Adaptive Digital Scent Technology for Improved Sleep Quality

Abstract

A smell delivery system 1010 comprising: an odour chamber 1052 containing a scent substance; an outlet connected to the odour chamber and configured to release the substance; and a control unit. The c

Description

The present disclosure relates to a smell delivery system, in particular to a smell delivery system for sleep. The present disclosure also relates to a method of delivering an olfactory output, in particular to a method of delivering an olfactory output for sleep improvement Insufficient sleep is a common problem, affecting up to 60% of the adult population. Abnormal sleeping patterns can have significant adverse effects on mental or physical health and performance. Sleep can be characterised by four distinct stages (phases) which change throughout the night, as sleepers move between states, usually in a cyclical order. There are typically several cycles of sleep stages (three to five a night). The sleep stages include NREM (non-rapid eye movement) or REM stages. In particular, the four stages are NREM stages 1 to 3 and REM. The person usually cycles through the sleep stages repeatedly, with each full cycle typically lasting about 90 to 110 minutes. Stage 1: The person transitions from being awake to being asleep. Sensations include a loss in awareness of surroundings (the feeling of drowsiness when not completely awake). The person can be easily woken from this state, which usually lasts 5 to 10 minutes. Stage 2: Sleeping but not particularly deep (easy to wake). The heart rate, breathing rate, and brain activity slows down, and the body completely relaxes. The person usually spends about half the night sleeping in this state, which usually lasts 10 to 25 minutes at a time. Stage 3: Deep, slow wave sleep (SWS). Research indicates that, during this time, the body renews and repairs itself. It can take up to half an hour to reach this deepest part of sleep and it takes far more effort to wake up. The person’s breathing rate becomes more regular, blood pressure falls, and the pulse rate slows. The amount of deep sleep varies with age, with older individuals typically experiencing a decrease in deep sleep and an increase in lighter sleep. Rapid Eye Movement (REM): During this stage, the person’s eyes move rapidly beneath closed eyelids, and the person experiences most of their dreams. The mind races, while the body is virtually paralysed. It is believed that this stage facilitates learning and if woken from this state, the person tends to remember that they were dreaming. This can happen particularly if REM is followed by light sleep (i.e., the start of a new cycle). Good sleep health requires a balance between deep, light, and REM sleep. Most healthy adults require 7 to 9 hours of sleep, with experts recommending 8 hours. However, there is great variability, with some people requiring only 6 hours, but others may require 10 hours of quality sleep. If deep sleep is restricted, a person may wake up unrefreshed, no matter how long they have been trying to sleep. The progression from light sleep to deep sleep may be more rapid for people who have been sleep deprived. Stress, including that created by workplace activities, can detrimentally affect sleep quality. The health consequences due to sleep disorders include obesity, diabetes, heart diseases, hypertension, mood disorders, weakened immune system, and increased mortality risk. The relationship between mental health and sleep is an active area of research. Neurochemistry studies indicate that good quality sleep helps enhance both mental and emotional resilience. Poor sleep quality may create negative emotions, heightening anxiety, while treating sleep disorders may help alleviate symptoms of anxiety disorder. Chronically disrupted sleep patterns increase worker absences, and sleep deprivation is believed to cost the UK economy £37 billion a year in lost productivity (RAND Europe, 2016). In addition, driver fatigue due to poor sleep health may be a contributory factor in a significant number of road accidents, and up to one quarter of fatal and serious accidents in the UK. Despite this heightened awareness of sleep health, the number of people with insufficient sleep or sleep disorders has continuously increased. In 2022, it was reported that 1 in 10 people in the UK were only getting 2 to 4 hours of sleep per night. Poor sleep health is particularly prevalent in young adults, with those aged 35 to 44 getting the least sleep, with almost 50% only getting 5 to 6 hours per night and only 33% getting the recommended 7 to 8 hours. A significant number of people with sleep disorders go undiagnosed due to the lack of monitoring and diagnosis. Although sleep disorders are widespread, they are easily dismissed because their symptoms are hard to detect while asleep. Research shows that sleep and the olfactory system are interconnected and share some neural substrates. In fact, it has been shown that odours can affect different aspects of sleep: the right smell at the right time can help facilitate falling asleep, maintaining sleep, and can increase the quality and duration of sleep. Conventionally, the use of essential oils to treat sleep disorders is increasing, both in domestic and healthcare settings. Lavender oil has long been used to create a calming effect and studies have indicated that exposure can be used to improve sleep quality. Recently, a synthetic jasmine scent known as Vertacetal Coeur (VC) has been shown to have a sedative effect. The scent contains 1,3 dioxanes which are known to potentiate GABA-induced inhibitory channel currents and appear to act in a similar manner to benzodiazepines. However, current efforts to use smells to improve sleep quality merely involve static smells, for example where a single smell is used (such as lavender). This means that the smell delivery cannot be adapted or adjusted. This also means that the smell delivery cannot be personalised to the user based on their sleep state or individual reactions to smell delivery. The present disclosure seeks to address one or more of the above problems. Aspects of the invention are set out in the independent claims and preferred features are set out in the dependent claims. According to a first aspect of the present disclosure, there is provided a smell delivery system, comprising: a smell delivery device for delivering an olfactory output, the smell delivery device comprising: an odour chamber for containing a substance configured to produce an olfactory output; and an outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device; and a control unit, comprising: an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user; a processor configured to: determine a sleep state of the user based at least on the sensor data; and determine settings for delivery of an olfactory output based at least on the sleep state of the user; and an output interface configured to provide instructions to the smell delivery device corresponding to the settings; wherein the smell delivery device is configured to deliver the olfactory output based on the instructions. The smell delivery system of the present disclosure is generally provided to deliver an olfactory output, for example to a user. In particular, the smell delivery system can be intended for use while the user is sleeping. Generally, sensor data is used to determine a sleep state of the user (for example, whether the user is asleep, or which particular sleep stage they are in), and the smell delivery settings can be adjusted based on that sleep state. This can be used to adapt the smell delivery based on the user’s sleep state. For example, if the user is in a deep sleep state, the system may provide a particular smell for providing a calming effect and maintaining deep sleep. Alternatively, a particular smell may be used to promote waking the user up or transitioning to a different sleep stage. In the first aspect, the smell delivery system comprises a smell delivery device and a control unit. As described herein, the control unit may be integrated with the smell delivery device, for example provided within the same physical assembly. For example, the control unit may be integrated into the smell delivery device. In such examples, the control unit may be the control unit for the smell delivery device and may be able to control the smell delivery device to deliver the olfactory output. In other examples, the control unit may be separate from the smell delivery device, and for example the control unit may be remote from the smell delivery device. In such examples, the control unit may communicate with the smell delivery device over a wired or wireless connection. In the first aspect, the smell delivery device is for delivering an olfactory output. The smell delivery device may otherwise be referred to as a smell delivery module or a smell delivery component. As used herein, the term “olfactory output” preferably refers to a substance having an odour, a smell, a scent, or a fragrance. For example, an olfactory output can be perceived by a user (e.g., a human or animal, preferably a human) through their olfactory system (their sense of smell). In other words, the substance is able to be detected by olfactory receptors as an odour or smell. Alternatively, the substance may be referred to as a substance configured to produce an odour or a smell. Thus, the smell delivery device may be for delivering an odour or a smell. In some examples, the substance may be referred to as an odorant or an aroma compound. The substance may therefore be a chemical (e.g., a chemical compound). The substance may be capable of being carried to the olfactory receptors of a user, such as by a carrier gas (e.g., air). For example, the substance may volatilise (or vaporise) into the odour chamber so that the substance can be carried by a carrier gas. For example, the substance may be a volatile compound so that the substance readily volatilises. In other examples, the substance may be in gaseous form for mixing with a carrier gas. In other examples, the substance may be heated or otherwise treated or affected in order to volatise or promote volatilisation. In the first aspect, the smell delivery device comprises an odour chamber for containing a substance configured to produce an olfactory output. As used herein, the term “chamber” preferably refers to a region or a container. For example, the odour chamber can be a region in which the substance is contained. The odour chamber may have various shapes and sizes. For example, the odour chamber may be in the form of a pipe through which a carrier gas (e.g., air) is passed to pick up the substance for delivery from the smell delivery device. In the first aspect, the odour chamber is suitable for containing a substance configured to produce an olfactory output. In some examples, the odour chamber contains the substance. In other examples, the odour chamber may be configured to receive a removable container which contains the substance. For example, the odour chamber may be configured to receive a removable container (e.g., a removable canister) comprising the substance, and wherein the removable container is configured to release the substance into the odour chamber. For example, the odour chamber may be provided without a substance, and a substance may be loaded into the odour chamber, for example by receiving a removable container. The removable container can be removed and replaced by another container which may have the same substance, a different substance (such as to produce a different odour), or no substance at all for generating odourless carrier gas. In some examples, the smell delivery device may further comprise the removable container loaded into the odour chamber. In some examples, the container is not removable, but the entire odour chamber is removable and replaceable. In other examples, the substance is provided in the odour chamber, such as integrally within the odour chamber. Thus, in some examples, the odour chamber may be configured to contain a substance configured to produce an olfactory output. In some examples, the odour chamber is suitable for containing a plurality of substances each configured to produce an olfactory output. For example, the odour chamber may contain a plurality of substances. In other examples, the odour chamber may be configured to receive a plurality of removable containers. Different substances may be used to produce different olfactory outputs. For example, multiple containers may be coupled to the odour chamber at the same time, so that different containers can be selected at different times (for example, by operating valves). In this manner, different substances can be delivered as required without physically replacing the containers. In other examples, multiple odour chambers may be provided, for example each with a different substance. In some examples, each odour chamber may be configured to receive a removable container. In the first aspect, the smell delivery device also comprises an outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device. The outlet may otherwise be referred to as an opening, a hole, or an exit. The outlet may be an opening in the odour chamber to allow the substance to leave the odour chamber. In some examples, the outlet may comprise an outlet