Amygdala and Insula Retraining (AIR) Significantly Reduces Fatigue and Increases Energy in People with Long COVID

Long COVID affects approximately 10–30% of individuals after an acute COVID-19 infection (Ceban, Ling, et al. 2022; Ortona and Malorni, 2022). Numerous symptoms, including extreme fatigue, can persist for months, resulting in social and economic hardship for individuals and their families (Ortona and Malorni 2022). Therefore, approaches that offer some relief from Long COVID are urgently needed. Research suggests that Long COVID symptoms are akin to those of chronic conditions, such as myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS) and are likely caused by inflammation and immune dysfunction (Scordo et al., 2021). Amygdala and Insula Retraining (AIR), a neuroplasticity program, has successfully alleviated chronic conditions (Gupta 2010; Sanabria-Mazo et al. 2020; Toussaint et al. 2012). In this randomized controlled trial, AIR was tested against a structurally equivalent health and wellness intervention for its effectiveness in treating the symptom of fatigue among Long COVID sufferers. Results showed a significant decrease in participants' fatigue and a significant increase in their energy after the 3-month AIR intervention. Additionally, the AIR group experienced more significant outcomes than the active control group. The AIR group demonstrated a fatigue reduction effect size four times that of the active control group, and the absolute reduction in mean scores for the AIR group was more than double that of the control group. Furthermore, the AIR group showed an effect size in energy enhancement twice that of the active control group, and the absolute increase in energy mean scores for the AIR group was almost double that of the control group. These novel findings suggest AIR is a viable means of reducing fatigue and increasing energy among Long COVID patients. Limitations and future research are discussed.


Introduction
Long COVID refers to a range of 200 possible symptoms that can persist for several months [1] for roughly 10-30% of individuals after their acute COVID-19 infection has been resolved [1,2]. Symptoms include extreme fatigue, myalgia, and cognitive impairment [3], and efective means of coping with Long COVID are urgently needed. Pharmaceutical, biological, dietary, homeopathic, and rehabilitative remedies have been tested, yet help for Long COVID remains elusive [4].
Long COVID symptoms are likely caused by chronic infammation and immunological dysregulation in the body [1]. Tey are similar to those of myalgic encephalomyelitis/ chronic fatigue syndrome (ME/CFS), a chronic disease characterized by immune and cognitive dysfunction [5]. Terefore, it is possible that treatments for ME/CFS could be efective in resolving Long COVID [5]. An amygdala and insula retraining (AIR) intervention, known as Te Gupta Program, has been successful in alleviating symptoms of ME/CFS [6,7] and other chronic conditions, like fbromyalgia [7,8]. Tus, the current study aimed to test the efectiveness of the AIR intervention for Long COVID.
Te AIR intervention is hypothesized to strengthen neurological inhibitory mechanisms in the prefrontal cortex, insula, amygdala, and anterior and posterior cingulate. Tis process helps reduce the magnifcation of incoming bodily signals and the hyper-stimulation of the autonomic nervous and immune systems by the amygdala and the insula, thereby allowing the systems to return to a normal state of balance [6]. Te AIR intervention primarily involves specialized neuroplasticity techniques to retrain the immune and nervous systems' hyperactivity.
Te study aimed to test the AIR program's efectiveness in reducing prolonged symptoms of fatigue following COVID-19 infection. Additionally, it aimed to evaluate the efectiveness of AIR compared to a general health intervention. To our knowledge, no research has yet examined the impact of a neuroplasticity program in remedying Long COVID symptoms. Consequently, this novel study attempted to generate potential insights that could inform future treatment. Te following hypotheses were explored: (H1) Tere will be a signifcant decrease in participants' fatigue after the AIR intervention (H2) Compared to an active control group, the experimental group will experience a signifcant decrease in participants' fatigue after the AIR intervention  ), and Long COVID Support Group. Te sample comprised n � 86 women (86%), n � 12 men (12%), n � 1 nonbinary, and n � 1 who declined to answer. Te mean age was 43.6 years. One-ffth (n � 20) had less than a college education, n � 43 had a bachelor's degree, n � 32 had a master's degree, and n � 5 held a doctorate. Half of the participants (n � 50) were randomly assigned to the AIR intervention and the other half (n � 50) to the active control group using an Excel function that generated a list of random assignments following enrollment in the study.

