A study applying time series analysis to examine the positive effects of music therapy on sleep quality
Publié en ligne: 17 mars 2025
Reçu: 13 oct. 2024
Accepté: 26 janv. 2025
DOI: https://doi.org/10.2478/amns-2025-0342
Mots clés
© 2025 Yupeng He, published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.
Insomnia is the most common sleep problem encountered in the clinic. With the rapid technological advancement, the shortening of the cycle of updating knowledge and skills, the tightening of the beat of life, and the increasing external and self-imposed pressures faced by human beings, the number of patients with insomnia as the main complaint has shown an upward trend [1-2]. As a subjective experience, insomnia usually refers to complaints about dissatisfaction with sleep duration and/or quality when the patient has sufficient sleep opportunities, and even affects social functioning during the day, mainly in the form of being ready to fall asleep, but unable to fall asleep within 30 minutes, after falling asleep, waking up no less than twice in the middle of the journey, waking up prematurely in the morning, and maintaining sleep for less than 6 hours, which not only decreases in quality and quantity, but is also accompanied by daytime dysfunction [3-5].
As an emerging cross-cutting applied discipline, modern music therapy was established in the 1840s, while modern music therapy in China started in the 1880s [6-7]. There are many versions of the definition of music therapy, and the most widely used is the one proposed by Prof. K. Bruscia, which can be briefly summarized as a systematic process of motivating patients to achieve health goals through music and the relationships established in therapy [8-9]. There are many classifications of music therapy, which are most often categorized as receptive music therapy, recreational music therapy, and improvisational music therapy. Of these, receptive music therapy is the most widely used in various fields, such as the common technique of listening to songs [10-11]. The fusion of music therapy and China’s unique traditional Chinese medicine culture has also produced the characteristic five-tone therapy and music electrotherapy [12]. At present, music therapy has been used in the fields of insomnia, schizophrenia, various kinds of pain, anxiety, depression, hypertension, coronary heart disease, cancer, obstetric labor and delivery, post-stroke rehabilitation, Alzheimer’s disease, autism, and special education for mentally retarded children [13-14].
In today’s society, sleep disorders have become a more common problem, and the decline in sleep quality will have a double adverse effect on the physiological and psychological health of the human body, and it is urgent to improve the sleep quality of the social population. Currently, 2 types of treatments are often used, namely pharmacological treatment and non-pharmacological treatment [15-17]. And music therapy, as a non-pharmacological intervention to improve sleep quality, is gradually accepted and recognized by the society [18].
In this paper, patients with sleep disorders are treated with somatosensory sound wave physiotherapy technology, along with psychological intervention techniques, to improve the patient’s insomnia effect. The EEG signals of the human body during sleep are extracted through polysomnography, and the EEG signals are analyzed in the frequency domain using the Fast Fourier Transform algorithm. The multichannel sleep timing model is built, and the PLV is employed to create the adjacency matrix of polysomnograms to extract the characteristics of the graph structure within it. TCN is used to capture the dependency of the long time series, and local and global features are fully explored to comprehensively understand the spatio-temporal characteristics of EEG data and detect the sleep of the human body. Through the comparison experiment before and after the intervention, the sleep quality of the subjects was assessed, and the effect of music therapy on their sleep quality was studied.
Sleep disorders refer to abnormalities in the quantity and quality of sleep or the occurrence of certain clinical symptoms during sleep, such as reduced or excessive sleep, sleep walking disorder, etc., of which insomnia is the most common [19].
1) Insomnia: It includes psychophysiological insomnia, ambivalent insomnia, affective disorder-related sleep disorder, panic disorder-related sleep disorder, post-traumatic stress disorder, poor sleep hygiene, insomnia caused by drug or substance abuse, insomnia caused by medical diseases, and adaptive sleep disorder.
2) Breathing-related sleep disorders: including obstructive sleep apnea syndrome and central sleep apnea syndrome.
3) Central hypersomnia: including episodic sleeping sickness, sleep deprivation syndrome due to action problems, etc.
4) Circadian rhythm disorder sleep disorders: including sleep phase advancement syndrome, sleep phase delay syndrome, etc.
5) Heteromorphic sleep: including repeated arc sleep paralysis, nightmare, intracranial explosive sound syndrome, jet lag change syndrome, shift work sleep disorder, etc.
6) Movement-related sleep disorders: restless legs syndrome, periodic limb movement disorder, nocturnal lower limb painful spasms, etc.
In music therapy, there are two main therapeutic ideas, resource orientation and problem orientation [20].
Problem orientation refers to intervening in response to specific problems of the visitor, which also coincides with traditional medical and psychological treatment ideas. Traditional psychoanalytic psychotherapy is committed to finding the causes of the visitor’s psychological problems, such as through hypnosis, free association, and other ways to explore the visitor’s subconscious level of conflict. Behaviorist psychology, cognitive psychology, and other schools of psychology also use problem-oriented therapy mainly.
Resource-oriented music therapy pays more attention to the positive forces originating from the visitors themselves, such as the sense of achievement, self-confidence, good memories, etc. This therapeutic idea originates from the humanistic school of psychology. In our practice, we have found that many visitors lack basic self-confidence due to long-term anxiety, depression, low self-esteem and other negative emotions, so they lack sufficient “psychological energy” to solve their own problems. Resource-based psychological interventions aim to stimulate the positive experience of the visitor’s heart, stimulate positive potential, and help them recover their self-confidence and ability to solve their own problems. In music therapy, some methods are problem-oriented, such as desensitization to music systems. Some methods are resource-oriented, such as guided musical imagery, safe island techniques, targeted and non-targeted positive resource reinforcement, and so on. There are also methods that incorporate both problem-oriented and resource-oriented thinking, such as song discussion, improvisational music therapy, and music-guided imagery.