channel to direct the substance from the odour chamber. Thus, an outlet channel may be connected to the odour chamber, and the outlet channel may comprise the outlet (at the other end of the outlet channel). The substance from the odour chamber can pass through the outlet for release from the smell delivery device (for example, after passing along the outlet channel). In other words, the outlet may be in communication with an external environment from the smell delivery device. In examples with multiple odour chambers, each odour chamber may have an outlet, and each of the outlets may deliver the olfactory output from the odour chamber from the smell delivery device (e.g., to the external environment). In some examples, the multiple outlets may be connected via an outlet manifold to deliver each olfactory output through a combined outlet (allowing for mixing of smells, or delivery of individual smells through the same outlet). In some examples, the outlet channel may direct the olfactory output towards the user, such as towards the headspace of the user (e.g., towards their head, or towards their nose). For example, the olfactory output may be directed towards the bed, or specifically towards the head-end of the bed, to provide the olfactory output near where the user’s nose will be positioned during sleeping. In the first aspect, the control unit comprises an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user. The input interface preferably refers to any component which is capable of receiving input data (specifically, at least sensor data). In some examples, the input interface is configured to receive the sensor data at an external connection. For example, if the sensor is remote to the control unit, the input interface may comprise a receiver which is configured to receive the sensor data, e.g., over a wireless connection. In some examples, the input interface may receive the sensor data over a wired connection from an external sensor (for example, a removable connection such as USB). In some examples, the input interface is configured to receive the sensor data at an internal connection. For example, if the sensor is integral with the control unit, the sensor data may be received by a direct connection (such as internal wiring or electrical connections, for example a permanent connection). For example, the sensor may provide the sensor data directly to the control unit. The sensor may also be integrated with a second device, such as a watch or headband. In the first aspect, the sensor data is received from at least one sensor. For example, there may be a plurality of sensors. In some examples, there may be two or more sensors, such as three or more sensors, or even four or more sensors. Each sensor may be different. For example, these sensors may each measure a signal corresponding to a different parameter. For example, the sensors may measure acceleration, position, and heart rate. Providing sensor data from multiple different sensors can provide additional context which allows for improved processing. For example, sensor data from multiple sensors can be used to more accurately determine a particular sleep state, such as a sleep stage. In the first aspect, the sensor data is received from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of the user. The sleep state may refer to whether the user is generally awake or asleep, or it may refer to a specific sleep stage of the user. The parameter may be used to indicate the sleep state of the user. For example, the parameter may depend on the sleep state of the user. For example, the heart rate of a user may depend on the sleep state of the user. In particular, a low heart rate can indicate that the user is asleep, or it may be used in determining a particular sleep stage. In other words, the parameter may be said to depend on, correspond to, or correlate to, the sleep state. In some examples, the sensor data is real-time sensor data. Thus, the sensor may measure the signal and provide the sensor data to the control unit in real-time. This allows live analysis and adaptive smell delivery. The signal measured by the sensor is indicative of the parameter. The signal measured by the sensor refers to the quantity that is detected. For example, a heart rate sensor may be configured to measure an electrical signal that is indicative of the heart rate. In this case, the electrical signal is the signal measured by the sensor, and the heart rate is the parameter - where the signal (electrical signal) is indicative of the parameter (heart rate). The electrical signal is directly related to the heart rate in this case since the electrical signal is a direct representation of the heart rate. Therefore, in some examples, the signal may be directly related to a parameter associated with the sleep state. In other words, the signal may be directly indicative of the parameter. Said another way, the signal may be a direct measure of the parameter. For example, the electrical signal of the heart rate sensor is a direct measurement of the heart rate. In other examples, the signal may be indirectly related to the parameter associated with the sleep state. For example, a microphone (or sound sensor) may be used to detect a signal corresponding to sound. Therefore, the sensor (microphone) may measure a signal indicative of a parameter (sound). The sound may in turn be associated with another parameter - specifically, a parameter associated with a sleep state. In particular, the sound may correspond to a breathing sound, or snoring sound, of the user, or it may correspond to sounds related to movement of the user or sounds related to movement of bedsheets that indicate movement of the user. In these cases, the sound is indicative of a parameter associated with a sleep state. In particular, the sound is indicative of breathing of the user or movement of the user, which depends on the sleep state of the user. Therefore, the signal may be indirectly indicative of the parameter. In other words, the signal may be a direct measure of a particular parameter, and that parameter may in turn be indicative of another parameter associated with the sleep state of the user. For example, the microphone signal is a direct measurement of the sound of a user, which in turn is an indirect measurement of the breathing or movement of the user. In some examples, the sensor is not part of the smell delivery system, and the sensor data is received from a sensor external to the smell delivery system. In other examples, the sensor is part of the smell delivery system. Thus, in some cases, the sensor may be provided. The sensor may be configured to measure a signal indicative of a parameter associated with a sleep state of the user. For example, the sensor may be an EEG (electroencephalogram) sensor that measures brain activity (indicative of sleep state), for example attached to a headband worn by the user. In another example, the sensor may be a PPG (photoplethysmography) sensor for monitoring heart rate, for example integrated in a smartwatch attached to the wrist of the user. In the first aspect, the control unit also comprises a processor. The processor may otherwise be referred to as a processing unit or processing component. In the first aspect, the processor is configured to determine a sleep state of the user based at least on the sensor data. The processor may therefore process the sensor data and analyse it. The processor may determine the sleep state by analysing the sensor data. Since the sensor data relates to a signal indicative of a parameter associated with the sleep state, the sleep state can be derived from the sensor data. For example, by analysing the sensor data and determining that the user’s heart rate has dropped, the processor may determine that the user has fallen asleep. In some examples, the processor may determine the sleep state based on pre-programmed correlations between the parameter and the sleep state. For example, when the heart rate drops below a threshold, the sleep state may be determined to be an “asleep” state. Alternatively, the sleep state may be determined based on historical sleep data or user input, and optionally using machine learning, as explained herein. In the first aspect, the processor is configured to determine settings for delivery of an olfactory output based at least on the sleep state of the user. Once the processor has determined a sleep state, it can determine corresponding settings for delivery of an olfactory output. For example, if it has been determined that the user is asleep, it may be desirable to provide a particular smell. For example, this can be used to improve sleep quality. In other examples, an adjustment to smell delivery (e.g., selection of different smells at different times or providing different intensities) can be used to coincide with different stages of sleep to improve sleep quality at a more granular level. In some examples, the processor may determine the settings for delivery of the olfactory output based on pre-programmed correlations between the sleep state and the settings. For example, when the user is determined to be asleep, the settings may correspond to providing a lavender smell. Alternatively, the settings may be determined based on historical sleep data or user input, and optionally using machine learning, as explained herein. In the first aspect, the control unit also comprises an output interface configured to provide instructions to the smell delivery device corresponding to the settings. The output interface preferably refers to any component which is capable of providing output data (specifically, at least instructions to the smell delivery device). In some examples, the output interface is configured to transmit the instructions via an external connection. For example, if the smell delivery device is remote to the control unit, the output interface may comprise a transmitter which is configured to transmit the instructions, e.g., over a wireless connection. In some examples, the output interface may provide the instructions over a wired connection to an external location (for example, a removable connection such as USB). In some examples, the output interface is configured to provide the instructions via an internal connection. For example, if the smell delivery device is integral with the control unit, the instructions may be provided by a direct connection (such as internal wiring or electrical connections, for example a permanent connection). For example, the control unit may provide the instructions directly to the smell delivery device. In some examples, the output interface may be the same as the input interface. For example, a connector may be configured to receive the sensor data as well as provide the instructions. In other words, the control unit may comprise an input / output connection. This may be a wired connector, or it may be a wireless transceiver. In the first aspect, the smell delivery device is configured to deliver the olfactory output based on the instructions. In response to receiving the instructions, the smell delivery device may deliver the olfactory output. In other words, the substance may be released from the outlet. For example, the substance may be released into the odour chamber and delivered out of the outlet to the environment external of the smell delivery device. In some examples, this may involve operating a flow generator (such as a fan or pump) to deliver the olfactory output at a controlled rate. The olfactory output is delivered based on the instructions, and therefore the settings can be applied at the smell delivery device. For example, the instructions may comprise instructions to deliver a particular olfactory output. This may involve selection of a particular substance by the smell delivery device, such as by operating various valves. In other examples, the instructions may comprise instructions to deliver the olfactory output at a particular flow rate or intensity. In this case, the smell delivery device may operate the flow generator at a particular flow rate. In some examples, the instructions may comprise instructions to deliver a particular olfactory output. For example, the instructions may instruct the smell delivery device to deliver a lavender smell. In response to the instructions, the smell delivery device may deliver a lavender smell, for example by selecting an appropriate container for delivering that smell. For example, the smell delivery device may comprise a processor to determine a configuration of the smell delivery device that corresponds to the instructions. In this case, it is selection of a particular container for delivering the desired smell, and the smell delivery device may then operate valves as necessary to deliver the smell. In this manner, the control unit does not need to be aware of the particular configuration of the smell delivery device, such as which containers need to be used to deliver a particular smell. In other examples, the control unit may determine the configuration of the smell delivery device, for example which container needs to be used to deliver the desired smell. In some examples, the processor of the smell delivery device can be the same as the processor of the control unit. The smell delivery system of the present disclosure allows for flexible delivery of an olfactory output