Multifunctional mattress, sleep system and operation method

Abstract

The present invention relates to a multifunctional mattress comprising a base support layer, an upper layer and a comfort layer, wherein the upper layer comprises: a combination of at least two integrated therapy systems, selected from the group comprising pressotherapy systems, vibration therapy systems and heat therapy systems; a biosignal sensor configured to obtain a user's physiological data, the physiological data comprising at least HRV; and a data processing unit configured to establish a reference HRV, calculate stress level by comparing HRV with the reference HRV and analysing a low frequency / high frequency index, infer a sleep phase and determine the automatic activation, adjustment or interruption of the operation of the therapy systems. The invention also relates to a sleep system comprising the mattress and to an operation method.

Description

[0001] DESCRIPTION

[0002] MULTIFUNCTIONAL MATTRESS, SLEEP SYSTEM AND OPERATING METHOD

[0003] TECHNICAL SECTOR

[0004] The present invention relates to the field of sleep systems, especially to a multifunctional mattress for improving sleep quality, to a sleep system and to a method of operation.

[0005] BACKGROUND OF THE INVENTION

[0006] Traditional mattresses primarily focus on providing comfort and support. However, there is a growing demand for products that also offer therapeutic benefits to improve sleep quality, reduce sleep latency, and enhance overall health and well-being. Compression therapy, a technique that uses air pressure to stimulate blood flow and lymphatic drainage, is commonly used in medical and wellness treatments. Vibration therapy aids muscle relaxation and pain relief, while heat therapy provides soothing warmth or cooling relief. When applied correctly, all of these therapies have proven effects on the central nervous system, resulting in relaxation and sleep induction.Traditional devices for these therapies are often bulky and not designed for comfort, materials, or functionality suitable for continuous use during the typical eight hours of sleep. They are designed for use while the user is awake, thus losing a significant portion of their potential therapeutic effect or sleep-enhancing potential. This is because, as is well known, muscle tension levels change during the different sleep stages, and therefore, so do the therapeutic needs of the various devices.The present invention integrates these therapeutic elements directly into a mattress, adds sensor(s) to monitor the user's vital signs, sleep patterns, and possible external parameters such as temperature and humidity, and a method using automatic data processing via AI (Artificial Intelligence) to determine the most appropriate therapy at any given time based on the sleep or wakefulness phase and the specific needs of each user at that particular moment. This provides a perfect and comfortable solution for users seeking continuous therapeutic benefits during rest and an improvement in sleep quality. EXPLANATION OF THE INVENTION.

[0007] The present invention relates to an advanced mattress design, which integrates therapeutic modalities, selected from pressotherapy, vibration therapy and thermal therapy, together with one or more sensors to monitor sleep, one or more physiological parameters and possible external variables related to it.

[0008] As used in this document, the term “mattress” should be understood in a broad and non-limiting sense, encompassing a mattress, a mattress topper, or any other similar means of rest.

[0009] Specifically, the object of the present invention is a multifunctional mattress comprising a base support layer, a top layer, and a comfort layer, wherein said top layer comprises:

[0010] - a combination of at least two integrated therapy systems, selected from the group comprising a pressotherapy system, a vibratory therapy system, and a thermal therapy system;

[0011] - a biosignal sensor, configured to obtain physiological data from a user during use of the mattress, the physiological data comprising at least heart rate variability (HRV);

[0012] - and a data processing unit, configured to:

[0013] ■ Establish a dynamic reference value for the user's heart rate variability, called the "HRV reference", based on successive uses of the mattress, and / or based on reference tables according to age, gender and physical condition

[0014] ■ Calculate the stress level in real time by comparing HRV with the reference HRV, and, optionally, by analyzing an LF / HF index;

[0015] ■ infer a sleep stage the user is in from physiological data, using classification algorithms comprising decision trees, nearest neighbor (k-NN) algorithms and / or recurrent neural networks RNN / LSTM trained with polysomnography recordings;

[0016] ■ Determine the automatic activation, adjustment, or interruption of the operation of the therapy systems, including selecting a single therapy or a combination of therapies based on the sleep stage and stress level. According to a particular embodiment, the upper layer comprises the three therapy systems, namely the pressotherapy system, the vibration therapy system, and the thermal therapy system.

[0017] According to one particular embodiment, the biosignal sensor is a heart rate variability sensor, e.g., a BCG sensor; and the processing unit is configured to calculate the stress level in real time by comparing the heart rate variability (HRV) with the HRV reference, wherein the HRV reference is initialized from population reference values ​​based on the user's age, gender, and fitness, and is dynamically updated based on machine learning and the user's physiological condition over successive uses of the mattress.

[0018] According to a particular embodiment, the mattress comprises one or more biosignal sensors selected from the group comprising a heart rate sensor, a respiratory rate sensor, and a user oxygen saturation sensor.