Materials and Methods
A series of chi-square tests and an independent t-test indicated that the AIR intervention and active control groups did not difer signifcantly on any of the demographic variables: gender [χ 2 (2) � 0.99, p � 0.610], age [t (98) � − 0.053, p � 0.957 (two-tailed)], and education [χ 2 (7) � 8.70, p � 0.275]. Additionally, a series of chi-square tests and an independent t-test were performed to compare the demographic variables of participants who completed the full 3-month study (n � 40) versus those who dropped out at some point (n � 60). No signifcant diference was found between groups for gender [χ 2 (2) � 4.00, p � 0.136] and education [χ 2 (7) � 6.52, p � 0.481]. However, there was a signifcant diference in age across the two groups [t (98) � −2.46, p � 0.016 (two-tailed)]. Te mean age of those who dropped out was 41.6, while the mean age of those who completed the study was 46.3.

2.2.
Procedure. Participants were randomly assigned to the AIR intervention or an active control group. Te AIR intervention group attended an online workshop to introduce participants to 6+---------------------------the educational neuroplasticity program of Amygdala and Insula Retraining (AIR) and received supporting materials online and by mail. Te main AIR practices include specialized neuroplasticity techniques. Te primary process is designed to interrupt adverse somatic signals and mental patterns and direct individuals to create new, positive neural pathways that indicate safety to the brain through repetition. Secondary activities support the neuroplasticity techniques. Tese include mindfulness-based meditation, where individuals focus on the present; alternate nostril breathing, where individuals hold one nostril shut while breathing through the other, then change nostrils and repeat; and other lifestyle therapies, such as suggested general health supplements and cultivating a calming morning ritual that eases people into their day.
Participants were asked to practice the AIR intervention for 40-60 minutes daily, including the main neuroplasticity processes, a few minutes of alternate nostril breathing, and a simple 20-minute mindfulness meditation practice. In addition, throughout the day, they were asked to practice abbreviated versions of the neuroplasticity techniques to interrupt somatic signals further and retrain the Amygdala and Insula's hypothesized responses. Tese retraining techniques took about 30-60 seconds to enact each time. Te participants received weekly webinars to support their training and the opportunity to ask questions within the webinars. Finally, participants had timely access to the study investigators for any immediate questions or challenges that arose with the practice and were ofered optional online support with a coach trained in AIR.
Te active control group received an online educational program for general health and well-being (12 weeks to wellness). It involved general advice on diet, exercise, energy, nutrition, sleep, and other lifestyle interventions. Te group was encouraged to live an overall healthier lifestyle and to put the knowledge gained from the program into practice. Participants in this program learned about eating healthy foods, managing stress, and developing an active work-life balance. In addition to making behavioral changes, the program emphasized the importance of individuals shifting their internal attitudes and beliefs about wellness. Participants attended weekly webinars, were provided with online resources, and were ofered optional online support with a coach trained in the 12 weeks to wellness program. Evidence-Based Complementary and Alternative Medicine Te AIR and active control interventions were free for participants, and no incentives were used to encourage participation. Te control group was also ofered complimentary access to AIR after the 3-month study. Tere were four points of measurement: baseline (before the intervention), 1 month, 2 months, and 3 months (upon completion of the intervention). Figure 1 displays the fowchart of participants and sample sizes for the groups at each point during the study. Te fnal number of participants in month 3 was N � 42. However, two respondents did not complete the measures at all four time intervals. Terefore, the overall retention rate for full completion of the study (i.e., participants completed all four measures), from baseline (N � 100) to month 3 (N � 40), was 40%. Tis retention rate for an online intervention is higher than typically observed [9]. Notably, the retention rate for the AIR group was higher (n � 23; 46%) than that for the control group (n � 17; 34%).