depending on the user and their sleep. In particular, the smell delivery can be adaptive and dynamic, and can be controlled depending on the user’s sleep state. This allows improved control of smell delivery for improved impact on the user’s sleep. Optionally, the parameter associated with a sleep state of the user comprises a physiological parameter associated with a sleep state of the user. As used herein, preferably the term “physiological parameter” refers to a parameter associated with an internal characteristic of the user’s body, in particular to a function of the user’s body, such as a physiological function. The parameter may otherwise be referred to as a biological parameter or a medical parameter. In one example, the parameter corresponds to a biosignal. Optionally, the physiological parameter comprises: a heart rate of the user, a blood pressure of the user, a blood volume of the user, a blood oxygen saturation of the user, brain activity of the user, a body temperature of the user, sweating of the user, body composition of the user, and / or breathing of the user. The physiological parameter may include one or more of these parameters (such as a plurality of these parameters). For example, multiple sensors may be used, for example to indicate multiple physiological parameters, such as those listed above. In some examples, the parameter may be heart rate of the user. For example, the heart rate of the user may be measured by a heart rate sensor. The heart rate sensor may be configured to measure a signal indicative of the heart rate of the user. For example, the signal may be an electrical signal. Thus, the heart rate sensor may be an electrical heart rate sensor. For example, this may involve electrocardiography (ECG or EKG). The voltage of the signal may correspond to the heart’s electrical activity. The signal may be detected by placing electrodes on the user’s chest. In some examples, a raw heart rate sensor may be used. In some examples, the signal may be an optical signal. Thus, the heart rate sensor may be an optical heart rate sensor. For example, this may involve photoplethysmography (PPG). In some examples, the heart rate may be inferred by measuring a pulse rate of the user. For example, pulse rate sensors (such as optical sensors) can measure the pulse rate (such as on the user’s wrist). This is particularly preferable since it is less invasive than requiring electrodes on the user’s chest. The sensor may also measure other heart-related parameters. In some examples, the parameter may be blood pressure of the user. For example, the blood pressure may be measured by a blood pressure sensor. The blood pressure sensor may be configured to measure a signal indicative of the blood pressure. For example, the blood pressure sensor may include a blood pressure monitor, or a blood pressure gauge. This typically has an inflatable cuff for the arm, or for the wrist or even the finger. In other examples, the blood pressure sensor may be cuffless, such as an optical or bioimpedance sensor. In some examples, the parameter may be blood volume of the user. For example, the blood volume may be measured by a blood volume sensor. The blood volume sensor may be configured to measure a signal indicative of the blood volume. In particular, the blood volume may be measured by a PPG sensor. This may use optical techniques to measure the blood volume. For example, a pulse oximeter may be used to obtain a photoplethysmogram. In some examples, the parameter may be blood oxygen saturation of the user. For example, the blood oxygen saturation may be measured by a blood oxygen saturation sensor. The blood oxygen saturation sensor may be configured to measure a signal indicative of the blood oxygen saturation. For example, the blood oxygen saturation may be measured using non-invasive techniques such as peripheral oxygen saturation (SpO2). In particular, the blood oxygen saturation may be measured by a pulse oximeter. This involves optical techniques. The pulse oximeter may be used to produce a PPG. The pulse oximeter can be used to provide information relating to blood oxygen saturation, as well as blood volume, and heart rate. In some examples, the sensor may comprise an irregular heart rhythm notification (IHRN) sensor that detects irregular heartbeats. For example, the parameter may be heart rhythm (such as using an ECG). The IHRN may be implemented as part of a heart rate sensor, a blood pressure sensor, a blood volume sensor, and / or a blood oxygen saturation sensor. In some examples, the parameter may be brain activity of the user. For example, the brain activity may be measured by a brain activity sensor. The brain activity sensor may be configured to measure a signal indicative of the electrical brain activity. In particular, electroencephalography (EEG) may be used to measure electrical activity of the brain. The signal may be detected by placing electrodes on the user’s head. In some examples, the parameter may be body temperature of the user. For example, the body temperature may be measured by a temperature sensor. The temperature sensor may be configured to measure a signal indicative of the body temperature. For example, the body temperature may be measured by a thermistor. In other examples, the body temperature may be measured by an infrared detector. In some examples, a sleep cycle monitor (such as a smartwatch) may include a temperature sensor. In some examples, the parameter may be sweating of the user. The sweating may be determined by measuring electrodermal activity (EDA). This may be referred to as skin conductance, such as galvanic skin response. For example, the sweating may be measured by a skin conductance sensor. The skin conductance sensor may be configured to measure a signal indicative of the electrical conductance of the skin. Skin resistance varies based on the state of sweat glands, with increased sweat gland activity from physiological (or psychological) arousal leading to increased skin conductance. The skin conductance may also be indicative of a stress level of the user. In some examples, a sweat loss sensor may be used to measure fluid loss from sweating. In some examples, the parameter may be body composition of the user. For example, the body composition may include body fat and muscle of the user. For example, the body composition may be measured by a body composition sensor. The body composition sensor may be configured to measure a signal indicative of the body composition. For example, bioelectrical impedance analysis (BIA) may be used. BIA involves measuring the impedance of the body, which is related to the amount of water in the body, which in turn corresponds to an amount of muscle - thus providing an indication of body composition. In some examples, the parameter may be breathing of the user. For example, the breathing of the user may include the breathing rate or respiratory rate. For example, the breathing may be measured by a breathing sensor. The breathing sensor may be configured to measure a signal indicative of the breathing of the user. In one example, the breathing may be determined by measuring sound. The sound can also classify whether the user is snoring. Thus, the breathing sensor may be a snoring sensor. In other examples, the breathing may be determined by measuring movement, such as movement of the user’s chest. In some examples, the sensor may be an apnoea sensor. For example, the apnoea sensor may be used to monitor signs of obstructive sleep apnoea (OSA) during sleep. In some examples, the parameter may be maximum oxygen consumption of the user. For example, the sensor may be a VO2 sensor. Other physiological parameters may be used. For example, the parameter may include eye movements, such as using electrooculography (EOG). In another example, the parameter may include muscle activity or skeletal muscle activation, such as using electromyography (EMG). In some examples, a piezo-film sensor or a photo optic sensor may be used to infer heart rate and / or respiratory rate. In some examples, an analogue front end (AFE) sensor may be used for bioimpedance analysis, electrical, and / or optical biosensing, for example for ECG or other heart rate measurements. In some examples, a device comprises one or more sensors. For example, the device may be a smartwatch. The device may include multiple sensors. The multiple sensors may be configured to measure different physiological parameters. For example, a smartwatch may include a heart rate sensor (an optical sensor on the user’s wrist) and a body temperature sensor (a thermistor in contact with the user’s skin). Data from multiple sensors can be packaged and transmitted together to the control unit, or separate sensor data may be transmitted. In some examples, the parameter is indicative of whether the user is awake or asleep. In some examples, the parameter comprises a parameter associated with a sleep stage of the user. In other words, the parameter may depend on the sleep stage. The sleep stage may be as described herein. For example, the sleep stage may be awake, light sleep, deep sleep, or REM. For example, some parameters may indicate whether the user is awake or asleep, or may indicate a specific sleep stage. For example, a physiological parameter may be used. In some examples, the physiological parameter comprises: a heart rate of the user, a blood pressure of the user, a blood volume of the user, brain activity of the user, a body temperature of the user, and / or breathing of the user. For example, each of these parameters may indicate whether the user is awake or asleep, or may indicate a specific sleep stage. Optionally, the parameter associated with a sleep state of the user comprises a physical parameter associated with a sleep state of the user. As used herein, preferably the term “physical parameter” refers to a parameter associated with an external characteristic of the user’s body. Optionally, the physical parameter comprises: a position of the user and / or movement of the user. It is particularly preferable to use the movement of the user. In some examples, the parameter may be position of the user. For example, the position of the user may be measured by a position sensor. The position sensor may be configured to measure a signal indicative of the position of the user. In some examples, the position may be determined based on an absolute position. In other examples, the position may be determined based on a relative position, such as a relative distance from the sensor. For example, a proximity sensor may be used to determine the location of the user relative to the proximity sensor. A signal strength may be used to determine a distance. In other examples, GPS may be used to determine the position of the user. In other examples, a gravity sensor (e.g., a magnetometer) may be used to determine a position of the user. The sensor may also determine whether the user is sitting up or lying down. For example, the gravity sensor may determine whether the user is lying down. For example, the sensor may be arranged on a particular body part, such as the head, and may be used to determine that the user has their head on the pillow. In some examples, the position sensor may be a gyroscope. In some examples, the parameter may be movement of the user. For example, the movement of the user may be measured by a movement sensor. The movement sensor may be configured to measure a signal indicative of the movement of the user. The movement may be determined by changes in the position of the user. Thus, the position sensor may be used to determine changes in the position, and therefore to determine movement. For example, a position sensor (such as a proximity sensor, GPS, or a gravity sensor) may be used to determine movement based on changes in position. In some examples, a sensor configured to measure movement may be used. For example, an accelerometer may be used to determine movement. In other examples, movement may be measured by using a rotation vector sensor. For example, a gyroscope may be used to determine movement. In some examples, a magnetometer may be used to determine movement. In some examples, gesture recognition or wrist tilt sensors may be used. In some examples, a step detector and counter may be used to indicate movement. In other examples, an inactive time sensor may be used to determine inactivity and therefore lack of movement. In some examples, movement of the user may be measured using a microwave or PIR (passive infrared) sensor. In other examples, movement of the user may be indirectly determined by measuring another signal, such as sound. For example, the sound associated with the user (for example, sound of movement) may be measured by a microphone. The microphone may be configured to measure a signal indicative of the sound associated with the user. In some examples, the sound may correspond to rustling of the bedsheets, which in turn corresponds to movement of the user. In some examples, a device comprises one or more sensors. For example, the device may be a smartwatch. The device may include multiple sensors. The multiple sensors may be configured to measure different physical parameters. The device may also include multiple sensors, such as one or more sensors configured to measure a physiological parameter and one or more sensors configured to