Participants who already use the investigational pulsed magnetic field device in daily routine (existing clients) but are not already having any experience in using the new application designed for improving sleep problems used this new application two times per day for 3 weeks with a frequency range between 5 Hertz and 970 Hertz

Same as application group A but deactivated PEMF device (sham application)

Participants were volunteers who have one or more of the sleep related problems non-restorative sleep, problems of falling asleep or frequently awakening at night and would profit from some self-help treatment in their general wellbeing, coping with particular issues, such as sleep problems, low affect or lack of energy. They gave informed consent to participate prior to any study specific procedure.

Volunteers were advised to participate only if they would not violate any restrictions for use as given in the Instructions for Use of the MagHealy device (not intended for pregnant women, persons with epilepsy, heart disease or those with a pacemaker).

According to the MagHealy product specification the device generates a pulsating magnetic field. Key specifications include a frequency range of 0-12.5 MHz, a portable, compact design, and a Bluetooth range of up to 10 meters for app connectivity. It creates an electromagnetic field up to 300 µT. 

Participants assigned to group A applied the frequency program Night Harmony Female/Male. This program is especially designed for improving sleep quality. The program was applied according to a predefined treatment plan once per day before going to bed for a total of 3 weeks.

Participants of group B are assigned to use the same program, but the devices of these participants used a placebo program (deactivated device), which enables the LED of the MagHealy device to work normally.

A combination of validated patient-reported outcome measures and daily electronic sleep diaries were used to assess treatment effects on sleep quality, sleep continuity, mood, daytime functioning, and overall well-being.

Sleep quality and functional impact were assessed using the Pittsburgh Sleep Quality Index 1920, a validated 19-item instrument evaluating perceived sleep parameters over the previous month. The PSQI yields:

· Global Score (0–21): higher scores indicate poorer sleep quality

· Seven Component Scores, including Sleep latency; Sleep duration, Sleep efficiency, Sleep disturbances, Daytime dysfunction/restorativeness, Sleep quality, Use of sleep medication

PSQI Components 2, 5b, and 7 were prespecified as primary efficacy measures depending on the participant’s enrolled stratum.

Participants completed a structured daily electronic sleep diary based on PSQI domains, including the following: Sleep onset latency (minutes), Number and duration of nighttime awakenings, Total sleep time (minutes), Time in bed and calculated sleep efficiency (%), Morning restorativeness (0–10 rating scale), Daytime fatigue, alertness, and adverse symptoms (0–10 scales)

Daily sleep diary entries were aggregated into weekly means for efficacy analyses to reduce within-subject variability and improve the stability of outcome measures. This approach minimizes the influence of day-to-day fluctuations in sleep patterns, enhances precision in estimating treatment effects, and aligns the temporal resolution of sleep outcomes with other weekly clinical assessments. Aggregation also mitigates the impact of occasional missing diary entries, providing more robust and interpretable efficacy results.

General psychological well-being was evaluated using the WHO-5 212223, a five-item measure assessing positive mood and life satisfaction over the preceding 2 weeks. Each item is rated 0–5, summed and converted to a 0–100 score, with higher scores reflecting better well-being.

Individualized symptom burden and personalized treatment benefit were evaluated using MYMOP 242526, enabling each participant to identify their most distressing symptoms and associated functional limitations. Scores for the primary symptom were assessed weekly on a 11-point scale (0 to 10), where higher scores indicate greater severity or impairment.

Daytime sleepiness was measured using the Epworth Sleepiness Scale 272829, an 8-item questionnaire assessing the likelihood of dozing in common situations. Total scores range from 0–24, with higher scores representing increased excessive daytime sleepiness.

The primary efficacy endpoint was the change from baseline to week 3 in the Pittsburgh Sleep Quality Index (PSQI) global score, assessing overall subjective sleep quality.

In addition, the self-assessments of general satisfaction with sleep quality, restfulness of sleep and difficulties falling asleep were evaluated as co-primary endpoints, enabling a multidimensional assessment of treatment effects on specific sleep domains.

Both the global PSQI scores and the self-assessment scores were analyzed as continuous variables using analysis of covariance (ANCOVA) with baseline values as covariates and treatment group as fixed factor. Adjusted mean changes from baseline and between-group differences with 95% confidence intervals were reported.

To preserve interpretability across sleep phenotypes, exploratory subgroup-specific analyses were also conducted to examine differential response patterns among participants with (1) severe difficulty falling asleep, (2) awakening at night, or (3) non-restorative sleep at baseline.