Multidimensional Fatigue Inventory.
Fatigue was measured using the multidimensional fatigue inventory (MFI) [10]. It is a 20-item scale that assesses fve dimensions of fatigue (four items per dimension): general fatigue, physical fatigue, reduced motivation, reduced activity, and mental fatigue. Te MFI scale developers recommended using the general fatigue subscale as a global fatigue index because of its sensitivity to changes in fatigue level [10,11]. Participants respond to each item using a 5-point Likert-type scale that ranges from the lowest score of 1, which represents a response of yes that is true, to the highest score of 5, which represents a response of no that is not true. Higher total scores indicate greater levels of fatigue. Te developers of the scale reported internal reliability ranging from α � 0.53 to α � 0.93 [10]. In the present study, the internal reliability score for general fatigue was acceptable across most points of measurement (baseline: α � 0.54; month 1: α � 0.72; month 2: α � 0.90; month 3: α � 0.86).

Short Form Health Survey (SF-36).
Te SF-36 questionnaire [12][13][14] measured several quality-of-life factors. Te 36-item scale assesses eight dimensions (ranging from two to 10 items per dimension): physical functioning, role limitations due to physical health, role limitations due to emotional problems, energy, emotional well-being, social functioning, pain, and general health. Te energy subscale was of most interest in the present study. Items include binary yes/no responses and some Likert-type scales that vary from lowest scores representing responses such as poor, much worse, and limited a lot and highest scores representing responses such as excellent, much better, and not limited at all. Higher total scores indicate a more favorable health status. Validation of the scale reported internal reliability of α � 0.85 [15]. In the present study, the internal reliability score for energy was acceptable across all measurement points (baseline: α � 0.73; month 1: α � 0.87; month 2: α � 0.90; month 3: α � 0.91).

Analyses.
A series of two-way mixed ANOVAs were used to analyze the data in IBM SPSS v26.0. Given the study's emphasis on reducing symptoms of fatigue, the analysis focused only on the fatigue-related dimensions of each scale. As noted above, N � 100 participants completed the baseline measure (n � 50 AIR; n � 50 control), but only N � 40 completed all four measures (n � 23 AIR; n � 17 control). Terefore, expectation maximization, last-observation-carried-forward, and listwise methods for missing data were used to ensure unbiased statistical estimates.
Te data were normal, with skewness and kurtosis scores within the ±2 acceptable ranges. Other assumptions required for a two-way mixed ANOVA were met, except for sphericity, which was violated for both subscales. Terefore, the Greenhouse-Geisser correction was used in reporting. Additionally, boxplots indicated three outliers in the MFI general fatigue data and one outlier in the SF-36 energy data. However, none of the outliers were extreme or considered Evidence-Based Complementary and Alternative Medicine aberrant in the data set. Terefore, all outliers were retained for analysis. When running the two-way mixed ANOVAs, tests of within-subjects were consulted frst to establish if there was a signifcant interaction between time and group. Subsequently, diferences between groups at each time interval were examined, as well as changes within each group over time. Te Bonferroni correction was used for pairwise comparisons, except where otherwise noted.