measure a physical parameter. For example, a smartwatch may include a heart rate sensor (an optical sensor on the user’s wrist) and a movement sensor (such as an accelerometer). In another example, a smartwatch may include a heart rate sensor, a body temperature sensor, and a movement sensor. In some examples, the system comprises an off body detector. The off body detector may be referred to as a low latency off body detector. The off body detector may be configured to determine whether the device (such as the sensor) is being worn. For example, if the sensor (e.g., heart rate sensor) is contained within a smartwatch, the off body detector may determine whether the smartwatch is currently being worn or not. In some examples, the system comprises a time synchronisation unit. The time synchronisation unit may be referred to as a time sync sensor. The time synchronisation unit may be configured to determine the correct time and day (for example, considering daylight saving time) to synchronise all sensors. In some examples, time synchronisation may be performed by the processor of the control unit. In other examples, the time synchronisation may be performed by the sensor, or the device comprising the sensor, such as a smartwatch. In some examples, the system comprises a statistics unit. The statistics unit may be referred to as a statistics sensor. The statistics unit may be configured to gather statistics from different sensors about different parameters (such as physiological and / or physical parameters). In some examples, the parameter comprises: a heart rate of the user, a blood pressure of the user, a blood volume of the user, a blood oxygen saturation of the user, brain activity of the user, a body temperature of the user, sweating of the user, body composition of the user, breathing of the user, position of the user, and / or movement of the user. As explained above, the processor determines a sleep state based on the sensor data. In some examples, the sleep state comprises an asleep state or an awake state. In other words, the sleep state may refer to whether the user is asleep or awake. Said another way, the sensor data may be used to determine whether the user is asleep or awake. For example, if the sensor data relates to a heart rate, the processor may determine that the heart rate is below a threshold, and therefore the determined sleep state may be asleep. As another example, if the sensor data relates to a sound signal which indicates the user is snoring, the determined sleep state may be asleep. In some examples, the processor may be configured to determine a characteristic of the sensor data in determining the sleep state. In other words, determining the sleep state may comprise determining a characteristic of the sensor data. The characteristic may be determined in the signal. In some examples, the processor may be configured to compare the sensor data to a threshold in determining the sleep state. Thus, the characteristic may be whether the sensor data is above or below the threshold. For example, if the sensor data is below a threshold, a first sleep state may be determined, whereas if the sensor data is above a threshold, a second sleep state may be determined. As one example, if the heart rate is below a threshold, the sleep state may be determined as asleep, and if the heart rate is above a threshold, the sleep state may be determined as awake. Thus, in some examples, a value of the sensor data may be compared to a threshold value. In other examples, a trend in the sensor data may be used. For example, the processor may determine whether the sensor data is increasing or decreasing over time, and may determine a trend (e.g., rising or falling). Determining the characteristic may therefore involve determining the trend. The trend may then be used in determining the sleep state. In other examples, the processor may calculate a rate of change in the sensor data over time. This rate of change can then be compared to a rate of change threshold. In some examples, the sensor data values may be used in determining the sleep state. For example, if the sensor signal is above a threshold, a particular sleep state may be determined. Therefore, the threshold may be a value or a rate of change. In some examples, the sensor data may be used to determine the parameter associated with the sleep state. For example, a body temperature signal may be converted into an indication of the body temperature. For example, rather than determining whether the raw signal from the sensor (e.g., a voltage) is above a threshold, the voltage may be converted into a body temperature (e.g., in °C), and the value of the parameter may be compared to a threshold. The parameter can then be used to determine the sleep state. Thus, the processor may be configured to determine a characteristic of the parameter in determining the sleep state, such as by comparison to a threshold or trend. In some examples, different features of sensor data may be extracted. In some examples, the feature may be in the time domain. For example, the feature may be average, standard deviation, maximum, minimum, or a count of events. In some examples, the feature may be in the frequency domain. For example, the feature may be magnitude of frequencies in a specific band associated with a particular parameter. The algorithms can then classify the data into a sleep state based on the features. In some examples, sleep events may be determined based on the sensor data. In some examples, the sleep state may correspond to a sleep event. The sleep events may be desirable or undesirable sleep events. Desirable sleep events may improve quality of sleep. For example, these may correspond to normal transitions of sleep stages based on healthy sleep. Undesirable sleep events may worsen quality of sleep. For example, these may correspond to awakening episodes during the sleep session. The olfactory output may be selected to aim to reduce the number of undesirable sleep events, such as awakening episodes. For example, this may be achieved by reducing the duration of light sleep stages, or increasing the length of deep sleep stages (such as by anticipating or delaying transition between the stages). In some examples, the processor may further determine a quality of sleep. In some examples, the sleep state may comprise a quality of sleep. For example, the parameter may indicate a quality of sleep. For example, the sensor data may indicate that the sleep quality is poor. In some examples, the sleep quality may be good quality sleep or bad quality sleep. Other categorisations may be possible in other examples. For example, the sleep state may indicate a sleep stage such as deep sleep, and further indicate the sleep quality (e.g., good or bad) forthat sleep stage or the overall sleep session. In one example, regular movement of the user over a period of time may indicate that the user is restless. This may either indicate that the user was not asleep for a period, or even if the user was asleep, that the sleep quality is poor. In another example, high levels of sweating may indicate that the user is not sleeping well. The user may also self-report sleep quality via a user input as described herein. The processor may further determine a duration of sleep, for example for each sleep state. The processor may determine an overall duration of sleep, such as a cumulative time the user was asleep or in particular sleep states. This information can be presented to the user, for example informing them that they slept for a total of 8 hours, with a total of 4 hours of deep sleep. The processor may also determine a number of awake events. This may correspond to when the processor determines that the sleep state is awake. This information can be presented to the user, for example informing them that they awoke 4 times during the night. In some examples, the sleep quality may be provided to the user. For example, the system may output a determined sleep quality assessment, such as at different points during sleep. For example, the system may comprise a display, or may transmit data to the user (such as a smartphone or computer) for display. The system may also provide additional context that may affect sleep quality, for example the olfactory output settings at that time. The user can then make informed choices, such as setting preferences for repeating or not repeating that olfactory output under those conditions based on the determined sleep quality following use of that olfactory output. Similarly, the system may also provide environmental data, such as from an environmental sensor, that may affect sleep quality, for example the environmental conditions at that time. For example, this may indicate that the room temperature was high when the system determined poor sleep quality, and the user may make informed choices, such as to control the temperature of the room to improve sleep quality. Alternatively, the system may also provide a recommendation based on the sleep quality, for example recommending control of the temperature based on analysis of the environmental sensor data. As another example, the system may recommend repeating olfactory output settings based on analysis of the impact of those settings on sleep quality. In some examples, the sleep state comprises an awake state or an asleep state. In some examples, the sleep state comprises a sleep quality. For example, the sleep quality may comprise good quality sleep or poor quality sleep, or in other examples may comprise different categories. Optionally, the sleep state comprises a sleep stage. In some examples, the sleep stage comprises a non-rapid eye movement (NREM) stage or a rapid eye movement (REM) stage. For example, the sleep stage may comprise one or more of: deep sleep, light sleep, and REM sleep. For example, the sleep stage may comprise one of: non-rapid eye movement stage one, non-rapid eye movement stage two, non-rapid eye movement stage three, and rapid eye movement. These correspond to the established four sleep stages. Being awake may also be included in this. For example, the sleep stage may comprise one or more of: deep sleep, light sleep, REM sleep, and awake. For example, the sleep stage may comprise one of: awake, non-rapid eye movement stage one, non-rapid eye movement stage two, non-rapid eye movement stage three, and rapid eye movement. In one preferred example, the processor determines the sleep state between awake, REM sleep, light sleep, and deep sleep. Therefore, the sleep state may change between these stages during the sleep session. A sleep cycle is generally defined as three non-rapid eye movement stages followed by a rapid eye movement stage. In a typical night sleep of a healthy individual, a user might go through four to six sleep cycles. The processor may therefore be configured to determine a sleep stage based at least on the sensor data. In other words, the processor may be configured to determine that the user is in a particular sleep stage. For example, based on the sensor data indicating a parameter falls within a particular range, the processor may determine the user is in a particular sleep stage. As one example, if the heart rate falls from above a threshold to below the threshold, the processor may determine that the user has fallen asleep and entered NREM stage 1. Each sleep stage is characterised by a specific pattern in different parameters. For example, brain activity, heart rate, respiratory rate, eye movements, and muscle movements will change depending on the sleep stage. For example, the heart rate is often higher during awake time, and decreases during light sleep, and is at a minimum during deep sleep. During REM sleep, the heart rate increases again and shows higher variability. Thus, changes in the heart rate can be used to determine changes in sleep state. As another example, blood pressure is often higher while the user is awake and is at a minimum during sleep. Blood pressure usually starts rising a few hours before the user wakes up. Thus, changes in the blood pressure can be used to determine changes in sleep state. As another example, blood volume increases during sleep, in particular during REM sleep. Thus, changes in the blood volume can be used to determine changes in sleep state. As another example, movement and / or position sensors can be used to determine activity levels, such as whether the user is lying in bed, and to determine different stages of sleep. For example, during initial stages of sleep, the body is not completely relaxed, and movement (particular limb and trunk movements) is still common. With transition to deeper sleep stages, the body relaxes, and movement (as well as variability in movement) decreases. During REM sleep, the body is temporarily almost paralyzed (atonia, lack of muscle tone), and this can be determined based on data from the movement sensors. As yet another example, body temperature changes during different sleep stages, being lower in deeper sleep stages. As another example, breathing rate decreases and becomes less variable during sleep, but it can increase and show more variability during REM sleep. In addition, an abnormally lower respiratory rate can be a sign of sleep apnoea. Therefore, by analysing the parameters, the sleep state can be determined, along with changes to sleep state. Specifically, some