Post-stroke sleep disorders (PSSD) affect up to 78% of stroke survivors and are significant yet insufficiently recognized complications (1). These disorders impede functional recovery and rehabilitation after stroke, as sleep promotes neurogenesis and neural rewiring process in damaged brains (2, 3). Indeed, insomnia in acute stroke patients is associated with poorer outcomes, reduced quality of life, and an increased risk of secondary strokes (4). Despite their high prevalence and serious consequences, PSSD remains insufficiently addressed in standardized care guidelines, highlighting a critical gap in stroke management (5).

Based on the diagnostic tools, including the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) and Fourth Edition (DSM-IV) (6, 7), up to 32% of stroke patients are diagnosed with insomnia as part of PSSDs within the first few months of stroke onset, compared to a 6% prevalence in the general population (5, 8). This high prevalence rate denotes the need for early diagnosis and treatment of post-stroke insomnia in acute settings. However, existing treatments are primarily tailored to chronic insomnia and are unsuitable for the acute stroke context (9). For instance, Cognitive Behavioral Therapy for Insomnia (CBT-I), the gold standard for treating chronic insomnia, is effective when delivered in four to eight weekly sessions over a 4–12-week period and requires active participation in physiologically and mentally demanding techniques, including techniques such as stimulus control, sleep restriction, and cognitive restructuring (10). Although effective, this approach is not ideal for acute stroke patients due to its delayed onset of action and the high prevalence of stroke-related motor and cognitive impairments.

Pharmacotherapy is commonly used to manage insomnia due to its accessibility and rapid onset of action. Recommended drug classes for chronic insomnia include benzodiazepines, non-benzodiazepine hypnotics, and orexin receptor antagonists. However, acute insomnia in stroke patients remains largely unaddressed, with no specific pharmacological guidelines available (11, 12). Additionally, pharmacological options carry distinct risks for stroke patients. For instance, benzodiazepines, commonly prescribed for insomnia, are contraindicated because they can exacerbate breathing-related sleep disorders and motor deficits (8, 13). Other drug classes, while potentially usable, carry risks of oversedation, delirium, and hypotension, which can hinder rehabilitation and exacerbate stroke related impairments (14).

These limitations underscore the need for alternative interventions tailored to the unique challenges faced by stroke patients. Smartphone-based Digital Cognitive Behavioral Therapy (dCBT) offers a scalable and cost-effective solution with proven efficacy in managing chronic insomnia (15, 16). However, similar to traditional CBT, it requires extended sessions and cognitive engagement, typically involving six or more sessions over a minimum of 4 weeks to achieve significant results (17). Non-invasive brain stimulation techniques, such as Repetitive Transcranial Magnetic Stimulation (rTMS) and Transcranial Direct Current Stimulation (tDCS), have shown promise in enhancing sleep quality, yet they are constrained by the need for specialized equipment and operators, high costs, and the potential for discomfort and side effects (18–20).

Biofeedback therapy, which provides real-time feedback on physiological parameters such as breathing patterns and neural activity, offers a non-invasive approach to modulate the autonomic nervous system (ANS). This method has demonstrated effectiveness in improving sleep quality across diverse populations (21–24), yet it is limited by the requirement for bulky equipment, specialized expertise, and extensive patient training. Given these challenges, there is a pressing need for accessible and effective solutions tailored to stroke patients suffering from insomnia in acute care settings. A recent proof-of-concept study experimented with mobile-based bio-sonification therapy, a modified form of biofeedback, and found that its use in individuals with insomnia symptoms led to significant improvements in sleep onset and maintenance (25). This approach offers rapid relief while eliminating the need for specialized equipment, extensive patient training, and expert oversight, highlighting the potential of digital biofeedback therapy for insomnia treatment. This study aims to assess the feasibility and effectiveness of a smartphone-based biofeedback intervention for post-stroke insomnia refractory to conventional pharmacological treatments, using an on–off–on design to evaluate its impact on sleep quality (25).

Figure 3 illustrates K-RCSQ, ISI-K, and PSQI-K results plotted at each assessment point. The primary assessment, subjective sleep quality measured by the K-RCSQ, demonstrated clear on–off patterns across all three cases, with scores improving during intervention periods, declining during discontinuation, and improving again upon reintroduction of the intervention. Notably, Case 2 demonstrated a progression from “poor sleep” to “very good sleep,” while Case 3 exhibited a striking improvement from “very poor sleep” to “very good sleep” by the end of the trial.

Secondary assessments reflected similar improvements. ISI-K scores consistently declined across all three cases; by the end of both intervention periods, all patients had achieved the insomnia threshold defined in the validation study of ISI-K (ISI < 15.5 points) (29). Reductions exceeding the score of 6, determined as the MCID (31), were observed during both intervention phases in all three cases. Moreover, remission thresholds (ISI < 8 points) (32) were met by Cases 2 and 3 at the end of the first intervention period, and Case 3 achieved remission again after the shorter second intervention period. Similarly, PSQI-K scores demonstrated significant improvements across all three cases as Cases 2 and 3 achieved scores below the insomnia threshold at the end of both intervention periods. All patients exceeded the native version’s threshold for a clinically meaningful response (∆PSQI > 3 points) by the end of each intervention phase (33–35).