In psychological intervention, when the therapist is not sure whether the visitor’s heart is stable enough, the use of problem-oriented approach is a certain risk, it is easy to cause secondary trauma to the visitor, the use of resource-oriented approach to psychological intervention helps to stabilize the visitor’s mood, in order to prepare for further treatment. The same applies to music therapy. In practice, it has been found that when positive resource-oriented music therapy is used to help stabilize the mood of visitors, some of them say that they do not need further treatment. This is because they have recovered their ability to solve their own problems in the process.
Somatosensory sound wave physical therapy technology, is through the patented audio transducer technology and specific resonance music, the music in the 16 ~ 150 Hz low-frequency signals, through the amplification and physical transducer, converted into a precise sound vibration, through the human body “bone conduction” and other roles, so that a person produces a rapid depth of relaxation can effectively improve insomnia, anxiety, tension and other physical and mental symptoms. This can effectively improve insomnia, anxiety, tension, and other physical and mental symptoms. This study can to a certain extent promote the development and popularization of the clinical application of somatosensory sound wave technology in insomnia in China. At present, in insomnia, the use of somatosensory sound wave technology music therapy (somatosensory music) means more and more with other drugs or traditional Chinese medicine, together with the combination of experimental clinical research, and has achieved certain results. For example, somatosensory music therapy with Chinese medicine acupuncture method for the treatment of primary insomnia will have better efficacy. Somatosensory sound wave technology can also be used in conjunction with psychological interventions to help individuals suffering from mental illnesses.
Polysomnography (PSG) is a method used to detect the sleep state of human beings, which records a series of physiological indicators including electroencephalography (EEG), electrooculography (EOG), electromyography of the jaw (EMG), electrocardiography (ECG), respiratory flow, and so on [21]. In polysomnographic data, each physiological indicator changes over time and can be considered as a time series. The sleep time series referred to in this paper is either one signal or multiple signals in polysomnography. One signal constitutes a one-dimensional time series, while multiple signals constitute a multidimensional time series. Under different sleep stages, various signals of PSG have different characteristics, and several of the important signals are briefly described below.
Electroencephalographic (EEG) signals are the main signals used in sleep staging work. There are approximately 100 billion neurons in the human brain, which have a large number of synapses. These neurons communicate with each other and generate weak synaptic potentials during brain activity. The synchronized occurrence of synaptic potentials in a large number of neurons generates an electric field strong enough to be detected on the surface of the scalp. The electric field produced by brain activity can be detected by applying electrodes to the scalp and recording the change in potentials over time to generate an EEG signal.
Respiratory flow is the rate and volume of air passing through the airway during breathing. During sleep, respiratory flow can be used to assess the state of airway obstruction and respiratory function, and is also important in determining sleep staging. During sleep, respiratory flow is usually measured by sensors in the nose, mouth, or throat. The magnitude and variability of respiratory flow can be used to determine the presence of sleep breathing disorders like apnea and hypoventilation during sleep. Also, respiratory flow can be used to determine sleep stages. During sleep, respiratory flow usually shows different patterns and changes. For example, respiratory flow is usually lower during REM sleep and higher during non-REM sleep. Therefore, sleep staging can be more accurately determined by respiratory flow measurements, thus helping to diagnose sleep disorders and develop appropriate treatment programs.
In summary, human sleep is a very complex physiological and neurological behavior that involves the functional activities of multiple organs and systems. Polysomnography contains signals collected from multiple parts of the human body to assess and monitor these physiological and neurological activities, helping physicians to monitor the status and quality of sleep, and thus to diagnose and treat sleep disorders and neurodegenerative diseases.
Frequency domain analysis of EEG signals is a technique for converting EEG signals from the time domain to the frequency domain to gain a better understanding of their characteristics and meaning. EEG signals are generated by the electrical activities of a large number of neurons, which show different oscillatory characteristics in different frequency ranges. Different physiological activities also generate EEG signals with different frequencies. Therefore, analyzing EEG signals from the frequency domain perspective can extract features that contribute to sleep staging.
The Fourier Transform (FT) is a common method of converting a time domain signal into the frequency domain [22]. It decomposes a signal into a series of weighted sums of sine and cosine waves and represents them as amplitudes and phases at different frequencies. Since the Fourier transform is too computationally intensive, researchers later proposed the Fast Fourier Transform (FFT). It is an efficient algorithm for calculating the Fourier transform, which can reduce the calculation time of the Fourier transform of a discrete sequence of length N from
where
In order to analyze for different typical brain waves, a power spectral density (PSD) estimation method was introduced, which is the square of the Fourier transform that represents the power or energy of the signal at each frequency. For the energy
In this paper, phase-locked values (PLVs) were utilized to construct the topological map structure of polysomnograms when extracting features from data using graph convolutional neural networks [23]. Phase-locked value is a metric used to analyze the degree of phase synchronization of EEG signals. It can be used to study functional connectivity patterns between different brain regions, as well as the interrelationships between different frequencies in the EEG signal. The phase-locked value usually involves dividing the EEG signal into different frequency bands and calculating the phase difference between different brain regions within each band to calculate the phase-locked value. The phase-locked values are between 0 and 1, with larger values indicating a greater degree of phase synchronization between the two signals. Meanwhile, sleep is a periodic physiological state, which involves coordination and synchronization between multiple brain regions and multiple body parts. Therefore, the use of phase-locked values to analyze multiple signal sources in polysomnography can reveal the differences that exist in the degree of functional connectivity between different parts during different sleep stages, so as to better understand the mechanism of information transfer and coordination between the parts.
