A neck-worn sleep apnea and blood oxygen monitoring necklace device and a monitoring method thereof

Abstract

This invention discloses a neck-worn sleep apnea and blood oxygen monitoring necklace device and its monitoring method, belonging to the field of sleep monitoring technology. The device includes: a flexible, skin-friendly neck ring, which integrates a reflective blood oxygen sensor, a piezoelectric thin film sensor, and an accelerometer; a pendant-style main control box, which has a built-in battery, microprocessor, wireless communication module, and alarm unit. This invention also includes: (1) a method for extracting weak respiratory vibration signals from the neck, which removes cardiac pulsation interference through adaptive filtering; (2) an adaptive filtering method for turning over interference, which identifies changes in body position in real time based on an accelerometer and dynamically adjusts the judgment window; (3) a multi-parameter fusion method for judging and classifying apnea events, which automatically classifies OSA, CSA, and MSA based on snoring energy and blood oxygen decrease; (4) a graded low-power all-night monitoring method; and (5) an apnea event trend early warning method. This invention adopts a necklace-style, imperceptible wearing design, which has the advantages of high sensitivity, anti-interference, automatic classification, and long battery life, and is suitable for long-term continuous monitoring and forward-looking early warning of sleep apnea in the home.

Description

Technical Field

[0001] This invention relates to the field of sleep breathing monitoring and signal processing technology, specifically a neck-worn sleep apnea and blood oxygen monitoring necklace device and its monitoring method. Background Technology

[0002] The incidence of sleep apnea syndrome (OSA) is increasing year by year in both children and adults. Traditional monitoring methods have the following shortcomings:

[0003] Polysomnography (PSG) requires laboratory testing, involves numerous electrodes, disrupts natural sleep, and cannot enable long-term home monitoring.

[0004] 2. Home-use finger clip pulse oximeters are prone to falling off, are not suitable for wearing overnight, and cannot provide respiratory airflow information;

[0005] 3. Mask-type or chest strap-type respiratory monitoring devices are uncomfortable, have low patient compliance, and less than 30% of patients accept wearing them overnight;

[0006] 4. Existing neckband products (earphones, massagers) do not have medical monitoring functions and are not designed for comfortable overnight wear;

[0007] 5. The neck breathing vibration signal is weak (amplitude <0.1m / s²), and is easily interfered with by movements such as swallowing, talking, and turning over. Existing signal processing methods are difficult to achieve accurate detection of sleep apnea.

[0008] 6. There is a lack of automatic identification and classification methods for sleep apnea events based on the fusion of multimodal sensors in the neck (respiratory vibration + blood oxygen + snoring + body position). Summary of the Invention

[0009] Purpose of the invention

[0010] This invention provides a neck-worn sleep apnea and blood oxygen monitoring necklace device and its monitoring method, aiming to solve the problems of existing sleep breathing monitoring devices such as discomfort when worn, easy to fall off, large signal interference, inability to monitor continuously all night, and lack of automatic classification ability.

[0011] Technical solution

[0012] A neck-worn sleep apnea and blood oxygen monitoring necklace device, characterized in that it comprises:

[0013] The flexible, skin-friendly neckband (A1) integrates the following internally or on its surface: a reflective blood oxygen sensor (A2) that conforms to the skin of the neck; a piezoelectric film sensor (A3) for picking up breathing vibrations and snoring; and an accelerometer (A4) for detecting sleep position and turning over disturbances.

[0014] The pendant-style main control box (A5) contains a built-in battery (A6), microprocessor (A7), wireless communication module (A8), and alarm unit (A9).

[0015] The microprocessor (A7) is configured to: identify apnea events based on respiratory vibration signals, and determine whether it is obstructive apnea by combining the blood oxygen drop threshold and snoring grade.