parameters can be used to determine a sleep stage. Algorithms may be used to determine the sleep state based at least on the sensor data. For example, the information from the at least one sensor can be fed into a pre-trained machine learning algorithm. The algorithm can compute a probability for a specific sleep state (such as a sleep stage) based on the sensor data. For example, for each time period (such as 30 second epoch), the algorithm can compute a probability based on the sensor data from that time period. The processor can therefore use these algorithms to determine the sleep state. In other examples, classical algorithms may be used. In some examples, parameters can be used to indicate a sleep quality. In some examples, parameters can be used to indicate a sleep event. For example, the sleep event may be an undesirable sleep event. For example, the sleep event may be a sleep disturbance. For example, the sleep event may be sleep apnoea or sleep terrors. For example, blood oxygen saturation may not change significantly during normal sleep, and a drop in blood oxygen saturation may indicate sleep apnoea. Similarly, skin conductance may not change significantly during normal sleep, and an increase in skin conductance can indicate arousal associated with night terrors or nightmares, for example. High skin conductance can also be associated with sleep deprivation. Body composition / BIA may not change significantly during normal sleep, but sleep can be negatively affected by body composition. For example, overweight users may experience more incidence of sleep apnoea. In some examples, the input interface is configured to receive sensor data from a plurality of sensors each configured to measure a signal indicative of a parameter associated with a sleep state of the user. For example, multiple different sensors may be used. The parameters may be different between sensors. The signals measured may also be different. For example, a first parameter may be a physiological parameter and a second parameter may be a physical parameter. As one example, a first parameter may heart rate and a second parameter may be movement. In some examples, different signals may be measured for the same parameter. For example, different movement sensors may be used to infer different properties associated with movement of the user (such as an accelerometer and a sound sensor). In some examples, the processor is configured to determine the sleep state of the user based on the sensor data from the plurality of sensors. This can be used to more accurately determine the sleep state, especially when used to determine a sleep stage. For example, data from multiple sensors can be used to differentiate between situations where sensor data from one sensor could be indicative of multiple potential sleep states. Optionally, the processor is configured to determine the settings for delivery of the olfactory output to facilitate maintaining the sleep state or to facilitate transitioning the sleep state. In particular, the settings may be selected for maintaining the sleep state. In other words, the olfactory output is intended to keep the sleep state the same, for example to continue a period of deep sleep. For example, if a particular smell can be used to maintain a sleep state, such as a particular sleep stage, then the settings may correspond to selection of an olfactory output to generate that smell. The settings may also be selected for transitioning the sleep state. In other words, the olfactory output is intended to change the sleep state to a different sleep state, for example from a first sleep stage to a second sleep stage, or from an asleep state to an awake state. Optionally, the processor is further configured to determine a predicted subsequent sleep state, and wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the predicted subsequent sleep state. In some examples, the processor may be configured to determine the settings for delivery of the olfactory output based at least one the sleep state and the predicted subsequent sleep state. In some examples, the predicted subsequent sleep state is determined based on the current sleep state. For example, once the processor has determined the current sleep state, the processor may then determine the predicted subsequent sleep state. In one example, the current sleep state may be asleep, and the subsequent sleep state may be awake. In another example, the current sleep state may be NREM stage 3, and the predicted subsequent sleep state may be REM. The predicted subsequent sleep state may be predetermined based on stored subsequent sleep states for a given current sleep state. For example, the subsequent sleep state may be the next subsequent stage of sleep in the sleep cycle. The subsequent sleep state may therefore be retrieved based on the current sleep state. Alternatively, the subsequent sleep state may be specific to that user. For example, historical data may be used to determine the most common subsequent sleep state that usually follows the current sleep state. The predicted subsequent sleep state may also be determined dynamically based on the current sensor data. For example, a slowing heart rate may indicate that the next sleep state is likely to be deep sleep. The settings may be determined based on the predicted subsequent sleep state. For example, if it is determined that the subsequent sleep state is deep sleep, particular settings may be applied that are desired for that subsequent sleep state. In addition, particular settings may be used to affect the timing of the transition to the subsequent sleep state. For example, the settings may be selected to trigger or delay the transition to the subsequent sleep state. Optionally, the processor is further configured to determine a progress of the sleep state, and wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the progress of the sleep state. In some examples, the processor may be configured to determine the settings for delivery of the olfactory output based at least on the sleep state and the progress of the sleep state. In some examples, the processor may be configured to determine the settings for delivery of the olfactory output based at least on the sleep state, the progress of the sleep state, and the predicted subsequent sleep state. In some examples, the progress of the sleep state comprises: a time elapsed since a start of the sleep state, a predicted time remaining of the sleep state, a predicted time of a start of a subsequent sleep state, or any measure of how far the user is through the sleep state. The progress of the sleep state may refer to how far through the sleep state the user is. For example, if a particular sleep state has a typical duration of 60 minutes (either predetermined, or based on historical user data), the progress of the sleep state may indicate that the user is 30 minutes into the sleep state. For example, the processor may determine this based on a detected transition into the current sleep state (i e., an indication of the start of the sleep state). The control unit may store an expected duration of each sleep state, and based on the current time, the processor can determine the progress of the sleep state. In some examples, the progress of the sleep state corresponds to a time elapsed since the start of the sleep state. In some examples, the progress of the sleep state corresponds to a predicted time remaining of the sleep state. In some examples, the progress of the sleep state corresponds to a predicted time of the start of the subsequent sleep state. In some examples, the progress of the sleep state corresponds to a measure of how far the user is through the sleep state. For example, the progress of the sleep state may correspond to a proportion of the duration of the sleep state elapsed (or remaining), such as a percentage completion of the sleep state. The settings may be determined based on the progress of the sleep state. For example, if it is determined that the user is reaching the end of the sleep state, particular settings may be applied to trigger or delay the transition to the subsequent sleep state. The combination of the progress of the sleep state and the predicted subsequent sleep state may be used in determining the settings. For example, the progress may indicate the approaching end of the current sleep state, and based on the subsequent sleep state, the processor may determine particular settings to apply that will be desired for the subsequent sleep state, and to trigger or delay the transition to that sleep state. In some examples, the settings for delivery of the olfactory output comprise predetermined settings. In other words, the settings may be fixed, with the particular setting depending on the sleep state. In other words, when a particular sleep state is determined, predetermined settings may be retrieved and applied. For example, the predetermined settings may be based on known links between particular olfactory outputs and sleep states. In other words, when a change between sleep states is determined, the same olfactory output may be delivered for that sleep state each time and for each user. This allows for a simplified system, which can adjust delivery of the olfactory output based on the sleep state, whilst minimising data collection and processing requirements. In some examples, the settings for delivery of the olfactory output are customised to the user. In other words, rather than having predetermined settings that are the same for every user, the settings may be specific for that user. For example, the user may prefer a particular smell to wake up. The user may be presented with different options for different scenarios, such as selecting a particular smell for a particular sleep state or time of night. The user can then select various settings to customise the settings for their preferences. The user selection may involve the user selecting settings that have a perceived improvement in sleep quality. For example, if the user determines that a particular olfactory output resulted in good quality sleep, they may select that in the future. In some examples, the system may initially operate with default settings including the predetermined settings, until overridden by user preferences. This allows the smell delivery to be tailored to the user, further improving sleep quality. Optionally, the input interface is further configured to receive subsequent sensor data from the at least one sensor, and wherein the processor is further configured to determine an impact of the settings for delivery of the olfactory output on the sleep state based at least on the subsequent sensor data. The subsequent sensor data preferably refers to sensor data that is received at a later period in time than the sensor data. More specifically, the subsequent sensor data preferably refers to sensor data that is received after the smell delivery device delivers the olfactory output based on the instructions. The subsequent sensor data may be from the same sensor or a different sensor as the earlier sensor data. In other words, if the instructions to the smell delivery device resulted in delivering a particular olfactory output, the subsequent sensor data may indicate a response of the user to that olfactory output. For example, this may indicate a response of the sleep state to the olfactory output. This may be used to determine an impact of the settings on the sleep state. For example, if the subsequent sensor data indicates a change in sleep state, it may be determined that the olfactory output had the impact of changing the sleep state. This may be used to determine whether the settings had the desired effect or not. Optionally, the processor is further configured to determine subsequent settings for delivery of the olfactory output based at least on the impact of the settings on the sleep state. For example, if it is determined that the settings had a particular impact on the sleep state, such as causing a desired transition of a sleep state, those settings may be applied subsequently to achieve the desired effect (for example, when it is next desired to trigger that same transition of sleep state). Therefore, this feedback mechanism can be used to improve the effectiveness of delivery of the olfactory output. Optionally, the processor is further configured to determine the settings for delivery of the olfactory output based at least on an impact of previous settings on the sleep state. The impact may be determined by the processor, as described below, or it may be received by the control unit, such as in historical sleep data. This allows the system to improve the effectiveness of the current settings. Optionally, the processor is further configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state, optionally wherein the processor is configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state based on historical sleep data. The processor may then be configured to determine the settings for delivery of the olfactory output based at least on the impact of the previous settings on the sleep state. In other words, based on the effectiveness of the previous settings, those same settings can either be used or not. This allows the effectiveness of the current settings to be improved. The impact of the previous settings may be determined