The model architecture of the proposed Sleep Graph Structure Data Comparison Learning (SGECL) in this chapter is shown in Fig. 1, where the deep neural network
Overall illustration of the proposed framework
In addition, PLV is utilized to construct the connected edge weights between two nodes (electrode channels) in a polysomnogram to form an adjacency matrix. This adjacency matrix is used in graph convolutional neural networks to extract graph structure features on one hand, and on the other hand, it is used to mine positive samples in contrast learning. During the pre-training phase of contrast learning, in order to address the sampling bias problem, time-nearest neighbor samples are sampled as positive samples. On this basis, the confidence of positive and negative samples is further calculated based on the similarity of topological graph structure between samples. According to the characteristics of the contrast loss function, the temperature coefficients of the samples are adaptively adjusted by combining the confidence of the positive and negative samples to obtain better pre-training results. Finally, multiple experiments on two public datasets verify the effectiveness of the proposed method.
Some existing sleep staging methods usually use CNN to extract local features of EEG signals and RNN to obtain contextual information of temporal signals. However, traditional CNNs have certain limitations and are unable to mine the deep features of sleep signals to learn sleep-related events, while traditional RNN models have difficulty in effectively extracting the long-term dependencies of sleep data and learning the transformation rules of sleep stages.
In order to solve the above problems, this chapter proposes a sleep staging method, which first encodes each sub-calendar element of multiple epochs of the EEG signal into the corresponding representative features, and learns the features related to the sleep events using an improved ResNet, and the improved ResNet structure solves the problem of information loss and information confusion caused by the deep level of the traditional CNN superposition network, and on this basis A normalization-based attention mechanism (NAM) is added to dynamically adjust the model’s ability to pay attention to different features, ignore insignificant features, and improve the model performance. Finally, TCN is used to capture the dependencies of long time series, fully exploiting local and global features to fully understand the spatiotemporal characteristics of EEG data.
The fine-grained feature extraction module is mainly composed of ResNet and the attention mechanism. Since the parameters of ResNet layer can be shared, the parameters can be updated through back propagation, and its residual connection can improve the generalization ability of the model and help the network to better capture the detailed changes of the signal, and then the combination of the attention mechanism can ensure that the network can ignore the influence of insignificant features when extracting the features, improve the nonlinear expression of the effective features, and enhance the performance of the network.
Since professional practitioners will consider sleep events (e.g., sleep rhythm waves such as spindle waves,
Although the performance of deep neural networks can be improved by stacking the layers of the network, new problems may be encountered as the layers are stacked, resulting in the updating of the parameters in the underlying layers of the network becoming more difficult and affecting the model’s ability to learn. A residual network consists of a series of residual blocks that combine constant mappings with residual connections, allowing the network to train deep structures efficiently. Where
In the channel attention module, the scaling factor in batch normalization (BN) is used to apply to the channel dimension:
where
The output characterization formula of the channel attention module is:
The formula for the weights is:
The spatial attention submodule uses the scaling factor in BN applied to the spatial dimensions and outputs features:
The weights are:
Both sub-modules
where
In summary, the use of the NAM attention mechanism allows for attentional correction of the feature map in the channel and spatial dimensions, thereby enhancing the ability to accurately describe and distinguish between features related to sleep staging, making the analysis of sleep stages more comprehensive and reliable.
Due to the limitations of the convolutional kernel, traditional convolutional neural networks are not suitable to be used to model temporal sequential problems, and in some studies it has been found that specific convolutional structures can achieve better results. The earliest approach was to use the extended convolution of TCNs to capture long time patterns, which was done not only to avoid the problems of gradient vanishing and gradient explosion of RNNs, but also to achieve the effect of modeling long time series. Sequences are predicted by using the transform of a stacked multilayer one-dimensional fully convolutional network, allowing sequence information to be transmitted through the network layer by layer until a prediction is obtained.
TCN mainly uses a 1*1 convolution to capture the local patterns of sleep data to extract the information of different time steps with a sliding window, and can satisfy mapping relations like
where
In addition to causal and inflationary convolution, TCN also uses residual connections to train the deep network and to enable transfer of information across layers, with two convolutional and nonlinear mappings for each residual block, and uses a weight normalization operation and a regularization operation to accelerate the convergence and stabilize the training process, and to improve the model generalization ability. In addition, the GeLU function is used to replace the original ReLu function to improve the expressive ability of the model. After this series of transformations, the final can be expressed as:
The output feature sequence after the fine-grained feature extraction module is:
After TCN can effectively capture the characteristic patterns of sleep data, learn the transition rules of sleep stages, and produce equal-length prediction outputs:
Finally, the output of TCN will be processed by the fully connected layer for dimensionality reduction, further extracting and combining features, and then the Softmax function outputs the classification results.