General Fatigue (MFI Scale).
Analysis of the general fatigue subscale from the MFI scale using expectation maximization (EM) methods revealed that the groups changed signifcantly over time with a medium efect size: F (1.95, 191.10) � 7.04, p � 0.001, partial η 2 � 0.067. As shown in Table 1, there was no signifcant diference in general fatigue between the groups at baseline. However, the AIR group posted lower fatigue scores than the control group for months 1-3. Te diference between the AIR and active control groups at month 1 was signifcant with a medium efect size. It was also signifcant at month 2 with a small efect size and approaching a signifcant diference at month 3 with a small efect size. Tus, general fatigue mean scores were signifcantly lower for the AIR group than the active control group at months 1 and 2 and approached a signifcantly lower mean score at month 3.
Notably, as shown in Table 1, the total decrease in mean scores for general fatigue from baseline to month 3 for the AIR group (3.38) was more than double that of the active control group (1.40). Moreover, research validating the MFI scale found the mean score for the general fatigue subscale among US adults was 12.90 [16]. Terefore, participants using AIR more readily approached normal levels of general fatigue by month 3.
Over time, results for the active control group showed a signifcant reduction in general fatigue with a medium efect size, F (1.95, 95.54) � 4.72, p � 0.012, and partial η 2 � 0.088. Pairwise comparisons showed that this signifcant reduction in mean fatigue scores occurred between baseline and month 3 (p � 0.037), indicating that fatigue signifcantly declined for active control group participants by the end of their 12-week wellness program.
Over time, results for the AIR group were also signifcant and to a greater extent than those of the active control group. Te AIR group demonstrated a very large efect size that was more than four times the control group efect size: F (1.93, 94.43) � 33.85, p < 0.001, partial η 2 � 0.409. Pairwise comparisons showed a signifcant reduction in mean fatigue scores occurring between baseline and all other time points (p < 0.001), indicating that compared to the baseline, fatigue was signifcantly reduced at every time point (months 1, 2, and 3) for AIR participants. Figure 2 shows the estimated marginal means plot for general fatigue across time and group. Tough both groups posted reduced fatigue scores at each time interval, it is notable that the AIR group experienced a marked decline in just one month compared to the active control group.
Te same analyses were performed to verify fndings, using the last observed value carried forward (LOCF) method for missing data (N � 100) and among only the respondents who completed all the study measures at each time interval (i.e., N � 40). Both approaches returned nearly identical results. Te LOCF analysis indicated the groups changed signifcantly over time with a small efect size, F (1.81, 177.41) � 5.24, p � 0.008, and partial η 2 � 0.051. Tere was no signifcant diference between groups at any time interval (baseline: p � 0.286, month 1: p � 0.116, month 2: p � 0.167, and month 3: p � 0.115). However, both groups experienced a signifcant decline in fatigue over time, with the AIR group posting a more signifcant result and a larger efect size. Active control: F (1.91, 93.73) � 4.37, p � 0.017, and partial η 2 � 0.082. AIR group: F (1.72, 84.48) � 17.15, p < 0.001, and partial η 2 � 0.259. Pairwise comparisons indicated the active control group had a signifcant decrease in fatigue between baseline and month 3 (p � 0.040). In contrast, the AIR group experienced a signifcant decrease between the baseline and all time points (p < 0.001).
Te analysis among the 40 respondents who completed all the study measures at each interval indicated that the change in groups over time approached statistical signifcance, F (1.93, 73.48) � 3.10, p � 0.053, and partial η 2 � 0.075. Tere was no signifcant diference between groups at any time interval (baseline: p � 0.703, month 1: p � 0.117, month 2: p � 0.132, and month 3: p � 0.146). Over time, the change in fatigue in the active control group was nonsignifcant, F (1.97, 31.52) � 2.37, p � 0.111, and partial η 2 � 0.129. However, the AIR group experienced a signifcant decrease in fatigue over time with a very large Tus, these additional analyses provided further evidence that AIR group participants benefted from a greater reduction in fatigue than the active control group over time.