[0019] According to a particular embodiment, the mattress comprises:

[0020] - one or more environmental sensors, which are:

[0021] ■ one or more temperature sensors, configured to monitor the environmental conditions of a mattress sleeping surface;

[0022] ■ one or more humidity sensors, configured to track humidity levels; and

[0023] - one or more motion and position sensors, configured to detect the user's movements and position during sleep, which are inertial accelerometers and / or gyroscopes.

[0024] According to a particular embodiment, the pressotherapy system comprises at least one pressotherapy chamber and a pump system that controls the inflation and deflation of the pressotherapy chambers; the vibration therapy system comprises at least one vibration motor; and the thermal therapy system comprises at least one element selected from heating and cooling elements.

[0025] According to a particular embodiment, the mattress comprises a user interface for the manual input of user health information; the mattress being configured to also receive user health information via health platforms and / or external health trackers such as smartwatches or rings, wherein the health information received by the mattress includes activity level, HRV evolution during the day, distant body temperature and / or SpU2 oxygen saturation level.

[0026] According to one particular embodiment, the processing unit is configured to apply a hybrid architecture that combines a global model in the cloud, trained with multiple users, and a local personalization module based on individual physiological response.

[0027] According to one particular implementation, the mattress is configured to connect with other smart devices, such as motorized bases or pillows, and provide a synchronized therapeutic experience.

[0028] The present invention also relates to the rest system that allows the application of the aforementioned therapeutic modalities (selected from pressotherapy, vibration therapy and thermal therapy), particularly during times when the user is not conscious enough to activate said therapies manually, typically during sleep.

[0029] Specifically, the object of the present invention is also a sleep system comprising the mattress defined above. According to one particular embodiment, the data processing unit automatically processes data using AI and determines the most appropriate therapy at any given time based on the sleep or wake phase and the user's specific needs at any given moment. The sleep system is capable of detecting the best therapy for the user in order to improve their conditions, and more generally, the quality of their sleep through combined therapies.

[0030] According to another particular embodiment, the mattress comprises a base support layer, a top layer, and a comfort layer, wherein the top layer comprises:

[0031] - a combination of at least two integrated therapy systems, selected from the group comprising a pressotherapy system, a vibratory therapy system, and a thermal therapy system;

[0032] - a combination of at least two sensors selected from the group comprising environmental sensors, motion and position sensors, and biosignal sensors;

[0033] - and a data processing unit.

[0034] The present invention also relates to a method of operating the mattress or system defined above, wherein: - Physiological data from a user are obtained by means of the biosignal sensor during use of the mattress, the physiological data comprising at least heart rate variability (HRV); the processing unit performs the following actions:

[0035] ■ It establishes a dynamic reference value for the user's heart rate variability, called the HRV reference, based on successive uses of the mattress, and / or based on reference tables according to age, gender, and physical condition.

[0036] ■ Calculates the stress level in real time by comparing HRV with the reference HRV, and by analyzing an LF / HF index;

[0037] ■ infers a sleep stage the user is in from physiological data, preferably using classification algorithms comprising decision trees, nearest neighbor (k-NN) algorithms and / or recurrent neural networks (RNN / LSTM) trained with polysomnography records;

[0038] ■ Determines the automatic activation, adjustment, or interruption of the functioning of therapy systems, including the selection of a single therapy or combination of therapies based on the sleep phase and stress level.

[0039] In one particular implementation, the user's HRV is measured by the biosignal sensor, and the processing unit calculates the stress level in real time by comparing the HRV to the HRV reference. The HRV reference is initialized from population-based reference values ​​according to the user's age, gender, and physical condition, and is dynamically updated based on machine learning and the user's physiological condition over successive uses of the mattress. This ensures a valid HRV reference value is available from the first night of use, preventing false positives in stress detection. As user data accumulates, the HRV reference adapts progressively, reflecting actual physiological changes, allowing for more precise and sophisticated personalization compared to known static methods.According to a more particular embodiment, the processing unit dynamically updates the HRV reference using a moving average of the mattress's recent uses.

[0040] According to one particular embodiment, the processing unit analyzes HRV temporal patterns during sleep by applying RNN / LSTM recurrent neural networks to classify sleep stages and to predict the user's physiological events.

[0041] According to one particular embodiment, the stress level is considered high when HRV is less than 80% of the reference HRV, and the LF / HF ratio is greater than 3. According to one particular embodiment, the processing unit infers the user's sleep stage from physiological data corresponding to HRV, respiration, and movement. According to another particular embodiment, the processing unit infers the user's sleep stage from physiological data corresponding to HRV, respiration, movement, SpO2, and peripheral temperature.

[0042] According to a particular embodiment, the processing unit determines a sympathetic or parasympathetic predominance of the user's autonomic nervous system based on one or more heart rate variability indices selected from the group comprising SDNN, RMSSD, pNN50, LF, HF, LF / HF.