Sleep is a vital physiological activity for living organisms, taking up a significant portion of their entire life cycle. Sleep is a dynamic process, and rational stage staging is the basis for diagnosing sleep disorders, detecting early signs of disease, studying sleep quality, and other important medical diagnostic techniques. Sleep status is usually determined by sleep specialists using sleep staging criteria and polysomnography (PSG). In the current research in the field of sleep staging, sleep specialists usually use the sleep staging criteria developed by R&K Sleep Staging Criteria and the American Academy of Sleep Medicine (AASM) for sleep stage identification. Among the indicators, the main ones are electroencephalography, electrooculography, and electromyography, and the auxiliary parameters are electrocardiography and respiratory airflow. Sleep is classified into two major states, non-rapid eye movement (NREM) and rapid eye movement (REM) sleep, in addition to the arousal state.The NREM sleep stage characteristically begins with nighttime onset of sleep and progresses as sleep deepens. During this stage, a person’s breathing becomes shallow, slow and even, heart rate slows, blood pressure drops, muscles throughout the body are relaxed (still able to hold a certain position), and there is no significant eye movement. In this stage, it can also be divided into 3 stages: N1,N2 and slow wave sleep.N1 sleep is the stage of transition from the waking state to other sleep, or the transition stage after the appearance of body movements during sleep. Initially, slow eye movements may be present, and late EEG may show irregular high-amplitude cortical sharp waves (top waves, voltage 50-75 uV).Stage N2 is characterized by the appearance of sleep spindle waves or non-wake-related K-complex waves.Stage SWS is characterized by the appearance of sleep
In this section, sleep data is cited from a healthy male subject who was required to be in a standard sleep laboratory. This laboratory utilized PSG. The primary recordings throughout the experiment consisted of a 6-channel electroencephalogram (EEG).
In summary, for EEG, it has different waveform characteristics and components in different sleep stages, as shown in Figure 2. However, sleep conditions are complex and diverse, and the recognition of waveforms requires expert experience on one hand, while the judgment criteria are complex on the other. The data sampling frequency of the EEG quoted in this section is 128 Hz, and it can be seen from the figure that the voltage ranges of SWS, N2, N1, REM, and WAKE are different, which are [-50,5], [-50,5], [-40,40], [-20,20], and [-25,25], respectively.
Examples of EEG signals in different sleep stages
Strictly conforming to the inclusion and exclusion criteria of this study, the 141 selected insomnia patients in the medical center were grouped by applying RAND in Excel for complete randomization: they were randomly divided into three groups according to the ratio of 1:1:1: 47 people in the control group, 47 people in the ordinary music group, and 47 people in the somatosensory sound wave physiotherapy group.
Pre-intervention preparation
(1) Consulting experts and debugging equipment. The carrier equipment used in this study was provided by P Technology Co. Ltd, and the name of the equipment was Empower Frequency Music Therapy (EMT) system. Before the start of the intervention, we contacted a professional music therapist to select five audio tracks suitable for insomnia patients, edited them using audio editing software, and then the company’s technical experts used somatosensory sound wave technology to produce and debug the five audio tracks.
(2) Design the general information questionnaire and learn the use of each assessment scale, and master the use of the Frequent Music Therapy System instrument.
(3) Pre-experimentation was conducted from August to September 2021 at the Shoujia Medical and Nursing Center in TS. The sample size for the pre-experiment is 10%-20% of the sample size for the research design, and finally, 6 people will be taken in each group, totaling 18 people. To find the shortcomings in the research design and adjust the research program in time, this pre-trial was adjusted to the intervention time according to the work and rest time of the insomnia patients who participated in the intervention.
Before the intervention, general information questionnaires were collected from all subjects and baseline assessments were conducted using PSQI, DBAS-16, and other assessment tools. The control group was given the appropriate health knowledge education, the general music group added general music training on the basis of health knowledge education, and the somatosensory sound wave physiotherapy group added somatosensory sound wave training converted by BBT on the basis of health knowledge education.
(1) Control group: subjects were given 20-30 minutes of health education once a week to promote sleep and cognitive function, including maintaining good sleep habits, reducing water intake 2 hours before going to bed, using the toilet before going to bed, soaking feet in warm water for 30 minutes before going to bed, avoiding strenuous exercise and agitation before going to bed, improving the environment that is not conducive to sleep (quietness, moderate light, temperature and humidity), controlling behaviors that interfere with sleep, and prohibiting the consumption of strong tea before going to bed. Behavior, before going to bed, prohibit drinking strong tea, coffee and other stimulating beverages. Correct understanding of cognitive disorders, risk factors and protective factors: including quitting smoking and alcohol, correct use of medication, urging insomnia patients to maintain healthy living and dietary habits, encouraging insomnia patients to participate in group activities to maintain a good mood, cultivating hobbies and interests, increasing the frequency of physical exercise, and encouraging insomnia patients to talk about their inner depressing factors and helping to channel them. There are two ways to organize health lectures: playing video materials and organizing them.
(2) somatosensory sound wave physical therapy group: avoiding insomnia patients eating and treatment time, twice daily training, according to the wishes and habits of each subject insomnia patients to choose 12:00 ~ 15:00, 17:30 ~ 21:30 time periods, that is, close to the insomnia patients lunch break or night rest time, each person a total of one hour per day, 14 consecutive days of intervention for a cycle of intervention in three cycles. Each cycle is separated by a one-week rest period, the whole intervention lasts eight weeks, and the total number of interventions per person is 42 times, with a total length of 42 hours (the intervention can be continued within three days of discontinuation within the original intervention cycle, and those who have discontinued for more than three days are required to start from the beginning of this cycle of intervention training). The trainer assisted the insomnia patient to adjust the comfortable lying position, put on the eye mask (can be replaced by cotton cloth or towel) and accurately distinguish between left and right earphones, and adjusted the volume of the earphones according to the insomnia patient’s personal situation.