Secondary efficacy endpoints were evaluated across the entire study population and, where relevant, within each symptom-based stratum. These endpoints were designed to further characterize the impact of pulsed magnetic therapy on sleep continuity, sleep quality, next-day functioning, and psychological symptoms:

Safety and tolerability outcomes included: Change in Epworth Sleepiness Scale (ESS) total score (Survey) incidence, severity, and relatedness of Treatment-Emergent Adverse Events (TEAEs) Frequency of treatment discontinuations due to adverse events.

Subgroup analyses were prespecified in the statistical analysis plan to further characterize treatment effects across clinically relevant patient subsets and to explore potential moderators of therapeutic response. In addition to analyses performed in the overall study population, predefined subgroup analyses were conducted to examine possible differences in efficacy among participants presenting with distinct patterns of sleep disturbance at baseline. The subgroups were defined according to established clinical thresholds derived from the Pittsburgh Sleep Quality Index (PSQI) and a self-reported measure of perceived sleep restoration, perceived sleep quality and difficulties falling asleep.

1. Severe sleep problems: Participants with pronounced sleep problems were defined as those with a PSQI total score at baseline of higher 5 30. This criterion corresponds to a sleep quality typically considered clinically meaningful and indicative of significant insomnia symptoms.

2. Satisfaction with sleep quality, non-restorative sleep and difficulties falling asleep: Participants reporting low subjective satisfaction with sleep, low restoration by sleep and or distinct problems falling asleep were identified as those with a self-assessment score of less or equal to 5 on a 0–10 Likert scale, where higher values denote greater perceived satisfaction by sleep 31. This threshold was selected to capture individuals experiencing notably problems with the corresponding item. 

These subgroups were established a priori to assess whether the efficacy of pulsed magnetic therapy varied according to baseline sleep characteristics. The analyses were intended to identify potential modifiers of treatment response and to generate hypotheses for future targeted investigations.

Analyses will follow the intention-to-treat principle.

All primary and secondary efficacy analyses were performed on the intent-to-treat (ITT) population, defined as all randomized participants who completed at least one post-baseline assessment. Statistical significance tests were two-sided with α = 0.05, without adjustment for multiplicity unless otherwise specified in the hierarchical testing plan.

For continuous endpoints with normally distributed residuals and sufficient data completeness, analysis of covariance (ANCOVA) was used to compare treatment groups at the primary evaluation timepoint (Week X). Each model included:

· Treatment group (pulsed magnetic therapy vs placebo)

· Baseline value of the endpoint as a covariate

Adjusted mean differences between groups were estimated with 95% confidence intervals.

Longitudinal and Sensitivity Analyses

When repeated measures across time provided greater power or insight, repeated-measures ANOVA or mixed-effects models (MMRM) were used, including:

· Fixed effects for treatment, visit (time), treatment × visit interaction, and baseline

· An unstructured or autoregressive covariance structure, selected by model fit criteria

These models assume data missing at random (MAR).

For ordinal variables, non-normal distributions, or endpoints violating parametric assumptions, non-parametric tests were applied, including:

Wilcoxon rank-sum test for between-group comparisons

Wilcoxon signed-rank test for within-group change

Hodges–Lehmann estimator and 95% CI for effect size

Safety outcomes were summarized descriptively in the safety population (all participants who initiated treatment). Adverse events were coded and analyzed by incidence, severity, and relatedness.

For survey-based instruments (PSQI, WHO-5, MYMOP, Epworth Sleepiness Scale), missing post-baseline data were replaced using last observation carried forward (LOCF). The last available non-missing score was carried forward to all subsequent missing timepoints, provided the participant had at least one post-baseline assessment.

Diary-based data were processed using the following approach:

· Short-term missing days: If ≤3 consecutive diary days were missing within a given week, the missing values were replaced using LOCF from the most recent prior day.

· Extended missing days: If >3 consecutive diary days were missing within a week, the entire weekly summary was treated as missing.

Analyses were performed using validated statistical software (R v4.3.4).

Safety analyses will summarize adverse event frequencies descriptively. 

This pilot study was conducted without prior case number estimation. It was assumed that 120 evaluable cases per study group would be sufficient to detect significant group differences.

242 people consented to the study protocol and were randomly allocated to one of both study groups: active study group (A) or to the sham control group (B), resulting in 121 participants in each study group. Because of withdrawal of consent prior to study start, in total 24 participants who have given informed consent but did not participate in the study. 217 participants completed the first survey consisting of the study population according to the ITT principle. 16 participants terminated the study prematurely (8 in each group). Two hundred  and one participants completed both surveys and terminated the study according to the protocol.