[0043] According to one particular embodiment, the processing unit applies nearest neighbor (k-NN) algorithms to classify the user's stress states and sleep stages.

[0044] According to one particular implementation, user health information is entered through a user interface and / or by connecting the mattress to health platforms, in order to personalize therapies based on the user's daily activity.

[0045] According to a particular embodiment, environmental conditions are monitored by one or more environmental sensors, as well as the user's movements and position during sleep by one or more motion and position sensors.

[0046] According to a particular implementation, the mattress determines the application of therapies based on a global cloud model trained with multiple users and a local personalization module based on an individual physiological response.

[0047] According to a particular embodiment, the processing unit applies random forest techniques to determine the therapy or therapies to be applied.

[0048] According to a particular embodiment, one or more therapies are applied during a drowsy phase of the user, reduced when the user is entering a deep sleep phase, interrupted when the deep sleep phase is detected, and reactivated if physiological or comfort disturbances are identified.

[0049] According to a particular embodiment, during the drowsiness phase and in a high stress level situation, low frequency vibration (20-50 Hz) and cooling of the feet to 2-4 °C less than the user's core body temperature are activated in a combined manner or, alternatively to cooling, localized heating of the feet to 2-4 °C above the basal peripheral temperature, which also induces vasodilation and promotes sleep onset.

[0050] According to a particular implementation, the therapies are controlled with the aim of inducing parasympathetic predominance of the autonomic nervous system and increasing the user's HRV relative to their individual reference.

[0051] The multifunctional mattress of the present invention aims to provide enhanced therapeutic benefits and improve sleep quality, thereby enhancing overall health and well-being, by combining traditional comfort with the advantages of various therapies. In one particular embodiment, the mattress features strategically placed pressure therapy chambers, vibration motors, and heating and cooling elements that can be controlled manually or automatically via artificial intelligence (AI) algorithms to provide personalized treatments. Furthermore, the mattress may include sensors to measure environmental conditions, the user's physiological state, whether they are asleep and, if so, which sleep stage they are in, with connectivity features to input health information and synchronize with other smart devices, such as motorized bases or pillows.

[0052] Thanks to the features of the present invention, a perfect and convenient solution is provided for users seeking continuous therapeutic benefits during rest and improved sleep quality. Unlike existing systems that apply therapies generically and only while the user is conscious, the present invention allows for the application of specific therapies during unconsciousness, particularly in the drowsiness and deep sleep phases, based on objective physiological parameters such as heart rate variability (HRV) and the LF / HF ratio. In this way, the system not only reacts to static conditions but also actively induces parasympathetic dominance, reducing sleep latency and improving overall sleep quality through dynamic therapeutic control adapted to the autonomic nervous system.

[0053] PREFERRED EMBODIMENT OF THE INVENTION

[0054] The present invention relates to a multifunctional mattress and a corresponding sleep system that allow for comprehensive therapy through the use of combined therapies chosen from integrated pressotherapy, vibration therapy, and thermal therapy, by means of advanced physiological monitoring and data analysis by AI algorithms.

[0055] The mattress comprises a base support layer, a top layer, and a comfort layer. The top layer comprises:

[0056] - A combination of at least two integrated therapy systems, selected from the group comprising a pressotherapy system, a vibration therapy system, and a thermal therapy system. According to a particular embodiment, the top layer comprises all three therapy systems.

[0057] - A biosignal sensor, preferably a heart rate variability (HRV) sensor (e.g., a BCG sensor) that is configured to obtain physiological data from a user during use of the mattress. The physiological data includes at least heart rate variability (HRV).

[0058] - A data processing unit, which is configured to:

[0059] ■ establish a dynamic reference value for the user's heart rate variability, called "HRV reference", from successive uses of the mattress, and / or from reference tables based on age, gender and physical condition;

[0060] ■ Calculate the stress level in real time, preferably from a comparison of HRV with the reference HRV, and, optionally, from an analysis of an LF / HF index;

[0061] ■ infer the sleep stage the user is in from physiological data, preferably using classification algorithms comprising decision trees, nearest neighbor (k-NN) algorithms and / or recurrent neural networks (RNN / LSTM) trained with polysomnography recordings;

[0062] ■ Determine the automatic activation, adjustment, or interruption of the operation of therapy systems, including selecting a single therapy or a combination of therapies based on the sleep stage and stress level.

[0063] Preferably, the mattress further comprises one or more biosignal sensors selected from the group comprising a heart rate sensor, a respiratory rate sensor, and a user oxygen saturation sensor.