The Pittsburgh Sleep Quality Index Scale (PSQI), including 19 entries and 7 dimensions, of which 18 entries were involved in scoring, were subjective sleep quality, time to sleep, sleep duration, sleep efficiency, hypnotic medication and daytime dysfunction.The Cronbach’s alpha coefficient of the Pittsburgh Sleep Quality Index Scale (PSQI) was 0.853.The Cronbach’s alpha coefficient of the culturally-tuned Chinese version of the scale was 0.844. Each dimension scored between 0 and 3 points, with a total score of between 0 and 21 points. 0.844. The score for each dimension ranges from 0 to 3, and the total score ranges from 0 to 21, with higher scores indicating poorer sleep quality of the subjects. The Pittsburgh Sleep Quality Index >7 was used as the reference threshold for sleep quality problems in adults, and a score of >1 for a single dimension indicated sleep problems in that dimension.
Table 1 shows the comparison of total PSQI scores and scores of various dimensions for the three groups of study subjects before and after the intervention.
Before and after the intervention, the PSQI total score and each dimensional score
| Project | Group | Preintervention | Four weeks of intervention | Eight weeks of intervention | F | P |
|---|---|---|---|---|---|---|
| Sleep quality | Control group | 2.27±0.46 | 2.24±0.59 | 2.19±0.51 | 0.024 | 0.863 |
| Ordinary music group | 2.14±0.37 | 1.47±0.53 |
65.654 | <0.002 | ||
| Body sensory acoustic therapy group | 2.27±0.45 |
143.16 | <0.003 | |||
| F | 2.378 | 47.963 | 113.674 | / | / | |
| P | 0.097 | <0.001 | <0.002 | / | / | |
| Fall asleep time | Control group | 2.78±0.26 | 2.82±0.26 | 2.98±0.24 | 3.169 | 0.915 |
| Ordinary music group | 2.86±0.46 | 2.49±0.76 |
28.148 | <0.001 | ||
| Body sensory acoustic therapy group | 2.91±0.16 | 80.698 | <0.001 | |||
| F | 1.049 | 15.954 | 66.948 | / | / | |
| P | 0.296 | <0.002 | <0.001 | / | / | |
| Sleep time | Control group | 2.78±0.43 | 2.79±0.46 | 2.74±0.46 | 0.048 | 0.815 |
| Ordinary music group | 2.71±0.46 | 2.34±0.46 |
48.154 | <0.001 | ||
| Body sensory acoustic therapy group | 2.79±0.46 | 97.695 | <0.001 | |||
| F | 0.004 | 19.462 | 85.954 | / | / | |
| P | 0.936 | <0.001 | <0.002 | / | / | |
| Sleep efficiency | Control group | 2.97±0.15 | 2.78±0.26 | 2.94±0.24 | 1.245 | 0.548 |
| Ordinary music group | 2.91±0.24 | 2.48±0.39 | 7.835 | 0.002 | ||
| Body sensory acoustic therapy group | 3.02±0.00 | 31.485 | <0.005 | |||
| F | 1.036 | 1.765 | 19.365 | / | / | |
| P | 0.342 | 0.196 | <0.002 | / | / | |
| Sleep disorders | Control group | 1.45±0.51 | 1.48±0.56 | 1.49±0.48 | 1.452 | 0.945 |
| Ordinary music group | 1.48±0.59 | 1.32±0.48 | 0.915 | 0.385 | ||
| Body sensory acoustic therapy group | 1.48±0.53 | 34.582 | <0.002 | |||
| F | 0.048 | 12.965 | 13.485 | / | / | |
| P | 0.915 | <0.001 | <0.001 | / | / | |
| Hypnotic | Control group | 1.65±0.74 | 1.67±0.78 | 1.76±0.71 | 0.945 | 0.758 |
| Ordinary music group | 1.47±0.71 | 1.22±0.68 |
3.936 | 0.023 | ||
| Body sensory acoustic therapy group | 1.63±0.94 | 32.485 | <0.001 | |||
| F | 0.046 | 15.965 | 45.631 | / | / | |
| P | 0.931 | <0.001 | <0.002 | / | / | |
| Daytime function Obstacles | Control group | 1.45±0.59 | 1.85±0.49 | 1.34±0.41 | 0.085 | 0.915 |
| Ordinary music group | 1.48±0.56 | 1.36±0.49 | 1.341 | 0.242 | ||
| Body sensory acoustic therapy group | 1.49±0.58 | 35.185 | <0.001 | |||
| F | 0.028 | 12.945 | 12.632 | / | / | |
| P | 0.926 | <0.001 | <0.001 | / | / | |
| PSQI | Control group | 15.49±1.79 | 15.94±1.92 | 15.36±1.85 | 0.048 | 0.965 |
| Ordinary music group | 15.28±1.68 | 12.78±1.63 |
58.415 | <0.001 | ||
| Body sensory acoustic therapy group | 15.96±1.75 | 187.596 | <0.001 | |||
| F | 0.634 | 49.852 | 167.933 | / | / | |
| P | 0.536 | <0.001 | <0.002 | / | / |
Note: : Compared with the control group P<0.05,
1) Comparison between groups: before the intervention, the difference between the PSQI total scores and scores of various dimensions of the three groups of study subjects was not statistically significant (P>0.05).After four weeks of the intervention and after eight weeks of the intervention, the three groups of study subjects had the PSQI total scores, time of sleep onset, quality of sleep, efficiency of sleep, duration of sleep, hypnotic medication, and sleep disorders, Daytime dysfunction scores were compared and the differences were statistically significant (p<0.05). Further two-by-two comparisons showed that after four and eight weeks of intervention, the total PSQI scores of the three groups of study subjects were reduced, in which the reduction in the somatosensory sound wave physiotherapy group was more pronounced than that in the ordinary music group and the control group, and the difference was statistically significant (P<0.05). Sleep quality, time to sleep, sleep duration, sleep efficiency, sleep disturbance, hypnotic medication, daytime dysfunction, and total PSQI score scores were reduced in both the somatosonic physiotherapy group and the regular music group, and the reduction was greater in the somatosonic physiotherapy group than in the regular music group, with the difference in the reduction between the two groups in the eight-week comparison of the interventions being 0.35, 0.42, 0.38, 0.43, 0.32, 0.86, 0.23, and 3.73.