[0016] This invention also includes the following method technical features:

[0017] I. Method for Extracting Weak Respiratory Vibration Signals from the Neck

[0018] The respiratory vibration signal extraction (C4) is characterized by the following steps:

[0019] R1: Piezoelectric thin film sensor (A3) acquires vibration signals from the neck surface, with a sampling rate ≥200Hz;

[0020] •R2: Filters out low-frequency drift and high-frequency noise through a 0.5Hz-15Hz bandpass filter;

[0021] •R3: Adaptive moving average filtering is used to remove cardiac pulsation interference (frequency 1-2Hz) and extract pure respiratory envelope signal;

[0022] •R4: Perform peak detection on the respiratory envelope signal, identify the inspiratory start point and expiratory end point, and calculate the respiratory cycle T_breath and respiratory amplitude A_breath;

[0023] •R5: When A_breath is less than 30% of the normal baseline for two consecutive respiratory cycles, it is considered a respiratory atony event.

[0024] II. Adaptive Filtering Method for Turning Over Interference

[0025] The feature is that the turning-over interference filtering module (C6) performs the following steps:

[0026] •L1: Accelerometer (A4) collects triaxial acceleration data of the neck and calculates the composite acceleration amplitude and Euler angles;

[0027] •L2: When the magnitude of the synthesized acceleration exceeds 0.5g for 0.5 seconds for 0 consecutive seconds, it is determined to be a postural change event;

[0028] •L3: Record the start time t_start and end time t_end of the body position change, and calculate the interference window T_interfere = t_end - t_start + 1 second;

[0029] •L4: Within the interference window T_interfere, the judgment result output by the apnea event judgment (C7) is marked as "invalid" and is not included in the apnea statistics;

[0030] •L5: After the interference window ends, reset the apnea timer and restart the test;

[0031] •L6: When the number of times you turn over at night exceeds the preset threshold (≥15 times / night), the interference window length is automatically adjusted to 1.5 seconds to improve the filtering sensitivity.

[0032] III. Multi-parameter fusion method for apnea event identification and classification

[0033] The feature is that the apnea event determination (C7) and the snoring grading module (C8) jointly perform the following steps:

[0034] •S1: Real-time monitoring of respiratory envelope signals; when the duration of respiratory arrest T_stop ≥ 10 seconds, it is marked as a suspected apnea event.

[0035] •S2: Synchronously acquire the SpO2 value output by the blood oxygen saturation calculation (C5), and record the blood oxygen change ΔSpO2 at the beginning and end of the event;

[0036] •S3: If ΔSpO2 ≥ 3%, it is considered a clinically significant apnea event;

[0037] •S4: Within the suspected sleep apnea event window, extract the high-frequency components (50-500Hz) from the signal of the piezoelectric thin film sensor (A3) and calculate the snoring energy E_snore;

[0038] •S5: Compare E_snore with the nighttime baseline snoring energy threshold E_threshold:

[0039] If E_snore ≥ 2×E_threshold, it is determined to be obstructive apnea (OSA).

[0040] If E_snore < 0.5×E_threshold and the chest and abdominal motion sensor (if present) detects respiratory effort, it is determined to be central apnea (CSA).

[0041] If it falls between the two, it is classified as mixed apnea (MSA).

[0042] •S6: Counts the number of various types of apnea events throughout the night, calculates the apnea-hypopnea index (AHI), and automatically generates a sleep breathing report.

[0043] IV. Low-power overnight monitoring method

[0044] The device is characterized by employing a hierarchical operating mode:

[0045] •P1: The reflective pulse oximeter (A2) uses an intermittent sampling mode (sampling once every 5 seconds, each sampling lasting 2 seconds), reducing power consumption by 70%;

[0046] •P2: The piezoelectric thin film sensor (A3) and the accelerometer (A4) are always sampled at 50Hz, but the microprocessor (A7) is only awakened for real-time processing when there is respiratory activity;

[0047] •P3: When there is no effective respiratory signal and no change in body position for 30 consecutive seconds, it enters standby mode (the sensor sampling rate drops to 5Hz).

[0048] •P4: If a change in acceleration or signal recovery is detected in standby mode, full-function monitoring will resume within 100ms;

[0049] •P5: Average power consumption ≤3mA throughout the night, 200mAh battery (A6) can continuously monitor for 3 nights.