[0064] According to a particular embodiment, the mattress comprises:

[0065] - one or more environmental sensors, which are:

[0066] ■ one or more temperature sensors, configured to monitor the environmental conditions of a mattress sleeping surface;

[0067] ■ one or more humidity sensors, configured to track humidity levels; and - one or more motion and position sensors, configured to detect the user's movements and position during sleep, which are inertial accelerometers and / or gyroscopes.

[0068] According to another particular embodiment, the mattress comprises a base support layer, a top layer, and a comfort layer, wherein the top layer comprises:

[0069] - a combination of at least two integrated therapy systems, selected from the group comprising a pressotherapy system, a vibratory therapy system, and a thermal therapy system;

[0070] - a combination of at least two sensors selected from the group comprising environmental sensors, motion and position sensors, and biosignal sensors;

[0071] - and a data processing unit.

[0072] According to a particular embodiment, the pressotherapy system comprises at least one pressotherapy chamber and a pump system that controls the inflation and deflation of the pressotherapy chambers; the vibratory therapy system comprises at least one vibratory motor; and the thermal therapy system comprises at least one element selected from heating and cooling elements.

[0073] According to one particular implementation, the processing unit combines a global cloud-based model, trained on data from multiple users, with a local module that dynamically adapts to each user's individual physiological response. This hybrid approach allows for leveraging statistically valuable collective patterns while simultaneously personalizing therapy based on the user's specific characteristics.

[0074] The present invention also relates to a method of operating the mattress or system defined above, according to the following actions:

[0075] - Physiological data is obtained from a user using the biosignal sensor during use of the mattress. Preferably, the physiological data includes at least heart rate variability (HRV).

[0076] - The data processing unit performs the following actions:

[0077] ■ It establishes a dynamic reference value for the user's heart rate variability, called the HRV reference, based on successive uses of the mattress, and / or based on reference tables according to age, gender, and physical condition.

[0078] ■ Calculates the stress level in real time by comparing HRV with the reference HRV, and by analyzing an LF / HF index; ■ Infers the sleep stage the user is in from physiological data using classification algorithms comprising decision trees, nearest neighbor (k-NN) algorithms and / or recurrent neural networks (RNN / LSTM) trained with polysomnography recordings;

[0079] ■ Determines the automatic activation, adjustment, or interruption of the functioning of therapy systems, including the selection of a single therapy or combination of therapies based on the sleep phase and stress level.

[0080] According to a particular embodiment, one or more therapies are applied during a drowsy phase of the user, reduced when the user is entering a deep sleep phase, interrupted when the deep sleep phase is detected, and reactivated if physiological or comfort disturbances are identified.

[0081] According to a particular implementation, the therapies are controlled with the aim of inducing parasympathetic predominance of the autonomic nervous system and increasing the user's HRV relative to their individual reference.

[0082] According to a particular embodiment, the processing unit determines a sympathetic or parasympathetic predominance of the user's autonomic nervous system based on one or more HRV indices selected from the group comprising SDNN, RMSSD, pNN50, LF, HF, LF / HF.

[0083] Preferably, the user's HRV is measured using the biosignal sensor, and the processing unit calculates the stress level in real time by comparing the HRV to the HRV reference. The HRV reference is initialized from population reference values ​​based on the user's age, gender, and physical condition, and is dynamically updated based on machine learning and the user's physiological condition over successive uses of the mattress.

[0084] According to one particular embodiment, the processing unit applies nearest neighbor (k-NN) algorithms to classify the user's stress states and sleep stages.

[0085] According to one particular implementation, the sleep system initially applies therapies based on a global cloud-based model trained with data from multiple users. Subsequently, by observing the individual physiological response (for example, changes in heart rate variability [HRV] after each therapy session), a local module progressively adjusts the parameters to personalize the therapy for the specific user. This continuous cycle of therapy application, results measurement, and progressive adjustment transforms the mattress into an adaptive system capable of optimizing the effectiveness of the therapies.

[0086] Detailed description according to preferred implementation:

[0087] 1. Structure and composition

[0088] The mattress consists of several layers, including a base support layer, a top layer with therapeutic elements and one or more integrated sensors, and a comfort layer. The integrated elements layer can be a separate layer, or the therapeutic and measurement elements can be integrated directly into the surface of the support layer during the manufacturing process. The support layer can be made of any of the materials commonly used in a mattress, such as elastic material like polyurethane foam or latex, springs, or, preferably due to its proven effect on improving sleep quality, air chambers that can adjust the firmness of the support to the individual's needs.

[0089] The pressotherapy chambers, vibrating motors, and heating and cooling elements are made of durable and flexible materials that can withstand repeated use.

[0090] These therapeutic elements are strategically located to target key areas of the body, such as the legs, back, and shoulders.

[0091] 2. Therapeutic systems

[0092] Pressotherapy system: The mattress is equipped with a built-in pump system that controls the inflation and deflation of the pressotherapy chambers.

[0093] The pump system is programmable, allowing users to customize pressure settings, massage patterns, and intervals.