Energy/Fatigue (SF-36 Scale).
Analysis for the energy subscale from the SF-36 scale using expectation maximization (EM) methods revealed that the change in groups over time approached statistical signifcance: F (2.41, 236.09) � 2.38, p � 0.084, and partial η 2 � 0.024. However, no signifcant diference was found between groups at any time interval, as shown in Table 2. Results for both groups over time indicated a signifcant diference in energy scores: F (2.41, 236.09) � 31.26, p < 0.001, partial η 2 � 0.242. Te active control group demonstrated a signifcant increase in energy with a large efect size, F (2.44, 119.57) � 9.63, p < 0.001, partial η 2 � 0.164. Pairwise comparisons showed a signifcant increase in energy scores between baseline and month 1 (p < 0.001), baseline and month 2 (p � 0.004), and baseline and month 3 (p � 0.002), suggesting that compared to the baseline, energy signifcantly increased at every time point (months 1, 2, and 3) for active control participants.
Te AIR group also experienced a signifcant increase in energy over time, with a large efect size double that of the active control group: F (2.22, 108.79) � 24.23, p < 0.001, partial η 2 � 0.331. Pairwise comparisons indicated a signifcant increase in energy between baseline and all other time points (p < 0.001), suggesting that AIR group participants experienced an increase in energy at all intervals compared to the baseline. In addition, AIR participants experienced a signifcant increase in energy between months 1 and 3 (p � 0.007) and approached a signifcant increase between months 1 and 2 (p � 0.055). Figure 3 shows the estimated marginal means plot for energy across time and group. Tough the diference between groups was nonsignifcant at all time points, it is worth noting that the AIR group started at a marginally lower mean score at baseline and climbed to a higher fnal energy level at month 3. Indeed, as shown in Table 2, the absolute increase in mean scores for energy for the AIR group (18.26) was almost twice that of the active control group (10.40).
Te same analyses were performed using the LOCF method for missing data (N � 100) and among only the respondents who completed all the study measures at each time interval (i.e., N � 40). Te results for those who completed all measures were virtually identical to the EM imputation method fndings. However, using the LOCF method yielded more signifcant fndings. Te groups changed signifcantly over time with a small efect size: F (2.10, 205.48) � 3.65, p � 0.026, partial η 2 � 0.036. Still, there were no signifcant diferences between groups at any time interval (baseline: p � 0.274, month 1: p � 0.613, month 2: p � 0.305, month 3: p � 0.452).
Results for the active control group over time showed a signifcant increase in energy with a medium efect size: F (1.93, 94.45) � 4.26, p � 0.018, and partial η 2 � 0.080. Pairwise comparisons using the Bonferroni correction showed no signifcant result between time points, perhaps because of the overly conservative nature of the method [17]. However, with the Least Signifcant Diference (LSD) adjustment, there were signifcant increases between the baseline and all other points (month 1: p � 0.020, month 2: p � 0.021, and month 3: p � 0.020), suggesting that control group participants' energy improved across time. Results for the AIR group over time were more signifcant with a large efect size: F (2.20, 107.64) � 16.90, p < 0.001, and partial η 2 � 0.256. Pairwise comparisons showed a signifcant increase in mean energy scores between baseline and all other time points (p < 0.001), with both the Bonferroni and LSD adjustments, indicating that energy was signifcantly enhanced at month 1, month 2, and month 3 for AIR participants.

Discussion
Te aim of this study was twofold. First, to evaluate the AIR intervention's impact in reducing prolonged symptoms of fatigue from Long COVID. Second, to test the efectiveness of AIR compared to a structurally equivalent general health and wellness intervention. Results supported both hypotheses. Tere was a signifcant decrease in participants' fatigue after the 3-month AIR intervention, and compared to an active control group, the AIR group experienced a more signifcant decrease in participants' fatigue. For general fatigue, the efect size of the reduction in the AIR group was four times that of the active control group. Additionally, the absolute decrease in fatigue mean scores for the AIR group was more than double that experienced by the control group.

Evidence-Based Complementary and Alternative Medicine
Furthermore, the fnal general fatigue mean score for the AIR group approached standardized healthy norms [16]. Similarly, the AIR group also demonstrated a signifcant increase in energy over time, with a large efect size that was double that of the active control group, and the absolute increase in energy mean scores for the AIR group was almost double that of the active control group. Tese fndings provide initial evidence that AIR is a viable and efective approach to mitigating fatigue from Long COVID.
Te results may be explained by the AIR intervention hypothesis, namely, that the sympathetic nervous and immune systems are triggered into an overactive and hypervigilant state following a physical illness or stressor, such as COVID-19. Tis overactivity of the nervous and immune systems manifests as various bodily symptoms, such as prolonged fatigue [6,18]. Te AIR intervention employs neuroplasticity techniques designed to create new neural pathways and retrain the amygdala and insula so the immune and autonomic nervous systems can return to homeostasis. Indeed, other research has supported this theory of immune conditioning in the insula [19], reinforcing the concept of using neural retraining to reduce stimulation in the nervous and immune systems.
Finally, results from the present study support the hypothesis that interventions successful in treating chronic conditions, such as ME/CFS, might be transferrable to addressing Long COVID symptoms because such ailments are likely caused by infammation and immune dysfunction in the body [1,5]. Certainly, the current results are consistent with prior research that demonstrated the efectiveness of AIR in alleviating chronic conditions. For example, in combination with treatment-as-usual, AIR efectively reduced symptoms of fbromyalgia, including pain, fatigue, and depressive symptoms [8]. Similarly, another study found that AIR helped address ME/CFS symptoms such as energy, pain, and fatigue [7]. Tus, the efectiveness of AIR in alleviating chronic conditions seems to extend at least to reducing the symptom of fatigue in Long COVID.