[0050] V. Early Warning Methods for Sleep Apnea Event Trends

[0051] The alarm triggering module (C9) is characterized in that it further performs the following:

[0052] •W1: Record the frequency of apnea events f_AHI over the past 30 minutes;

[0053] •W2: If f_AHI shows an upward trend for three consecutive 10-minute windows (linear regression slope > 0), and the current blood oxygen SpO2 ≤ 90%, the alarm unit (A9) is triggered to issue a low-intensity vibration warning;

[0054] •W3: If the duration of a single apnea event is T_stop ≥ 20 seconds, or the blood oxygen SpO2 is ≤ 85%, a high-intensity alarm (vibration + buzzer) is triggered.

[0055] •W4: Alarm information is simultaneously sent to the guardian's terminal and the cloud server via the wireless communication module (A8).

[0056] Beneficial effects

[0057] 1. High signal extraction accuracy: By removing cardiac pulsation interference through adaptive moving average filtering, the respiratory envelope is accurately extracted from weak neck vibration signals, with a detection sensitivity ≥92%;

[0058] 2. Strong anti-interference capability: The adaptive filtering method for turning over interference is based on the real-time identification of body position changes by the accelerometer (A4) and dynamically adjusts the interference window, reducing the false alarm rate to ≤0.5 times / hour;

[0059] 3. Automatic classification capability: Based on the combined judgment of snoring energy and blood oxygenation decline, it realizes automatic classification of OSA, CSA and MSA to assist in clinical diagnosis;

[0060] 4. Low power consumption and long battery life: The hierarchical intermittent sampling and intelligent wake-up mechanism ensures that the average power consumption throughout the night is ≤3mA and supports continuous monitoring for 3 nights.

[0061] 5. Trend Early Warning: Analysis of the frequency trend of sleep apnea events enables proactive early warning rather than reactive alerts;

[0062] 6. Seamless wearing: The memory foam inner layer (B1) and the array of breathable holes (B4) ensure comfort all night long, and compliance is more than 3 times higher than that of the chest strap type. Attached Figure Description

[0063] Figure 1 The schematic diagram of the overall structure of the present invention includes: a flexible skin-friendly neckband (A1), a reflective blood oxygen sensor (A2), a piezoelectric thin film sensor (A3), an accelerometer (A4), a pendant-type main control box (A5), a battery (A6), a microprocessor (A7), a wireless communication module (A8), and an alarm unit (A9).

[0064] Figure 2 The diagram shows the layered structure of the neck collar of the present invention, which includes: a memory foam inner layer (B1), a silicone covering layer (B2), a sensor embedding groove (B3), and an array of vents (B4).

[0065] Figure 3 The signal processing flowchart of this invention includes: a reflective blood oxygen sensor (A2), a piezoelectric thin film sensor (A3), an accelerometer (A4), a respiratory vibration signal extraction (C4), a blood oxygen saturation calculation (C5), a turning-over interference filtering module (C6), a sleep apnea event judgment (C7), a snoring grading module (C8), and an alarm triggering module (C9).

[0066] Figure 4 This is a schematic diagram of the wearing state of the present invention, which includes: a neck-back fitting area (D1), a throat sensor area (D2), a clavicle support area (D3), and a magnetic charging contact (D4). Detailed Implementation

[0067] Example 1

[0068] This embodiment provides a neck-worn sleep apnea and blood oxygen monitoring necklace device and its monitoring method, as detailed below:

[0069] The flexible, skin-friendly neckband (A1) has a total length of 380mm, a width of 12mm, and a thickness of 4mm. Its layered structure consists of a memory foam inner layer (B1) with a thickness of 2.5mm and a silicone overlay layer (B2) with a thickness of 1.5mm. The silicone overlay layer (B2) has three sensor embedding slots (B3) on its inner side, used to fix a reflective pulse oximeter (A2) (model MAX30102), a piezoelectric film sensor (A3) (model LDT0-028K), and an accelerometer (A4) (MPU6050), respectively. The neckband surface has a uniformly distributed array of breathable holes (B4), with a hole diameter of 0.8mm and a spacing of 5mm.