2) Within-group comparison: after the intervention, sleep quality, time to fall asleep, sleep duration, sleep efficiency, sleep disorder, hypnotic medication, daytime dysfunction, and PSQI scores were reduced in the somatosonic physiotherapy group, and the differences were statistically significant (P<0.05), and the scores were reduced in the eight weeks of the intervention compared to the four weeks of the intervention by 0.38, 0.82, 0.73, 0.6, 0.03, respectively, 0.43, 0.03, 2.69.
Table 2 shows the repeated ANOVA, the repeated ANOVA of PSQI total score and each dimension scores of the three groups of study subjects before and after the intervention: the repeated measures ANOVA was used to compare the PSQI total score, time to fall asleep, sleep duration, sleep quality, sleep disorders, sleep efficiency, daytime dysfunction, and hypnosis medication scores of the three groups of study subjects at different time points, and the ball test was used to test the scores that did not satisfy the condition of ball symmetry of the covariance matrix, the Using the Greenhouse-Geisser test with corrected degrees of freedom, the ANOVA results showed that the differences between the PSQI total score, sleep quality, time to fall asleep, sleep efficiency, sleep duration, hypnotic medication, sleep disorder, and daytime dysfunction scores were statistically significant (P=0.001/<0.001) in terms of the main effect of the intervention, the main effect of time, and the interaction effect, indicating that There were different trends in PSQI total score, sleep quality, sleep disorder, sleep duration, sleep time, sleep onset time, sleep efficiency, hypnotic medication, sleep disorder, daytime dysfunction scores over time in the three groups of study subjects after four weeks and eight weeks of intervention, and the somatosonic physiotherapy group was superior to the ordinary music group than the control group.
Repeated variance analysis
| Project | Time master effect | Intervention main effect | Interaction effect | |||
|---|---|---|---|---|---|---|
| F | P | F | P | F | P | |
| Quality of sleep | 279.654 | <0.001 | 49.635 | <0.001 | 70.645 | <0.001 |
| Bedtime | 126.746 | 0.001 | 36.752 | <0.001 | 65.742 | <0.001 |
| Sleep time | 176.635 | <0.001 | 36.485 | <0.001 | 45.952 | <0.001 |
| Sleep efficiency | 35.736 | 0.001 | 11.967 | 0.001 | 23.821 | <0.001 |
| Sleep disorder | 12.934 | 0.001 | 7.945 | <0.001 | 13.948 | <0.001 |
| Hypnotic | 51.963 | 0.001 | 12.997 | <0.001 | 58.918 | <0.001 |
| Daytime dysfunction | 13.934 | <0.001 | 7.752 | 0.001 | 10.156 | <0.001 |
| PSQI | 246.964 | <0.001 | 65.963 | <0.001 | 146.765 | <0.001 |
Table 3 shows the sleep beliefs and attitudes, 1) Comparison between groups: before the intervention, the difference between the DBAS-16 total score and the scores of each dimension of the three groups of study subjects was not statistically significant (P>0.05), four weeks after the intervention and eight weeks after the intervention, the three groups of study subjects DBAS-16 total scores, exaggerating the adverse effects of insomnia, exaggerating the despair of insomnia worrying about insomnia, unrealistic expectation of insomnia, incorrect knowledge of assisted sleep scores comparison The difference was statistically significant (P<0.05), and the DBAS-16 total score, exaggerated insomnia adverse effect, exaggerated insomnia worry and despair, unrealistic expectation of insomnia, and incorrect understanding of assisted sleep scores of the three groups of study subjects increased, with the increase of the somatosensory phonophonic physiotherapy group being greater than that of the ordinary music group and greater than that of the control group. In a two-by-two comparison, the total DBAS-16 scores of the somatosensory sound wave physiotherapy group were better than those of the ordinary music group and better than those of the control group. 8 weeks after the intervention, for example, the total DBAS-16 scores of the control group, the ordinary music group, and the somatosensory sound wave physiotherapy group were 45.98, 53.42, and 64.95, respectively. The difference was statistically significant (P<0.05).