[0094] The system includes safety features such as pressure sensors and automatic shut-off to prevent overinflation and ensure user safety. Vibration therapy system: The mattress includes multiple strategically placed vibration motors to target different parts of the body.

[0095] These motors can be controlled to provide different intensities and vibration patterns, which aids muscle relaxation and pain relief.

[0096] Thermal therapy system: The heating and cooling elements within the mattress allow the user to adjust the temperature according to their preferences, providing soothing warmth or cooling relief as needed. Furthermore, its automatic activation via AI, in different sleep states, promotes and supports the user's varying and changing thermal needs throughout the night, improving sleep quality by preventing interruptions due to excessive or insufficient temperature and reducing sleep latency by assisting in the lowering of body temperature required to initiate sleep.

[0097] 3. Sensors and monitoring:

[0098] Environmental sensors: Temperature sensors monitor the ambient conditions of the sleeping surface, controlling thermoregulation elements as needed. Humidity sensors track humidity levels, which can be used to adjust therapies for optimal effect.

[0099] Motion and position sensors: These detect the user's movements and position during sleep, providing data that can be used to automatically adjust therapies. Such sensors may be inertial accelerometers and / or gyroscopes.

[0100] Biosignal sensors: These measure heart rate, respiratory rate, and heart rate variability (HRV), providing a comprehensive overview of the user's physiological state, including their stress level. HRV is well-established as a useful tool for determining an individual's stress level and the balance between the sympathetic and parasympathetic nervous systems. By monitoring this parameter, various therapeutic systems that induce relaxation can be activated with varying intensity and duration.

[0101] 4. User input and connectivity: The system includes a user interface for the manual entry of health information, including specific pathologies and other vital signs such as blood oxygen saturation level SpU2, or skin temperature that could be collected by other devices, such as external trackers (health trackers) such as a smart watch or ring connected to the mattress either directly through radio technologies (BLE, WiFi) or through health platforms (third-party platforms) such as Apple Health or Google Fit, platforms through which other data could be received regarding the level and type of activity that the person has performed during the day and the evolution of their heart rate, body temperature and HRV during the same.Connectivity with other smart devices, such as motorized bases, pillows, and adjustable beds, allows for a synchronized therapeutic experience, optimizing the benefits of integrated therapies.

[0102] 5. AI-powered personalization:

[0103] The system includes a data processing unit that enables automatic data processing using AI to determine the most appropriate therapy at any given time based on the sleep or wakefulness phase and the specific needs of each user at that particular moment, providing a perfect and convenient solution for users seeking continuous therapeutic benefits during rest and an improvement in sleep quality.

[0104] The mattress uses AI algorithms to analyze collected data and personalize therapeutic treatments. This ensures that therapies are tailored to the user's specific needs, improving their effectiveness. Furthermore, AI allows therapies to be activated and deactivated while the user is unconscious. For example, the AI ​​uses sensors to measure the person's stress level and other factors such as ambient temperature and relative humidity. It also considers information entered by the user, either directly or automatically through third-party health platforms, regarding their daily physical activity and activates specific therapies, such as targeted compression therapy for the calves and thermal therapy to induce sleep by warming the feet.The moment the sensors indicate that the person has fallen asleep, the therapies are turned off and will only be activated during the night if, for example, an extra need for thermoregulation is detected (activating, for example, in this case, thermal therapy) and then to wake the user up when the time programmed by them has approached and the sleeper is in light sleep, through other therapies that induce the activation of, in this case, the sympathetic system.

[0105] 6. Therapeutic benefits:

[0106] The mattress provides continuous, low-intensity compression therapy, which can improve blood circulation, reduce edema, and relieve muscle pain. Vibration therapy relaxes muscles and alleviates pain, while heat therapy provides soothing warmth or cooling relief. Furthermore, a combination of two or more of these elements can activate either the sympathetic or parasympathetic nervous system, which has proven effects on sleep quality, as demonstrated in various studies.

[0107] The sensors monitor the user's physiological state and environmental conditions, including sleep-wake cycles and different sleep stages, allowing for real-time adjustments to therapies. The mattress is particularly beneficial for people with conditions such as lymphedema, chronic venous insufficiency, and those needing muscle recovery, although its effects on improving sleep latency and quality are beneficial for all types of users.

[0108] 7. Portability and convenience:

[0109] If designed as a mattress topper, the product is portable, allowing users to benefit from its therapeutic features even while traveling. Whether it's a mattress or a topper, the product is designed to provide the comfort and support needed for a good night's sleep, which in turn enhances the effectiveness of applied therapies.

[0110] A more detailed description of some aspects of the present invention is included below.