based on historical sleep data. For example, the control unit may be configured to receive historical sleep data. In some examples, the control unit is configured to store sleep data over time. For example, the sleep data may be stored in a data storage unit, such as a memory. The sleep data may be stored locally (e.g., in the control unit), or it may be transmitted to a remote storage. The sleep data may then be retrieved by the control unit. In some examples, the historical sleep data may be received at the input interface. The historical sleep data allows for later analysis of the sleep state at different times throughout the sleep session. For example, the sleep data can indicate how the sleep state changed overnight, for example indicating when the user changed between sleep stages. The sleep data can also be collected over multiple nights and used to build a pattern of data for the user. The historical sleep data can indicate trends in sleep, such as periods of poor-quality sleep. This can also indicate common patterns such as average times for each sleep state for the user. Deviations from normal sleep can therefore be identified. This can also be used to predict future changes in sleep states. This historical sleep data can be output to the user. For example, the historical sleep data may be output on a display of the smell delivery system, or the data may be transmitted to a user device. For example, the data may be transmitted directly from the control unit or from the remote storage. The data may be transmitted live, on demand, or the data may be stored for later transmission such as after the sleep session. The historical sleep data preferably at least comprises the settings. In other words, the historical sleep data may indicate the settings overtime. For example, the historical sleep data can indicate the settings for olfactory output for a given sleep state or at a particular time. This can be used to determine how the sleep changed in response to the settings. The user can analyse this when receiving the data. Alternatively, the processor may analyse the data to determine the effectiveness of different settings. In some examples, the historical sleep data comprises the sleep state. In other words, the historical sleep data may comprise the settings and the sleep state. The historical sleep data may comprise the corresponding time associated with that sleep state. Thus, the historical sleep data may indicate how the sleep state changed over time, and also includes the corresponding settings. This can be used to determine how the settings affected sleep. In some examples, the processor is configured to determine the current sleep state of the user based at least on the historical sleep data. In some examples, the processor is configured to determine the sleep state of the user based at least on the sensor data and the historical sleep data. For example, if the historical sleep data indicates that a user typically experiences a particular sleep stage for an average time of 60 minutes, and the sensor data suggests that the current sleep state has lasted for approximately 60 minutes and then changes, this may indicate that the current sleep state corresponds to the particular sleep stage suggested by the historical sleep data. This can improve the accuracy of the determination. In addition, the impact of the settings on the sleep state can be determined from the settings in combination with the sleep state. For example, the historical sleep data can indicate how the sleep state changed in response to the settings. The processor can therefore determine the impact based on the historical sleep data. Alternatively, the impact may be determined and stored as part of the historical sleep data. For example, the historical sleep data may comprise the impact of the settings on the sleep state. This impact may be received at the processor, and used in the analysis, meaning that the impact does not need to be determined each time. In some examples, the historical sleep data comprises the sensor data. In other words, the historical sleep data may indicate the sensor data over time. In some examples, the historical sleep data may comprise the settings and the sensor data. In some examples, the sleep state is not required as part of the historical sleep data, since the sleep state can be inferred or determined based on the sensor data. For example, an impact of the settings may be determined based on changes in the sensor data. In response to a change in settings, any corresponding change in sensor data can indicate how those settings affected the sleep state. In particular, if subsequent sensor data indicates a change in sleep state, it may be determined that the olfactory output had the impact of changing the sleep state. In some examples, the historical sleep data comprises the settings, the sensor data, and the sleep state. In some examples, the historical sleep data may be used in determining the sleep state. For example, by associating sleep state with sensor data, it can be determined that a particular characteristic in the sensor data (such as a change in value of a particular parameter) is a good indicator of a change in particular sleep states. Therefore, when a particular characteristic in the sensor data occurs (such as a parameter passing a particular threshold value), then based on the historical sleep data, this may indicate a change in sleep state. In this manner, historical sleep data specific to the user can be used for determining the sleep state, instead of simply having a predetermined set of criteria (e.g., fixed thresholds) for determining the sleep state based on the sensor data. In some examples, different criteria / characteristics for determining sleep state, such as thresholds, may be adjusted based on the historical sleep data. As described above, historical sleep data may be used to indicate how the user responded to previous settings. The processor may determine whether the previous settings were effective, and whether the same settings should be applied now. For example, if particular settings to deliver a particular olfactory output resulted in improved sleep quality or achieved a desired transition of sleep state, those settings may be used again when the same situation arises. The settings may therefore be determined based on the impact which is determined based on the historical sleep data. The impact may also be determined in other ways. The control unit may store historical sleep data overtime, such as in real-time or in batches, such as after a sleep session. The control unit may store the settings, the sleep state, and / or the sensor data over time. By storing this data as historical sleep data, this can then be used to determine impacts of the settings on the sleep state in later analysis, such as for determining future settings. In some examples, having determined the impact, this impact may also be stored as part of the historical sleep data. Optionally, the input interface is further configured to receive a user input from the user indicating user feedback data associated with sleep, optionally wherein the processor is further configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state based at least on the user feedback data. In other examples, the user input is received through a different interface. In particular, the user feedback data may comprise an assessment of sleep quality. For example, the user feedback data may comprise an indication of sleep quality. This may be used to determine an effectiveness of the settings for delivery of the olfactory output. In other words, this may be used to determine that the settings were effective or not effective. For example, the user may indicate how well they slept, or how refreshed they felt when they woke up. The user feedback may also indicate if and when they woke up during the night. This information, in combination with the sensor data, can be used to determine changes in sleep state over the sleep session, and the effectiveness of the settings. The user feedback data may also be stored as part of the historical sleep data. The user feedback data may also comprise feedback on different smells and preferences. The impact of the previous settings may be determined based at least on the user feedback data. In some examples, the processor is configured to determine the settings for delivery of the olfactory output based at least on the impact of the previous settings on the sleep state. For example, the user feedback data may indicate that the user experienced good quality sleep, which may indicate that the settings were effective. This may be instead of, or in addition to, determining the impact based on the historical sleep data. In some examples, machine learning or Al can be used to implement these determining steps. For example, determining the settings for delivery of the olfactory output may comprise using machine learning algorithms. For example, classic or machine learning algorithms may be used to determine the settings. In some examples, this may be based on the historical sleep data. For example, by considering the impact of particular settings on sleep state, machine learning algorithms can determine which olfactory outputs have the desired effects for that particular user. Optionally, the input interface is further configured to receive environmental sensor data from at least one environmental sensor configured to measure a signal indicative of an environmental parameter in an environment of the user. Environmental conditions can affect how a smell is perceived by a user. For example, changes in air pressure (such as before a thunderstorm) result in a perceivable difference in experiencing smells. Other environmental parameters may also change a user’s perception of a smell. Therefore, depending on environmental parameters, a set intensity of smell delivery may result in different user perceptions. By measuring these environmental parameters, a difference in user perception can be determined and this can be compensated for. In some examples, the environmental parameter comprises one or more of: temperature, humidity, pressure, altitude, background smell, and / or light. Optionally, the processor is further configured to determine an impact of the environmental parameter on the delivery of the olfactory output, based at least on the environmental sensor data. Since different environmental parameters affect a user perception of the smell, based on the environmental sensor data, the processor can determine how these environmental conditions will affect delivery of the olfactory output For example, the environmental conditions may increase or decrease the ability of the user to detect a particular smell or at a particular intensity. In some examples, the environmental sensor may be a temperature sensor. The temperature sensor may be configured to measure the temperature of the room. The temperature of the room may be the environmental parameter. Higher temperatures can increase the volatility of odour molecules, increasing dispersion, and leading to increased intensity. For example, in response to detecting a high temperature, the processor may determine an impact on the delivery of the olfactory output of decreased intensity. In some examples, the environmental sensor may be a humidity sensor. The humidity sensor may be configured to measure the humidity of the room. The humidity of the room may be the environmental parameter. Humidity can affect the dispersion rate of chemicals, thus affecting the speed of diffusion within the space. More water in the air causes the odour molecules to mix and spread less. Accordingly, higher humidity can lead to reduced spreading of the olfactory output, reducing the intensity. Perception of the smell is therefore decreased. In some examples, the environmental sensor may be a pressure sensor. The pressure sensor may be configured to measure the pressure of the room. The pressure of the room may be the environmental parameter. Low air pressure, such as during stormy weather or in a plane cabin, can result in decreased odour sensitivity. In some examples, the environmental sensor may be an altitude sensor. The altitude sensor may be configured to measure the altitude. The altitude may be the environmental parameter. Altitude can affect the user’s perception of smell. Higher altitudes can decrease olfactory function, decreasing odour sensitivity. In some examples, the environmental sensor may be a sensor configured to measure a background smell. In particular, this sensor may measure contamination or pollutants in the environment. For example, pollutants such as a forest fire within the vicinity may result in a burning smell in the environment, which may overpower the smells delivered by the smell delivery device. The background smell, resulting from other external factors, as well as previous smells from the smell delivery device, may change how different olfactory outputs are perceived by the user. For example, the environmental sensor may be a gas sensor for determining air quality and / or chemicals in the environment. For example, the gas sensor may be a photoionization sensor. In another example, the gas sensor may be a metal oxide gas sensor. For example, there may be a single sensor, or an array of gas sensors, each for detecting a different