[0070] The pendant-style main control box (A5) weighs 15g and measures 40mm × 25mm × 10mm. It contains a battery (A6) (200mAh lithium polymer), a microprocessor (A7) (nRF52840), a wireless communication module (A8) (Bluetooth 5.2), and an alarm unit (A9) (miniature vibration motor + buzzer). Magnetic charging contacts (D4) are located at the bottom.

[0071] When worn, the flexible, skin-friendly neckband (A1) has a neck-fitting area (D1) behind the cervical spine, a throat sensor area (D2) above the thyroid cartilage, and a clavicle support area (D3) that naturally rests on the clavicle to prevent slippage.

[0072] Respiratory vibration signal extraction (C4): The piezoelectric thin-film sensor (A3) samples at 200Hz, and after bandpass filtering of 0.5-15Hz, an adaptive moving average filter with a window length of 2 seconds is used to remove cardiac pulsation interference. The peak detection algorithm identifies the respiratory cycle, and respiratory attenuation is determined when the respiratory amplitude is lower than 30% of the baseline for two consecutive cycles.

[0073] Turning-over interference filtering module (C6): Accelerometer (A4) calculates the synthetic acceleration in real time. When the acceleration exceeds 0.5g for 0.5 seconds, it is determined that the body position has changed. The interference window is 1 second after the end of the change. During this period, the apnea judgment result is marked as invalid.

[0074] Multi-parameter fusion judgment: When breathing stops for ≥10 seconds and SpO2 decreases by ≥3%, snoring energy is extracted. In this embodiment, the nighttime baseline snoring energy threshold E_threshold is set to 0.5μV². If E_snore ≥ 1.0μV², it is judged as OSA; if E_snore < 0.25μV² and breathing effort is present, it is judged as CSA. Automatic grading is performed after calculating the overnight AHI: AHI < 5 is normal, 5-15 is mild, 15-30 is moderate, and > 30 is severe.

[0075] Low power mode: The reflective pulse oximeter (A2) samples for 2 seconds every 5 seconds; the piezoelectric film sensor (A3) and accelerometer (A4) always sample at 50Hz; when there is no valid signal for 30 consecutive seconds, the sensor drops to 5Hz standby. The measured power consumption overnight is 2.8mA.

[0076] Trend warning: If the AHI frequency is detected to increase from 2 times / hour to 6 times / hour (slope > 0.5) in the past 30 minutes, a vibration warning will be triggered; if a single apnea lasts for 22 seconds, a buzzer alarm will be triggered.

[0077] Example 2

[0078] The difference from Example 1 is that the piezoelectric film sensor (A3) is replaced with a bone conduction microphone (model Vibration Sensor S18), increasing the sampling rate to 400Hz and improving the sensitivity of respiratory vibration extraction by 40%, making it suitable for users with light snoring or children. The flexible, skin-friendly neckband (A1) is reduced in width to 10mm to fit children's neck sizes.

[0079] Example 3

[0080] The difference from Example 1 is that the turning-over interference filtering module (C6) further introduces a machine learning classifier: 500 sets of acceleration samples of neck turning over, swallowing, speaking, coughing and other movements are collected offline, time domain and frequency domain features (peak value, energy, zero crossing rate, etc.) are extracted, a random forest classifier is trained, and online recognition of movement types is achieved. Interference is filtered out only for turning over and violent head rotation, while respiratory signals under swallowing and slight movements are retained, further reducing the invalid data rate.

Claims

1. A neck-worn sleep apnea and blood oxygen monitoring necklace device, characterized in that, include: A flexible, skin-friendly neckband (A1) integrates the following internally or on its surface: a reflective blood oxygen sensor (A2), a piezoelectric thin film sensor (A3), and an accelerometer (A4); The pendant-style main control box (A5) has a built-in battery (A6), microprocessor (A7), wireless communication module (A8), and alarm unit (A9). The microprocessor (A7) is configured to: identify apnea events based on respiratory vibration signals, and determine the type of apnea by combining the blood oxygen drop threshold and snoring level.

2. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The microprocessor (A7) is configured with respiratory vibration signal extraction (C4), including: R1: Piezoelectric thin film sensor (A3) acquires vibration signals from the neck surface; R2: Filters out low-frequency drift and high-frequency noise through a 0.5Hz-15Hz bandpass filter; R3: Adaptive moving average filtering is used to remove cardiac pulsation interference and extract pure respiratory envelope signal; R4: Perform peak detection on the respiratory envelope signal, identify the inspiratory start point and expiratory end point, and calculate the respiratory cycle and respiratory amplitude; R5: When the respiratory amplitude is less than 30% of the normal baseline for two consecutive respiratory cycles, it is considered a respiratory attenuation event.

3. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The microprocessor (A7) is equipped with a turning-over interference filtering module (C6), which includes: L1: Accelerometer (A4) collects triaxial acceleration data of the neck and calculates the composite acceleration amplitude; L2: When the magnitude of the synthetic acceleration exceeds 0.5g for 0.5 seconds for 0.5 seconds, it is determined to be a postural change event; L3: Record the start and end times of body position changes and calculate the interference window; L4: Within the interference window, the result of the apnea event determination is marked as invalid; L5: After the interference window ends, reset the apnea timer.

4. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The microprocessor (A7) is configured with a multi-parameter fusion method for apnea event identification and classification, including: S1: Real-time monitoring of respiratory envelope signals; when the duration of respiratory arrest T_stop ≥ 10 seconds, it is marked as a suspected respiratory arrest event. S2: Simultaneously acquire blood oxygen saturation SpO2 value and record blood oxygen change ΔSpO2; S3: If ΔSpO2 ≥ 3%, it is considered a clinically significant apnea event; S4: Within the suspected sleep apnea event window, extract the high-frequency components from the signal of the piezoelectric thin film sensor (A3) and calculate the snoring energy E_snore; S5: Compare E_snore with the nighttime baseline snoring energy threshold E_threshold and classify it as obstructive, central, or mixed apnea.

5. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 4, characterized in that, In the classification method: • If E_snore ≥ 2×E_threshold, it is determined to be obstructive apnea (OSA). • If E_snore < 0.5×E_threshold and respiratory effort is detected, it is determined to be central apnea (CSA). • Otherwise, it is determined to be mixed apnea (MSA).

6. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The device employs a graded low-power operating mode: P1: The reflective pulse oximeter (A2) uses an intermittent sampling mode; P2: The piezoelectric thin film sensor (A3) and accelerometer (A4) sample at 50Hz and wake up the microprocessor (A7) only when there is respiratory activity; P3: If there is no effective respiratory signal and no change in body position for 30 consecutive seconds, it enters standby mode; P4: When a change in acceleration or signal recovery is detected in standby mode, full-function monitoring will resume within 100ms.

7. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The alarm triggering module (C9) is configured with a sleep apnea event trend early warning method: W1: Record the frequency of apnea events in the past 30 minutes; W2: If the frequency shows an upward trend for three consecutive 10-minute windows, and the current blood oxygen SpO2 ≤ 90%, a low-intensity vibration warning is triggered; W3: If the duration of a single apnea event is T_stop ≥ 20 seconds, or the blood oxygen SpO2 is ≤ 85%, a high-intensity alarm is triggered.

8. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The flexible, skin-friendly neckband (A1) has a layered structure: an inner layer of memory foam (B1) and a silicone covering layer (B2) with a width ≤15mm. The silicone covering layer (B2) is provided with a sensor embedding groove (B3) and an array of breathable holes (B4).

9. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The flexible, skin-friendly neckband (A1) has a neck-fitting area (D1), a throat sensor area (D2), and a clavicle support area (D3). The piezoelectric film sensor (A3) and a reflective blood oxygen sensor (A2) are installed in the throat sensor area (D2).

10. The neck-worn sleep apnea and blood oxygen monitoring necklace device according to claim 1, characterized in that, The device will issue an alarm via alarm unit (A9) when blood oxygen levels drop below 92% and breathing stops for more than 10 seconds.

Images