[0111] The present invention is further enhanced by the use of heart rate variability (HRV) as an objective physiological parameter to calculate the user's stress level. The system establishes a baseline HRV value (HRV reference) and compares measurements in real time during sleep, such that a significant reduction in HRV triggers therapies aimed at restoring parasympathetic dominance. This mechanism leverages the relationship between the autonomic nervous system, stress, and HRV, applying it in a novel way in a therapeutic mattress.

[0112] - HRV reference value (HRV reference)

[0113] The HRV baseline (HRV reference) is determined in the initial phase of use by measuring HRV under resting and relaxed conditions (e.g., over several initial nights). For the first session, the sleep system may use a typical HRV value estimated from indicative tables that consider the user's age, sex, and physical condition. Subsequently, this initial HRV baseline is dynamically replaced and updated using a moving average of the user's most recent sessions. The HRV baseline is not fixed but evolves based on the system's learning process and the individual's physiological condition.

[0114] - Detection of the sleep stage the user is in:

[0115] The sleep system combines different physiological signals that present characteristic patterns according to the sleep phase:

[0116] Heart rate (HR): tends to decrease in light and deep sleep, increases in REM and wakefulness.

[0117] Heart rate variability (HRV): increases in deep sleep, decreases in REM and wakefulness.

[0118] Respiratory rate (RR): remains average, but more regular during deep sleep. Respiratory variability (RRV): low during deep sleep, increases during REM and wakefulness.

[0119] Respiratory depth (Rdepth): decreases in deep sleep and increases in REM and wakefulness.

[0120] These parameters are typically collected using a BCG sensor integrated into the mattress, and are optionally supplemented with SpO2 and peripheral temperature sensors.

[0121] In practice, reference tables such as the following can be used, which summarizes the trend of each physiological parameter in the different sleep phases.

[0122] To automatically differentiate between wakefulness, drowsiness, light sleep, deep sleep, and REM sleep, the system integrates these signals into classification models:

[0123] In a basic implementation, decision trees trained with simple physiological rules are used.

[0124] In a preferred embodiment, recurrent neural networks (RNN) or LSTMs are applied, capable of processing temporal sequences of HRV and respiration, which allows for more precise identification of transitions between sleep phases.

[0125] These models are initially trained using reference recordings from polysomnography studies (the gold standard), which provide validated (“ground truth”) sleep stage annotations. From this foundation, the system progressively adapts the models with each user's actual data, refining stage classification in a personalized way and improving detection accuracy.

[0126] - Specific therapy that is applied based on the level of stress and the sleep phase detected:

[0127] The sleep system determines the appropriate therapy by combining physiological stress levels and the detected sleep stage. The decision is based on quantitative thresholds obtained from HRV, the LF / HF ratio, and respiratory parameters.

[0128] Definition of physiological stress levels

[0129] The sleep system classifies stress into three levels (high, moderate, low) based on the combination of:

[0130] • Relative decrease in HRV compared to the individual HRV reference.

[0131] • LF / HF ratio as a marker of sympathetic predominance.

[0132] • Respiratory rate and its variability as a complementary indicator.

[0133] Example of classification (orientative values):

[0134] • High stress: HRV <80% of reference HRV and LF / HF >3.

[0135] • Moderate stress: HRV between 80-95% of the reference HRV or LF / HF between 2-3.

[0136] • Low stress: HRV >95% of reference HRV and LF / HF <2.

[0137] High stress: activate intensive parasympathetic therapies.

[0138] Moderate stress: apply a single therapy according to the predominant variable. Low stress: do not apply active therapies, only monitoring.

[0139] Depending on the sleep stage:

[0140] During drowsiness: apply vibration or cold to the feet to accelerate sleep onset or, alternatively to cold, localized heat to the feet to induce peripheral vasodilation and reduce sleep latency.

[0141] During light sleep: apply pressotherapy or localized heat, avoiding interference with deep sleep.

[0142] During deep sleep: intervention is minimized, except in special situations (e.g., chronic pain).

[0143] - Decision on whether to apply several therapies or just one:

[0144] The rest system employs a hierarchical decision model that determines whether single or combined therapies are applied according to the intensity of stress and the sleep phase: High stress + drowsiness phase — ► combination of vibration (20-50 Hz) + cold thermotherapy on feet (2-4 °C less than the core temperature) or, alternatively to cold thermotherapy, hot thermotherapy on feet (2-4 °C above the basal peripheral temperature).

[0145] High stress + light sleep phase — ► combination of sequential pressure therapy (30-40 mmHg) + lumbar heat (36-39 °C).

[0146] Moderate stress — ► single therapy selection: vibration if low HRV predominates, thermotherapy if high LF / HF predominates, pressotherapy if muscle fatigue is present.

[0147] Low stress — ► no active therapy is applied.

[0148] These rules are formalized using decision trees or random forest models trained with records from multiple users, allowing the therapeutic combination to be adapted based on observed patterns.