Since slightly more participants from the control group withdrew consent prior to study start, the groups were somewhat unevenly distributed. The active group consisted of 111 participants, while 106 study participants were assigned to the control group.

16 participants (8 in each group) did not terminate the study regularly and did not complete the final survey. This missing data was replaced according to the last observation carried forward principle.

As can be seen from Table 1, the randomization process yielded quite comparable groups. The majority of the participants, 89%, were female. Due to data-protection concerns age was only collected in rough categories. 70 percent or 80 participants belonged to the middle-aged group between 41 and 60, and nearly half of the rest were older than 60 years or younger than 40 years.

The baseline values of all endpoint parameters differed only marginally in both groups. The respective 95% confidence intervals overlapped and the t-tests for independent samples did not reveal any significant group differences (survey: t = 0.31, p = 0.76 for PSQI, t = 1.96, p = 0.06 for WHO-5; t = 1.52, p = 0.13 for Mymop, t = 1.85, p = 0.07 for ESS; df 215 for each; Diary : t = 1.43, p = 0.15 for restfulness of sleep, t = 1.33, p = 0.18 for awakening frequency; t = 0.59, p = 0.55 for time to fall asleep, t = 0.65, p = 0.52 for daytime tiredness; df 200 for each).

Participant’s compliance was determined by the frequency of study programs used during the study period. This is tracked in a cloud and can be read from there. In both groups, compliance, defined as at least 50% of the intended applications performed, was close to 70%. The two groups differed only very slightly (CHIQU-test, p = 0.89).

Both active and control groups demonstrated meaningful improvements in overall sleep quality over the study period (Figure 1). Mean reductions in PSQI global score were –3.24 (active) and –2.89 (control), corresponding to a small, non-significant between-group effect (Cohen’s d = 0.14, p = 0.314; Table 2).

Improvements were more pronounced for subjective sleep experiences. Participants receiving active treatment reported significantly greater satisfaction with sleep (mean increase = 1.78 vs. 1.18; Cohen’s d = 0.30, p = 0.029) and restfulness of sleep (mean increase = 2.15 vs. 1.63; Cohen’s d = 0.33, p = 0.042).

Changes in difficulty falling asleep were modest and did not differ significantly between groups (mean decrease = 1.99 vs. 1.78; Cohen’s d = 0.07, p = 0.64).

Collectively, these findings suggest that while pulsed magnetic therapy did not significantly outperform placebo in reducing global PSQI scores, it produced small but statistically significant gains in perceived satisfaction and restfulness of sleep.

A prespecified subgroup analysis was performed among participants who reported pronounced complaints on the corresponding PSQI item at baseline. ANCOVA models adjusting for baseline severity showed that participants in this subgroup experienced larger absolute improvements across all outcomes compared with the full sample.

Reductions in PSQI global score were comparable between groups (marginal mean change: active 3.865; control 3.586), and the between-group difference was not statistically significant (p = 0.501). The estimated effect size was small (Cohen’s d = 0.07).

In contrast, significantly greater improvements were observed for subjective sleeper-centered outcomes. Participants receiving active treatment reported a larger increase in satisfaction with sleep (2.427 vs. 1.770; p = 0.022; Cohen’s d = 0.349) and a markedly larger increase in restfulness of sleep (3.012 vs. 2.078; p = 0.007; Cohen’s d = 0.472). These effect sizes fall within the small-to-moderate range and exceed those observed in the overall sample, indicating that participants with more severe baseline impairment may derive greater benefit.

For difficulty falling asleep, both groups demonstrated substantial improvements (3.323 vs. 3.013), with no significant between-group difference (p = 0.249) and a small effect size (Cohen’s d = 0.20) (Table 3).

Overall, in participants with pronounced baseline symptoms, pulsed magnetic therapy yielded significantly greater improvements in satisfaction and restfulness of sleep, while improvements in global PSQI score and sleep initiation difficulty remained similar to placebo.

Both treatment groups demonstrated significant improvements over time across all diary-based sleep parameters (Figure 3). The frequency of nocturnal awakenings decreased by an average of 0.65 events per night in the active group during the final study week and by 0.38 events per night in the control group. Despite the numerically greater reduction in the active arm, an ANCOVA adjusting for baseline values found no significant between-group difference (F = 0.663, p = 0.43). The significant time effect in both arms indicates that participants generally experienced fewer nighttime awakenings regardless of treatment conditions.

For duration of sleep interruptions, the active group showed a mean reduction of 4 minutes, compared with a 5.5-minute reduction in the control group. Baseline-adjusted ANCOVA revealed no significant group effect (F = 0.427, p = 0.514).