chemical. In some examples, the environmental sensor may be a light sensor. The light sensor may be configured to measure light levels in the room. The light may be the environmental parameter. Light may affect the user’s sleep, such as sunlight in the morning causing the user to wake up. Turning off the bedside light may also be an indication of when the user begins attempting to sleep. Optionally, the processor is configured to determine the settings for delivery of the olfactory output based at least on the impact of the environmental parameter on the delivery of the olfactory output. In this manner, the processor can compensate for the impact of the environmental conditions on the delivery of the olfactory output. For example, in response to determining a particular humidity level, the processor may determine the settings to ensure that the user perception of the smell will be at the desired level. For example, if the humidity will result in lowering the user’s perception of the smell, the settings may be adjusted to increase the intensity of the olfactory output, to ensure that the smell has the intended effect on the user. The combination of at least one sensor (such as a physiological and / or a physical sensor) and an environmental sensor may be particularly beneficial to provide accurate delivery of the olfactory output. The settings for delivery of the olfactory output can be based on the sensor data, and this can be adjusted as necessary to compensate for the environmental sensor data (for example, avoiding a reduction in user perception due to environmental conditions). In some examples, the processor is further configured to determine an impact of the environmental sensor data on the sleep state based at least on the environmental sensor data. In other words, in some cases, the environmental parameter may affect the sleep state. For example, in extreme temperature, humidity, pressure, or altitude, this may affect the user’s ability to sleep, either keeping them awake, waking them up, or interrupting sleep cycles. In other examples, light may also affect the sleep state, for example sunlight in the morning may wake up the user. This information may be used in combination with the sensor data to confirm a particular sleep state. In some examples, the processor is configured to determine the settings for delivery of the olfactory output based at least on the impact of the environmental parameter on the sleep state. Based on the impact of the environmental parameter on the sleep state, the settings may be adjusted to compensate for this. Alternatively, the environmental parameters may be used to encourage changes in sleep state. For example, in response to detecting increasing light levels indicative of sunlight in the morning, the processor may determine that it is time for the user to wake up, and the processor may then determine settings for delivery of a particular olfactory output for that purpose. Optionally, the settings for delivery of the olfactory output correspond to one or more of: initiating delivery of an olfactory output, ceasing delivery of an olfactory output, selecting a particular olfactory output or combination of olfactory outputs for delivery, selecting an intensity of an olfactory output for delivery, and / or selecting a timing of an olfactory output for delivery. For example, the settings may correspond to starting release of the smell of lavender. This may be done when the user enters a deep sleep state. In some examples, the timing of an olfactory output may include the time at which delivery of the olfactory output occurs. In some examples, the timing of an olfactory output includes the duration of delivery of the olfactory output. In some examples, an olfactory output is delivered in pulses, and the timing of an olfactory output may include the duration of the total time window over which the pulses are delivered. In some examples, the timing may include the duration of each pulse. In some examples, the timing may include the time between pulses. In some examples, the timing may include the total on-time of all the pulses within the total time window. In some examples, the timing may refer to the number of pulses. In some examples, the timing may refer to the frequency of delivery of the olfactory output. In some examples, the settings for delivery of the olfactory output comprise an adjustment to current delivery of the olfactory output. For example, this may involve selection of a different olfactory output to the current olfactory output. In other examples, this may involve changing the intensity of the current olfactory output. In some examples, the delivery may include delivering a “relaxing” scent. For example, relaxing scents can be used to control the duration of some sleep stages that are known to be important for a restful and healthy sleep. Some such known relaxing scents include lavender, chamomile, jasmine, rose, and sweet marjoram. In other words, several predetermined scents may be used as relaxing scents. Thus, in order to maintain a sleep state, the settings for delivery of the olfactory output may comprise selection of a relaxing scent. In some examples, the delivery may include delivering an “activating” scent. For example, activating scents may be an arousing effect to trigger changes in sleep state, such as waking up. Some such known activating scents include coffee, eucalyptus, and peppermint. In other words, several predetermined scents may be used as activating scents. Thus, in order to change a sleep state, the settings for delivery of the olfactory output may comprise selection of an activating scent. In other examples, the scent may be personalised to the user. For example, specific users may respond differently to different scents. Based on user input and historical data, different scents can be classified as relaxing or activating. As one example, in order to increase deep sleep stage duration, a relaxing scent may be delivered as soon as the user enters the deep sleep stage. Alternatively, a relaxing scent can be delivered just before a predicted deep sleep state, for example during a light sleep stage. This can aim to accelerate the transition from light sleep to deep sleep. Optionally, the smell delivery device comprises a flow generator configured to provide a flow of carrier gas to the odour chamber for releasing the substance, and wherein the settings for delivery of the olfactory output correspond to adjusting a flow rate of the flow generator. In some examples, the carrier gas may be odourless. For example, the carrier gas may be air. The flow generator may be a pump or a fan. For example, the flow generator may draw in a supply of air through an inlet from the external environment and provide this flow to the odour chamber. This flow of carrier gas can then be used to pick up the substance for delivery through the outlet. The settings may correspond to adjusting a flow rate of the flow generator, for example by turning the flow generator on or off, or by adjusting the speed. This can result in delivering different intensities of the olfactory output. In some examples, the processor is configured to determine the settings for delivery of the olfactory output based at least on a scheduled time associated with the sleep state of the user. For example, if the user has been in a sleep state for 60 minutes, the settings may be selected to trigger a change in sleep state. Predetermined average sleep state durations, or historical sleep data specific to that user, may be used in determining how long a sleep state should typically last. This can be used to encourage changes of sleep state at desired times. For example, the user may set a scheduled time for waking up. The system may determine that the user desires to wake up in 30 minutes time. Therefore, the settings may be determined to trigger a switch to a light sleep, so that the user does not wake directly from a deep sleep, and in doing so can wake more refreshed. Therefore, not only can overall sleep quality be improved, but the sleep state immediately before waking can be controlled to improve the user’s wake-up cycle. The system may therefore determine the settings to control the sleep state before the user wakes up. Therefore, the processor may be configured to determine the settings based at least on a scheduled time for waking up. In some examples, the scheduled time comprises an alarm time for the user to wake up. For example, the system may receive a scheduled alarm time, such as from the user’s alarm clock or smartphone. This can then be used as the desired scheduled awake time. In some examples, the system can control the alarm. For example, the user may select a desired alarm time such as 7am. The system may determine that the user is coming out of a deep sleep stage, and it would be undesirable to wake up at exactly 7am. Therefore, the system may determine that it would be better if the user awoke at 7.15am. Accordingly, the scheduled alarm time may be adjusted to 7.15am. The user may be able to select an initial alarm time of 7am, and to select a desired range within which the alarm can be changed, such as within 15 minutes. In this manner, the system can ensure that the user awakes at a time to coincide with sleep states that may make the user less tired, whilst waking at a time desired by the user. In some examples, the system may output the adjusted alarm time to the user’s alarm such as an alarm clock or smartphone. In other examples, the system may directly control the alarm, such as by having an in-built alarm. Optionally, the control unit is integral with the smell delivery device. For example, the control unit may be part of the smell delivery device. The control unit and the smell delivery device may be formed as one single unit. For example, the control unit may be a processing unit of the smell delivery device. In examples with the control unit integral with the smell delivery device, providing the instructions to the smell delivery device may comprise internal communications. In some examples, the control unit may be configured to operate the smell delivery device to deliver the olfactory output. For example, the control unit may select different olfactory outputs by selecting particular odour chambers or substances (such as by controlling valves). The control unit may also control flow rate of the flow generator in order to control delivery and control intensity of the olfactory output. Optionally, the control unit is separate from the smell delivery device. For example, the control unit may not be part of the smell delivery device. The control unit and the smell delivery device may be formed as separate units. For example, the control unit may be a computing device, such as a smartphone, laptop or PC, or smartwatch. In some examples, the control unit is a standalone computing device. The control unit may be local to the smell delivery device. For example, the control unit may be provided in the same room as the smell delivery device (i.e., the control unit may be in the room in which the user is sleeping). In some examples, the control unit may be configured to provide the instructions to the smell delivery device by external communications. For example, the instructions may be transmitted wirelessly or provided over a wired connection. For example, the control unit may provide the instructions to the smell delivery device over a short-range wireless transmission protocol, such as Bluetooth™. In particular, the control unit and the smell delivery device may be capable of sending and / or receiving information with each other via Bluetooth™. Alternatively, the control unit may be connected to the smell delivery device by a wired connection, such as a USB cable. In some examples, the control unit comprises a remote computing device. For example, the control unit may be a remote server or remote computer. For example, the control unit may be a cloud-based server. The control unit may therefore not be local to the smell delivery device. For example, the control unit may not be provided in the same room as the smell delivery device (i.e., the control unit may not be in the room in which the user is sleeping). In some examples, the control unit may provide the instructions to the smell delivery device over a long-range wireless transmission protocol such as Wi-Fi™. In particular, the control unit and the smell delivery device may be capable of sending and / or receiving information with each other via Wi-Fi™. For example, the smell delivery device may connect to the internet via Wi-Fi™, for example by a local router in the vicinity, such as within or near the room. Alternatively, the smell delivery device may connect to the internet such as by cellular, 3G, 4G, or 5G, or via a wired connection such as Ethernet. The smell delivery device may therefore communicate with the control unit over the internet, which may be located remotely, such as in the cloud. In some examples, the smell delivery device may communicate with the control unit indirectly. For example, the smell delivery device may communicate with a local computing device, such as the user’s smartphone. For example, this may be over a direct short-range wireless transmission protocol such as Bluetooth™. The local computing device can then connect to the internet, such as by Wi-Fi™, for