Sleep beliefs and attitudes
| Project | Group | Preintervention | Four weeks of intervention | Eight weeks of intervention | F | P |
|---|---|---|---|---|---|---|
| Exaggerate the consequences of insomnia | Control group | 12.94±1.93 | 12.96±1.82 | 13.85±1.71 | 0.048 | 0.945 |
| Ordinary music group | 13.56±1.59 | 15.37±1.48 |
15.96±1.48 | 32.485 | <0.002 | |
| Body sensory acoustic therapy group | 13.85±1.63 | 18.62±1.24 |
19.69±1.39 |
183.64 | <0.001 | |
| F | 1.069 | 98.153 | 146.965 | / | / | |
| P | 0.385 | <0.002 | <0.001 | / | / | |
| Fear of insomnia | Control group | 16.36±1.85 | 16.96±1.62 | 16.99±1.52 | 0.918 | 0.348 |
| Ordinary music group | 16.08±2.06 | 19.85±1.93 |
21.95±1.94 |
76.254 | <0.001 | |
| Body sensory acoustic therapy group | 15.96±2.26 | 23.48±1.76 |
26.48±2.47 |
234.595 | <0.001 | |
| F | 0.029 | 124.95 | 206.65 | / | / | |
| P | 0.915 | <0.001 | <0.001 | / | / | |
| Unrealistic expectations of insomnia | Control group | 5.59±1.16 | 5.78±1.02 | 5.95±0.81 | 2.348 | 0.078 |
| Ordinary music group | 5.36±0.89 | 5.63±1.05 | 5.93±1.15 | 4.095 | 0.048 | |
| Body sensory acoustic therapy group | 5.71±0.78 | 6.18±0.91 |
7.05±0.93 |
26.655 | <0.001 | |
| F | 2.364 | 11.964 | 21.945 | / | / | |
| P | 0.185 | <0.002 | <0.001 | / | / | |
| Unconvinced of the help of sleep | Control group | 9.18±1.49 | 9.87±1.12 | 9.97±1.36 | 1.485 | 0.278 |
| Ordinary music group | 9.23±1.34 | 10.96±0.75 |
11.35±0.91 |
60.565 | <0.001 | |
| Body sensory acoustic therapy group | 9.08±1.45 | 11.39±1.11 |
12.85±1.59 |
86.965 | <0.001 | |
| F | 0.296 | 48.35 | 58.35 | / | / | |
| P | 0.715 | <0.001 | <0.001 | / | / | |
| DBAS-16 | Control group | 44.32±3.69 | 44.36±3.27 | 2.855 | 0.087 | |
| Ordinary music group | 43.18±2.67 | 51.63±3.37 |
145.185 | <0.001 | ||
| Body sensory acoustic therapy group | 315.985 | <0.001 | ||||
| F | 0.296 | 198.635 | 307.965 | / | / | |
| P | 0.754 | <0.001 | <0.001 | / | / |
Note: : Compared with control group P<0.05:
: Compared with common music group P<0.05:
2) Comparison within groups: after four weeks and eight weeks of intervention, the DBAS-16 total score, exaggerating the adverse effects of insomnia, worrying about insomnia and despairing about insomnia, unrealistic expectation of insomnia, and incorrect knowledge of sleep aids in the somatic sound wave physiotherapy group and the ordinary music group all increased, and compared with the pre-intervention period, the DBAS-16 total score of the somatic sound wave physiotherapy group in the four weeks of intervention and the eight weeks of intervention was 57.89, respectively, 64.95, the improvement effect is more obvious. The difference was statistically significant (P<0.05). The total DBAS-16 score and the scores of each dimension in the control group increased slightly, and the difference was not statistically significant (P>0.05).
Based on the scoring criteria of each item in the PSQI and its weighting, the scoring criteria that can quantitatively analyze the assessment results were completed by improving some of the entries, and Table 4 shows the sleep quality scoring criteria.
Sleep quality criteria
| Evaluation item | Inclusion entry | Evaluation score | Scoring interval | |||
|---|---|---|---|---|---|---|
| A | Time of sleep | ≤15min | 16~30min | 31~60min | ≥60min | 0~2.5 |
| 0 | 0.5 | 1 | 1.5 | |||
| 30min unsleep | NO | YES | ||||
| 0 | 1 | |||||
| B | Actual sleep time | ≥7h | 6~7h | 5~6h | ≤5h | 0~3 |
| 0 | 1 | 2 | 3 | |||
| C | Actual sleep time/in bed time | ≥85% | 75~84% | 65~74% | ≤65% | 0~3 |
| 0 | 1 | 2 | 3 | |||
| D | Wake up at night or wake up early | NO | YES | 0~1.5 | ||
| 0 points | 0.5 points | |||||
| Go to the toilet at night | NO | YES | ||||
| 0 | 0.5 | |||||
| Sleeping position (sleep comfort) | ≤10 times(Comfort) | >10 times(Discomfort) | ||||
| 0 | 0.5 | |||||
The scores of four assessment items were counted, and the entries contained in each assessment item were scored according to their weighting in the overall assessment. The results filled out by the subjects synchronously and autonomously were used as the reference standard, and the calculation results were obtained by quantifying the above assessment items and compared with the reference standard to verify the accuracy of the experimental protocol. Among them, the first entry in items A, B, C and D can be counted by awake and sleep recognition, the second entry in item D can be counted by off-bed discrimination, and the third entry sleep comfort is subjective from the subject’s point of view, and the frequency of switching sleeping positions is used to simulate this item, and the quality of sleep is relatively poor when the sleeping positions are switched frequently. Finally, the scores of the four assessment items were summed up, and the comprehensive sleep quality evaluation results were obtained according to the scoring intervals, with [0,2.5] as very good, (2.5,5] as good, (5,7.5] as poor, and (7.5,10] as very poor.