This case series provides preliminary evidence supporting the feasibility and effectiveness of a smartphone-based biofeedback intervention for managing insomnia in subacute stroke patients unresponsive to pharmacological treatment. Insomnia is a complex multifactorial disorder arising from a web of interconnected physiological mechanisms, with autonomic dysfunction recognized as one contributing factor (36–38). Although the exact mechanisms remain debated, a similar process is likely responsible for insomnia in stroke patients. Studies involving Heart Rate Variability (HRV) analysis and plasma norepinephrine levels indicate that acute cerebrovascular events disrupt the autonomic nervous system (ANS), characterized by heightened sympathetic (SNS) activity and reduced parasympathetic (PNS) activity (39–41). In particular, right-sided insular cortex lesions have been strongly associated with heightened SNS outflow, reduced PNS activity, and lower HRV, suggesting a potential link between stroke-related autonomic dysfunction and insomnia (42–44).

Building on this understanding of ANS dysfunction in insomnia patients, the biofeedback application used in this study offers an innovative approach to addressing insomnia by potentially modulating autonomic function. It is hypothesized to restore autonomic balance through mechanisms distinct from conventional treatments, including increasing the magnitude of respiratory sinus arrhythmia (RSA) via controlled breathing and stimulating slowly adapting pulmonary stretch receptors through regulated lung inflation (45, 46). These processes may synergistically activate the vagus nerve, enhance parasympathetic activity, suppress sympathetic activity, and restore ANS homeostasis, which in turn could contribute to improving post-stroke insomnia (47–49). This proposed mechanism is partially supported by prior studies that have demonstrated efficacy of vagus nerve stimulation (VNS) in treating insomnia and autonomic dysfunction (50, 51). However, VNS’s invasive nature and reliance on external electrical stimuli limit its applicability. In contrast, this biofeedback application may offer a non-invasive alternative that replicates the benefits of VNS.

The results of this study further validate the proposed mechanism, demonstrating its effectiveness in patients previously unresponsive or minimally responsive to pharmacotherapy. Unlike traditional treatments, the biofeedback application is hypothesized to stimulate the vagus nerve and may help circumvent some of the challenges associated with impaired monoaminergic and GABAergic systems often seen in stroke patients (52–54). Improvements across primary and secondary assessments underscore its efficacy in addressing post-stroke insomnia. Notably, categorical improvements in K-RCSQ scores for Cases 2 and 3, including progression to higher sleep quality categories, highlight its potential to alleviate significant symptoms. Marked reductions in ISI scores, with all three cases exceeding the MCID and Cases 2 and 3 reaching remission thresholds, along with substantial PSQI improvements surpassing the MCID for all three patients, further support its therapeutic benefits. The on–off design of the study suggests a potential causal relationship, as improvements appeared to align consistently with the application’s use and declined during the washout period. However, as this was a proof-of-concept study involving a small number of participants, the strength of this inference is limited. Furthermore, because no control group was incorporated, the influence of uncontrolled confounding factors such as natural stroke recovery, concurrent medications, and placebo effects cannot be completely ruled out.

This study highlights several advantages of biofeedback interventions for insomnia in PSSD, with safety being the primary benefit. Unlike pharmacological therapies, which carry risks of adverse effects that may impair cognition and motor control in stroke recovery patients, the biofeedback approach primarily relies on internal physiological adjustments. By avoiding the introduction of external chemicals or stimulants, this method minimizes the risk of disrupting the delicate neurochemical and neuronal balance, thereby ensuring a superior safety profile. Notably, no common side effects typically associated with conventional pharmacotherapy treatments - such as dizziness, oversedation, or headaches (55)—were observed during this trial in all three cases, further supporting the intervention’s favorable safety profile. Additionally, the first case subject, who had previously experienced dizziness as an adverse effect of zolpidem tartrate, did not report similar or new symptoms during the intervention phase. Such finding indicates that the intervention is well-tolerated, even in individuals with heightened sensitivity to drug-related adverse effects.

Moreover, the intervention demonstrated a rapid onset of effects, with noticeable benefits observed during the first treatment window, likely reflecting its direct modulation of the ANS function. This rapid onset of effect is particularly significant, as quality sleep during the early stages of rehabilitation is closely linked to an improved quality of life (56, 57). The ability to achieve significant results within a short time frame sets this intervention apart from therapies such as CBT or dCBT, which require extended patient engagement and cognitive restructuring over weeks or months to yield comparable benefits. By addressing sleep disturbances promptly, the intervention used in this study has the potential to improve both short-term and long-term recovery trajectories for stroke patients (58).

The application’s straightforward design, requiring only a smartphone, eliminates the need for specialized equipment or trained personnel. This simplicity enhances its feasibility in clinical settings and its potential for home use. Additionally, caregivers found it effective in helping patients sleep without overnight delirium, significantly reducing their workload and the need for constant monitoring. Primary physicians also noted that the application required no supervision, allowing them to focus on other tasks, while the reduction in shift calls minimized overnight interventions. Other patients in the same ward also experienced improved sleep quality, as the absence of disruptions caused by case patients created a more conducive sleep environment. These findings demonstrate that the application benefits not only individual users but also caregivers, healthcare staff, and other patients in shared environments.

This study has several limitations that should be addressed in future research. First, the findings are based on a small sample size of three subacute stroke inpatients. To confirm the application’s efficacy, larger and more diverse patient populations, including outpatients and those with chronic PSSD, should be studied within the framework of RCTs. Additionally, this study only evaluated short-term intervention use. Future research should investigate the effects of long-term application use to determine if the benefits persist over extended periods. Lastly, the study did not incorporate objective sleep measures such as actigraphy or polysomnography.

This case report underscores the potential of a smartphone-based biofeedback application as an effective and safe intervention for managing insomnia in PSSD patients. Future research should focus on validating the application’s efficacy through RCT with placebo comparisons, determining the optimal duration of intervention, and evaluating the long-term sustainability of its therapeutic benefits.