Strengths and Limitations.
Tis study had several strengths. Te primary strength is the novelty of this research. To our knowledge, this is the frst study demonstrating the efectiveness of a neuroplasticity program (AIR) in reducing the Long COVID symptom of prolonged fatigue. Second, an experimental research design was used to identify cause-and-efect outcomes. Finally, the research design included an active control group rather than a wait-list control group. Tus, the AIR intervention was tested against a structurally equivalent intervention intended to improve health and wellness.
However, the study was not without various limitations. Tese included a convenience sample, a relatively small sample size that may have reduced statistical power, selfreported measures, and a study design that may have impacted results. Te sample was recruited from Long COVID support groups on social media, which relied on participants' self-report of Long COVID symptoms for 3 months or more. Tus, it is possible some participants believed they had Long COVID when they did not. However, the study was conducted in the frst half of 2021, when the health community's diagnosis and understanding of Long COVID were nascent. Terefore, by necessity, recruitment heavily depended on participant self-diagnosis, and it is unclear if a more stringent verifcation method could have been used in the circumstances. Moreover, participants were not asked  if they used other therapies for their Long COVID condition. Instead, the exclusion criteria served as a proxy to eliminate the possibility of including participants with comorbidities, taking prescription drugs, or engaging in other treatments. Still, any potential confounds created by these limitations of diagnosis or additional therapies would have been mitigated by randomizing participants to the AIR or active control groups and would not have prevented the detection of the intervention impact.
Te study design also presented a limitation by only including an active control group and not a passive control group. Te active control group provided a robust test for the efectiveness of AIR. However, with no passive control group, it is impossible to know how much more efective AIR would be compared to Long COVID symptoms following their natural course over 3 months. Furthermore, the study design did not allow for a followup measure to assess the longer-term impacts of AIR or the active control treatment on the symptom of fatigue. Finally, the study relied on self-reported measures rather than observations by health professionals. Arguably, the evaluation of fatigue is best reported by the self. Still, including objective biomarkers, such as oxidative stress and antioxidative activity, could have enhanced the investigation.
Future research should address these limitations and explore new areas of investigation. Forthcoming studies should recruit participants from well-established Long COVID clinics, incorporate objective biomarkers and observations from health professionals, and include a waiting control condition. Other suggestions include measuring additional symptoms of Long COVID and conducting larger studies to further knowledge of AIR as an approach to easing fatigue in people living with Long COVID.

Conclusion
Tis novel study was the frst to demonstrate the efectiveness of a neuroplasticity program (AIR) in addressing the Long COVID symptom of prolonged fatigue. Participants who received the AIR intervention experienced a signifcant reduction in fatigue and an increase in energy. Moreover, the AIR intervention yielded more signifcant fatigue reduction and energy enhancement results than an active control group. Te efect size of AIR's impact on fatigue was four times greater than that of the active control group, and the absolute reduction in fatigue scores for the AIR cohort was more than twice that of the control. Similarly, the efect size of AIR's impact on energy was double that of the active control group, and the absolute increase in energy scores for the AIR group was almost twice that of the control group. Tese fndings are both timely and pertinent, as so little is known about how to treat Long COVID and so many patients sufer from it after the acute infection of COVID-19. Te present research suggests that AIR-a low-cost and widely available intervention-could help alleviate the common symptom of prolonged fatigue in Long COVID.

Data Availability
Te data are not publicly available because the sample consists of vulnerable adults with a chronic medical condition. Individuals did not agree in the consenting process to have their data shared and expect their data to be kept strictly confdential.

Ethical Approval
Te study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board (or Ethics Committee) of Luther College (protocol code 30.20 and October 22, 2020).

Consent
Informed consent was obtained from all subjects involved in the study.

Conflicts of Interest
Alexandra J. Bratty is the CEO of AB Research Consulting, which provides consulting services to Te Gupta Program, the commercial version of the Amygdala and Insula Retraining (AIR) intervention. Her company was compensated for this work by independent donors. She was not involved in data collection for this study.