Twelve subjects were randomly selected from among the subjects, and the results of sleep quality assessment calculated by BCG signals and the results of the reference standard filled in by the subjects themselves are shown in Table 5. As can be seen from the table, the calculated results of normal subjects are mostly consistent with the total score of the reference standard results, and the calculated results of two subjects with sleep disorders (D11, D12) are completely consistent with the total score of the reference standard results, which are 7 and 5.5, respectively. It indicates that the sleep quality assessment model constructed in this paper has a high accuracy in recognizing low-quality sleep. Some of the statistical results with differences have low differences, and considering that the final results are generated within a certain interval, some of the differences have little effect on the overall results. Among them, the differences produced by assessment items A and B and assessment item D are more, and the number of differences is 6, 4 and 4 times respectively. Statistically, since the assessment item A involves the time of falling asleep, and the intervals of the scoring criteria are more carefully divided, it is easy to produce differences in the scores of this item in the wakefulness/sleeping recognition results unless it achieves a very high degree of precision. item D involves a number of entries, among which the recognition of the nighttime toilet use is 100% correct, while sleep comfort is a subjective entry in the self-administered results, and by calculating the number of sleeping positions and the number of switches, it fails to match the subjective feelings. The number of switching times failed to fully align with subjective feelings, which is prone to producing poor values. In addition, the PSQI itself involves many entries, and the scoring criteria for some quantifiable entries selected in this paper are designed according to the original rules. Due to the research purpose of this paper, the assessment cycle is chosen in units of one day, so compared with the original rules of the PSQI, the results of the assessment model designed in this paper are more stringent in their discriminations.
Statistics of sleep quality assessment results
| Number | Evaluation item A | Evaluation item B | Evaluation item C | Evaluation item D | Total | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| Calculate | Reference | Calculate | Reference | Calculate | Reference | Calculate | Reference | Calculate | Reference | |
| D01 | 0 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 2 |
| D02 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 1 | 1 |
| D03 | 0 | 0 | 0 | 0 | 0 | 0 | 0.5 | 1.5 | 0.5 | 1.5 |
| D04 | 1 | 0.5 | 1 | 2 | 0 | 0 | 0.5 | 1 | 2.5 | 3.5 |
| D05 | 0.5 | 1 | 1 | 2 | 0 | 0 | 0.5 | 0 | 2 | 3 |
| D06 | 0 | 0 | 1 | 2 | 0 | 0 | 1 | 1 | 2 | 3 |
| D07 | 1 | 2 | 1 | 1 | 0 | 0 | 1 | 1 | 3 | 4 |
| D08 | 2 | 3 | 0 | 1 | 0 | 0 | 0.5 | 0.5 | 2.5 | 4.5 |
| D09 | 1 | 1 | 2 | 2 | 0 | 0 | 0.5 | 0.5 | 3.5 | 3.5 |
| D10 | 2.5 | 2 | 2 | 2 | 1 | 1 | 0.5 | 0.5 | 6 | 5.5 |
| D11 | 2 | 2 | 3 | 3 | 1 | 1 | 1 | 1 | ||
| D12 | 2 | 2 | 2 | 2 | 1 | 0.5 | 0.5 | 1 | ||
In order to demonstrate the validity of the assessment results of the model, the final comparison results are listed and a paired-sample t-test is performed, as shown in Table 6. Table 6 shows that the sleep quality assessment model based on PSQI assessment items proposed in this paper is able to agree with the results of the reference standard in terms of calculation results, and some of the results that produce deviations are due to the continuous scoring intervals, and the total scores are at the junction of the assessment results around the junction of the assessment results, which affects the final results of the assessment. The p-value of the total score of the two results was calculated to be 0.5936, indicating that there is no significant difference between the two final results (p>0.05), and that the calculated results can be used in place of the results of the autonomy fill in results, as well as the results of the wearable device. The fact that it is taken from PSQI makes this model have a strong theoretical value, and it also takes into account a number of sleep physiological parameter statistics, which helps to make a more comprehensive analysis of sleep quality during all-night monitoring.
Statistics of final results of sleep quality assessment
| Number | Sleep Quality Assessment Results | |
|---|---|---|
| Calculate | Reference | |
| D01 | Good | Good |
| D02 | Good | Good |
| D03 | Good | Good |
| D04 | Good | Better |
| D05 | Good | Better |
| D06 | Good | Better |
| D07 | Better | Better |
| D08 | Good | Better |
| D09 | Better | Better |
| D10 | Worse | Worse |
| D11 | Worse | Worse |
| D12 | Worse | Worse |
| Average Score | 3.042±2.648 | 3.667±2.678 |
| P | 0.5936 | |
Somatosensory sound wave physical therapy technology is applied in this paper to treat sleep disorders and a method for testing sleep quality based on temporal information learning to determine the prognostic effect of music therapy is proposed. The intervention group was divided into a control group, a normal music group, and a somatosensory sound wave physical therapy group. A PSQI questionnaire was used to complete the baseline assessment. In the sleep quality assessment, for example, at eight weeks of intervention, the sleep quality, time to sleep, sleep duration, sleep efficiency, sleep disturbance, hypnotic medication, daytime dysfunction, and PSQI scores of the somatosensory sound wave physiotherapy group were 0.35, 0.42, 0.38, 0.43, 0.32, 0.86, 0.23, and 3.73 lower than that of the general music group, indicating that among the three groups, somatosensory sound wave physiotherapy had a very good effect on the improvement of sleep quality. The total DBAS-16 scores of the control group, the ordinary music group, and the somatosensory sound wave physical therapy group were 45.98, 53.42, and 64.95 respectively, respectively, and the difference was statistically significant.