[0149] - Specific therapy or therapies that are applied depending on the circumstances:

[0150] Vibration: applied at low frequencies (20-50 Hz), amplitude <2 mm, induces muscle relaxation and improves peripheral circulation. Indicated when HRV drops significantly during numbness.

[0151] Thermal: • Cold feet (2-4 °C less than core temperature): promotes peripheral vasodilation and sleep onset.

[0152] • Heat in feet (2-4 °C higher than basal peripheral temperature): causes active peripheral vasodilation and reduction of sleep latency.

[0153] • Lumbar heat (36-39 °C): reduces pain, relaxes muscles and increases comfort.

[0154] Pressotherapy: sequential pneumatic cycles (30-50 mmHg) on ​​the extremities, lasting 15-20 minutes, promotes venous return and muscle recovery. Indicated when high muscle strain or fatigue is detected.

[0155] - Decision trees:

[0156] A simplified decision tree for therapy allocation can be structured in hierarchical nodes:

[0157] Node 1: HRV vs reference HRV.

[0158] Node 2: Sleep phase (wakefulness / drowsiness / deep).

[0159] Node 3: LF / HF index.

[0160] Sheet: specific therapeutic action (e.g. vibration, cold, heat, pressotherapy or monitoring).

[0161] In preferred implementations, random forests are used, which combine multiple trees trained with user records. Each tree generates a therapy proposal, and the final decision is made by consensus, improving robustness and reducing errors.

[0162] - Usefulness of analyzing HRV temporal patterns using RNN / LSTM

[0163] Recurrent neural networks (RNNs) and LSTM (Long Short-Term Memory) variants allow for the analysis of HRV, respiration, and movement sequences, providing predictive capabilities. They are used both for sleep stage classification and for the early prediction of physiological events.

[0164] They anticipate decreases in HRV before a stress episode occurs.

[0165] They predict sleep phase transitions (e.g. from drowsiness to REM).

[0166] They adjust therapies preventively and individually, avoiding unnecessary interventions. Unlike decision trees (which are static rules), RNN / LSTM adapts dynamically and improves with use, learning each user's individual patterns from their nightly recordings. This innovative, multifunctional mattress offers a unique combination of comfort and therapeutic benefits. By providing a convenient and effective solution for integrated therapies, this mattress has the potential to improve the quality of life for users with various conditions and those seeking greater well-being during rest and sleep. Sensors and AI-powered personalization further enhance the user experience, making it a revolutionary product in the field of sleep and health technology.

Claims

CLAIMS 1. Multifunctional mattress comprising a base support layer, a top layer and a comfort layer, characterized in that said top layer comprises: - a combination of at least two integrated therapy systems, selected from the group comprising a pressotherapy system, a vibratory therapy system, and a thermal therapy system; - a biosignal sensor, which is configured to obtain physiological data from a user during use of the mattress, the physiological data comprising at least heart rate variability (HRV); - and a data processing unit, which is configured to: ■ establish a dynamic reference value for the user's heart rate variability, called the HRV reference, from successive uses of the mattress, and / or from reference tables based on age, gender and physical condition; ■ Calculate the stress level in real time by comparing HRV with the reference HRV, and by analyzing an LF / HF index; ■ infer a sleep stage the user is in from physiological data, using classification algorithms comprising decision trees, nearest neighbor (k-NN) algorithms and / or recurrent neural networks RNN / LSTM trained with polysomnography recordings; ■ Determine the automatic activation, adjustment, or interruption of the operation of therapy systems, including selecting a single therapy or a combination of therapies based on the sleep stage and stress level.

2. Mattress according to claim 1, wherein the biosignal sensor is a heart rate variability sensor; and the processing unit is configured to calculate the stress level in real time by comparing the HRV with the HRV reference, wherein the HRV reference is initialized from population reference values ​​based on the user's age, gender, and physical condition, and is dynamically updated based on machine learning and the user's physiological condition over successive uses of the mattress.

3. Mattress according to claim 2, wherein the biosignal sensor is a BCG sensor.

4. Mattress according to any of the preceding claims, comprising one or more biosignal sensors selected from the group comprising heart rate sensor, respiratory rate sensor, and user oxygen saturation sensor.

5. Mattress according to any of the preceding claims, comprising: - one or more environmental sensors, which are: ■ one or more temperature sensors, configured to monitor the environmental conditions of a mattress sleeping surface; ■ one or more humidity sensors, configured to track humidity levels; and - one or more motion and position sensors, configured to detect the user's movements and position during sleep, which are inertial accelerometers and / or gyroscopes.

6. Mattress according to any of the preceding claims, wherein the pressotherapy system comprises at least one pressotherapy chamber and a pump system that controls the inflation and deflation of the pressotherapy chambers; the vibration therapy system comprises at least one vibration motor; and the thermal therapy system comprises at least one element selected from heating and cooling elements.