A comparable pattern was observed for sleep onset latency. Participants in the active group demonstrated a mean decrease of 12 minutes, whereas those in the control group improved by 11 minutes. The between-group comparison again did not reach statistical significance after baseline adjustment (F = 2.392, p = 0.124).

Overall, although sleep diary measures improved over the course of the study in both groups, none of the between-group differences were statistically significant, suggesting that the magnitude of change did not differ reliably between the active pulsed magnetic field intervention and control conditions.

Diary-based ratings of daytime functioning also improved over the course of the study.

Restoration by sleep increased by 0.65 points in the active group and 0.21 points in the control group (6-point Likert scale). Repeated-measures ANOVA confirmed significant time effects for both the active group (F = 3.18, p < 0.001) and the control group (F = 7.17, p < 0.001). Importantly, baseline-adjusted ANCOVA revealed a significant between-group (Figure 4) difference favoring the active intervention (F = 8.990, p = 0.003).

Daytime stamina improved by 0.66 points in the active group and 0.46 points in the control group. Time effects were significant in both groups (active: F = 18.853, p < 0.001; control: F = 6.559, p = 0.002). However, the between-group difference was not significant in baseline-adjusted ANCOVA (F = 1.330, p = 0.250).

Daytime exhaustion decreased (improved) by 0.54 points in the active group and 0.40 points in the control group. Time effects again reached significance in both groups (active: F = 20.04, p < 0.001; control: F = 5.517, p = 0.005), while ANCOVA showed no significant between-group difference (F = 1.71, p = 0.189).

Baseline and post-intervention survey measures demonstrated significant within-group improvements in both the active and control arms.

In the active group, paired-sample t-tests showed robust improvements across all three scales: WHO-5 well-being scores increased significantly (t(110) = 9.192, p < .001), MYMOP symptom scores decreased (t(102) = 7.426, p < .001), and Epworth Sleepiness Scale (ESS) scores also decreased (t(110) = 5.498, p < .001).

The control group likewise showed significant pre–post changes: WHO-5 scores improved (t(105) = 8.074, p < .001), MYMOP scores decreased (t(97) = 4.645, p < .001), and ESS scores decreased (t(105) = 6.015, p < .001).

Despite these substantial within-group improvements, baseline-adjusted between-group comparisons did not reveal significant differences. ANCOVA results indicated no significant treatment effect for WHO-5 (F = 1.091, p = 0.297), MYMOP (F = 1.739, p = 0.189), or ESS scores (F = 0.044, p = 0.835).

Overall, both groups demonstrated meaningful improvements in well-being, symptom burden, and daytime sleepiness across the study interval; however, the magnitude of change did not differ significantly between the active and control conditions (Figure 5).

Taken together, the daytime diary data indicate that participants in both groups experienced improvements in perceived restoration, stamina, and exhaustion across the intervention period. Among these variables, only the sense of restoration by sleep showed a statistically significant between-group difference, favoring the active pulsed magnetic field stimulation.

A total of 46 participants experienced 49 adverse events (AEs) during the study, with 21 AEs in the active group and 28 AEs in the control group. No serious adverse events (SAEs) occurred.

The most frequently reported AEs were upper respiratory infections, including colds (18 cases; 7 active, 11 control) and sinusitis (3 cases; 1 active, 2 control). Musculoskeletal complaints were also common (10 cases total), comprising joint and muscle pain reported in both groups. Allergic reactions occurred in 4 participants (2 per group). Gastrointestinal infections were reported only in the active group (2 cases).

A variety of isolated events were each reported once, including toothache, migraine, elevated blood pressure, stomach pain, endometriosis, influenza, dry skin, unspecified infection, increased urinary urgency, and dizziness. Overall, the distribution and nature of events were consistent with common, self-limiting conditions in the general adult population, with no notable imbalance between study arms.

One participant in the active treatment group experienced a dermatologic and systemic event considered of special interest due to its temporal proximity to device use. The participant reported the onset of a generalized pruritic rash beginning the day after the first application of the sleep program. Symptoms progressed over several days and were accompanied by nocturnal itching, malaise, cold-sweat sensations, and restlessness. The participant adjusted device intensity parameters but ultimately discontinued use.

The event was also associated with transient lymph node swelling, and symptoms gradually resolved following self-directed management. No clinical diagnosis or confirmatory medical assessment was obtained. The sponsor’s safety officer classified the event as possibly related to treatment, based solely on temporal association, with no evidence of a serious or lasting complication.

Across the study, three events were judged as having a possible relationship to treatment. Two involved a temporary worsening of sleep problems, one in each treatment arm, both of which improved after discontinuation. The dermatologic event described above represented the third case.