7. Mattress according to any of the preceding claims, comprising a user interface for the manual input of user health information; the mattress being configured to also receive user health information through health platforms and / or external health trackers such as smartwatches or rings, wherein the health information received by the mattress includes activity level, HRV evolution during the day, distant body temperature and / or SpU2 oxygen saturation level.

8. Mattress according to any of the preceding claims, wherein the processing unit is configured to implement a hybrid architecture that combines a global model in the cloud, trained with multiple users, and a local personalization module based on individual physiological response.

9. Mattress according to any of the preceding claims, which is configured to connect with other smart devices, such as motorized bases or pillows, and provide a synchronized therapeutic experience.

10. A sleep system comprising a mattress according to any one of the preceding claims, wherein the data processing unit automatically processes data using AI and determines the most appropriate therapy at any given time based on the sleep or wakefulness phase and the specific needs of the user at any given time.

11. Method of operation of a mattress according to any one of claims 1 to 9, or of a sleep system according to claim 10, characterized in that: - physiological data is obtained from a user using the biosignal sensor during use of the mattress, the physiological data comprising at least heart rate variability (HRV); - The processing unit performs the following actions: ■ establishes a dynamic reference value for the user's heart rate variability, called the HRV reference, based on successive uses of the mattress, and / or based on reference tables according to age, gender and physical condition; ■ Calculates the stress level in real time by comparing HRV with the reference HRV, and by analyzing an LF / HF index; ■ infers a sleep stage the user is in from physiological data using classification algorithms comprising decision trees, nearest neighbor (k-NN) algorithms and / or recurrent neural networks (RNN / LSTM) trained with polysomnography recordings; ■ Determines the automatic activation, adjustment, or interruption of the functioning of therapy systems, including the selection of a single therapy or combination of therapies based on the sleep phase and stress level.

12. Method according to claim 11, wherein the user's HRV is measured by the biosignal sensor, and the processing unit calculates the stress level in real time by comparing the HRV with the HRV reference, wherein the HRV reference is initialized from population reference values ​​based on the user's age, gender, and physical condition, and is dynamically updated based on machine learning and the user's physiological condition over successive uses of the mattress.

13. Method according to claim 12, wherein the processing unit updates the HRV reference dynamically by means of a moving average of the last uses of the mattress.

14. Method according to any of claims 11 to 13, wherein the stress level is considered high when the HRV is less than 80% of the reference HRV, and the LF / HF index is greater than 3.

15. Method according to any of claims 11 to 14, wherein the processing unit analyzes temporal patterns of HRV during sleep by applying recurrent neural networks (RNN / LSTM) to classify sleep stages and to predict physiological events of the user.

16. Method according to any of claims 11 to 15, wherein the processing unit infers the user's sleep stage from physiological data relating to HRV, respiration, and movement.

17. Method according to any of claims 11 to 15, wherein the processing unit infers the user's sleep stage from physiological data corresponding to HRV, respiration, movement, SpO2 and peripheral temperature.

18. Method according to any of claims 11 to 17, wherein the processing unit determines a sympathetic or parasympathetic predominance of the user's autonomic nervous system based on one or more HRV indices selected from the group comprising SDNN, RMSSD, pNN50, LF, HF, LF / HF.

19. Method according to any of claims 11 to 18, wherein the processing unit applies nearest neighbor (k-NN) algorithms to classify the user's stress states and sleep stages.

20. Method according to any of claims 11 to 19, wherein user health information is entered through a user interface and / or by connecting the mattress to health platforms, in order to personalize therapies based on the user's daily activity.

21. Method according to any of claims 11 to 20, wherein the environmental conditions are monitored by means of one or more environmental sensors, as well as the movements and position of the user during sleep by means of one or more motion and position sensors.

22. Method according to any of claims 11 to 21, wherein the mattress determines the application of therapies based on a global cloud model trained with multiple users and a local personalization module based on an individual physiological response.

23. Method according to any of claims 11 to 22, wherein the processing unit applies random forest techniques to determine the therapy or therapies to be applied.

24. Method according to any of claims 11 to 23, wherein one or more therapies are applied in a drowsy phase of the user, are reduced when the user is entering a deep sleep phase, are interrupted when the deep sleep phase is detected, and are reactivated if physiological or comfort disturbances are identified.

25. Method according to claim 24, wherein during the drowsiness phase and in a high stress level situation, low frequency vibration (20-50 Hz) and cooling of the feet to 2-4 °C less than a core body temperature of the user are activated in a combined manner or alternatively to cooling, localized heating of the feet to 2-4 °C above the basal peripheral temperature.

26. A method according to any of the preceding claims, wherein the therapies are controlled with the aim of inducing parasympathetic predominance of the autonomic nervous system and increasing the user's HRV relative to their individual reference.

Images