communication with the control unit. In this manner, the smell delivery device can communication with the control unit via the local computing device. This avoids the need of the smell delivery device to have a direct connection to the internet. Optionally, the smell delivery system further comprises the at least one sensor, optionally wherein the sensor is integral with the smell delivery device and / or with the control unit. In other words, the at least one sensor may be part of the smell delivery system. Thus, the smell delivery system may include the sensor. In other examples, the sensor is external to the smell delivery system, and the smell delivery system may receive the sensor data from the external sensor. The sensor may be integral with the smell delivery device. For example, the sensor may be provided as part of the same assembly as the smell delivery device. In other words, the smell delivery device may comprise the sensor (or multiple sensors). In some examples, the sensor is integral with the control unit. For example, the sensor may be provided as part of the same assembly as the control unit. In other words, the control unit may comprise the sensor (or multiple sensors). For example, the control unit may be a local computing device which includes a sensor. For example, the control unit may be a smartphone which includes a sensor such as a microphone. In this manner, the sensor data may be received internally by the control unit, such as by internal connections. In some examples, the sensor is integral with the smell delivery device and the control unit. In other words, the smell delivery device, the control unit, and the sensor are all part of the same assembly. For example, the smell delivery device may include the control unit and the sensor. In this manner, the sensor data may be received internally by the control unit, such as by internal connections. In some examples, the sensor is separate from the smell delivery device and / or the control unit. For example, the sensor may be separate from the smell delivery device. This may allow the sensor to be positioned on the user, for example. Some sensors may be separate from the smell delivery device, allowing the sensors to be positioned on the user, for example, while other sensors may be separate or may be provided with the smell delivery device. For example, the sensor may be separate from the control unit. In this manner, the sensor data may be received by the control unit over an external connection, such as Bluetooth™ or Wi-Fi™. In some examples, the sensor is provided on a wearable electronic device. In other words, the sensor may be a wearable sensor. For example, the sensor may be provided in a smartwatch. For example, the smartwatch may act as the sensor and measure a pulse rate of the user. The smartwatch may also act as the control unit by processing the sensor data. For example, an application running on the smartwatch may use the sensor data to determine the sleep state and thus determine the settings for delivery of the olfactory output. In other examples, the smartwatch may forward the sensor data to another device, such as a smartphone or other device to act as the control unit. For example, an application running on the smartphone may receive the sensor data and determine the sleep state and thus determine the settings for delivery of the olfactory output. In other examples, the sensor data may be passed to a separate control unit, such as a local computing device, a remote computing device, or integral with the smell delivery device. For example, the smartwatch or smartphone may upload the sensor data to the cloud for processing by the cloud-based control unit. The instructions may then be sent from the control unit directly to the smell delivery device, or via the smartwatch / smartphone. In other examples, the wearable electronic device may be a wristband, a headband, or any other device attached to the user or clothing. According to a second aspect, there is provided a control unit for controlling a smell delivery device, comprising: an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user; a processor configured to: determine a sleep state of the user based at least on the sensor data; and determine settings for delivery of an olfactory output based at least on the sleep state of the user; and an output interface configured to provide instructions to a smell delivery device corresponding to the settings, wherein the instructions are configured to cause the smell delivery device to deliver the olfactory output based on the settings. In other words, the control unit may be provided in isolation of the smell delivery device. In other words, features of the control unit may be provided without the smell delivery device. Features of the control unit of the first aspect may be readily applied to the control unit of the second aspect. According to a third aspect, there is provided a method of delivering an olfactory output, comprising: receiving sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user; determining a sleep state of the user based at least on the sensor data; determining settings for delivery of an olfactory output based at least on the sleep state of the user; and providing instructions to a smell delivery device corresponding to the settings. In some examples, the method is performed by a processor. For example, the method may be performed by a control unit. For example, the control unit may be the control unit of the first or second aspects. The method may further comprise steps corresponding to features of the smell delivery system of the first aspect. For example, the sensor data may be from one or more sensors, such as described above. The sleep state may be determined based on the values or changes in the sensor data, such as exceeding a threshold. In some examples, the sleep state may be asleep or awake, or in other examples may be a sleep stage such as described herein. Different sensors (such as measuring physical or physiological parameters) can be used to determine sleep states and changes in sleep state. For example, algorithms such as ML algorithms as described herein may be used to determining the sleep state. For example, historical data may be used. Settings for delivery of the olfactory output can then be determined based on the sleep state. For example, settings known to encourage deep sleep can be selected if desired, or settings to encourage transition of sleep state may be selected. The settings may be customised to the user, as described herein. In some examples, the settings may be adjusted based on environmental sensor data. The instructions can then be provided to the smell delivery device. For example, the instructions may be transmitted to a smell delivery device. For example, the transmission may be wireless, such as by Bluetooth™ or Wi-Fi™. Optionally, the method further comprises delivering, by the smell delivery device, the olfactory output based on the instructions. For example, this may be performed by the smell delivery device of the first aspect. Thus, in some examples, the method comprises operating the smell delivery device. In other examples, the method does not comprise operating the smell delivery device. According to a fourth aspect of the present disclosure, there is provided a computer program, computer program product, or computer-readable medium comprising instructions which, when executed by a processor, cause the processor to perform the method of the third aspect. Thus, the computer program may be implemented at a computing device such as the control unit. For example, the computer program may be in the form of an application for running on a smartphone or smartwatch. According to a fifth aspect of the present disclosure, there is provided a computing device comprising the computer program, computer program product, or computer-readable medium of the fourth aspect. For example, the computing device may be a smartphone or smartwatch for running an application. According to a sixth aspect of the present disclosure, there is provided a smell delivery system, comprising: a smell delivery device for delivering an olfactory output; a control unit, comprising: an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user; a processor configured to: determine a sleep state of the user based at least on the sensor data; and determine settings for delivery of an olfactory output based at least on the sleep state of the user; and an output interface configured to provide instructions to the smell delivery device corresponding to the settings; wherein the smell delivery device is configured to deliver the olfactory output based on the instructions. Thus, the features of the smell delivery device are optional. The smell delivery system of the sixth aspect may be similar to the smell delivery system of the first aspect, except that the features of the smell delivery device need not be provided. Other features of the first aspect may be readily applied to the sixth aspect and vice versa. According to a sixth aspect of the present disclosure, there is provided a smell delivery system, comprising: a smell delivery device for delivering an olfactory output, the smell delivery device comprising: an odour chamber for containing a substance configured to produce an olfactory output; and an outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device; and a control unit, comprising: an input interface configured to receive an input associated with a sleep state of a user; a processor configured to: determine a sleep state of the user based at least on the input; and determine settings for delivery of an olfactory output based at least on the sleep state of the user; and an output interface configured to provide instructions to the smell delivery device corresponding to the settings; wherein the smell delivery device is configured to deliver the olfactory output based on the instructions. Thus, the input associated with a sleep state of the user may be received instead of specific sensor data. In some examples, the input may include sensor data as in the first aspect. In other examples, the input may comprise a user input indicative of sleep, as explained below in relation to the seventh aspect. According to a seventh aspect of the present disclosure, there is provided a smell delivery system, comprising: a smell delivery device for delivering an olfactory output, the smell delivery device comprising: an odour chamber for containing a substance configured to produce an olfactory output; and an outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device; and a control unit, comprising: an input interface configured to receive a user input associated with sleep; a processor configured to: determine a sleep state of the user based at least on the input; and determine settings for delivery of an olfactory output based at least on the sleep state of the user; and an output interface configured to provide instructions to the smell delivery device corresponding to the settings; wherein the smell delivery device is configured to deliver the olfactory output based on the instructions. For example, the user input may indicate sleep quality of the previous sleep session. For example, the user may indicate that the sleep quality was good or bad. This information can be used to determine whether to apply the same settings or different settings. As described herein, the user input may be used to determine the sleep state, which may correspond to sleep quality (e.g., good or bad), and the settings may be selected to attempt to improve the sleep quality. Features of one aspect can be readily applied to the other aspect and vice versa. Apparatus feature of the smell delivery system of the first aspect may be applied to the method and vice versa. In particular, the method may further comprise providing the smell delivery system with one or more features described in relation to the first aspect. Embodiments of the present disclosure are described below, by way of example only, with reference to the accompanying Figures. Figure 1 shows a schematic view of a smell delivery system in accordance with a first embodiment of the present disclosure. Figure 2 shows a schematic view of a smell delivery system in accordance with a second embodiment of the present disclosure. Figure 3 shows a schematic view of a smell delivery system in accordance with a third embodiment of the present disclosure. Figure 4 shows a schematic view of a smell delivery system in accordance with a fourth embodiment of the present disclosure. Figure 5 shows a schematic view of a smell delivery system in accordance with a fifth embodiment of the present disclosure. Figure 6 shows a schematic view of a smell delivery system in accordance with a sixth embodiment of the present disclosure. Figure 7 shows a graph of predicted sleep stages based on sensor data. Figure 8 shows a flowchart corresponding to a method of operating the smell delivery system in accordance with an embodiment of the present disclosure. Figure 9 shows a flowchart corresponding to a method of operating the smell delivery system in accordance with an embodiment of the present disclosure. Figure 10 shows a perspective view of the interior of a smell delivery device for use with the smell delivery system of the present disclosure. Figure 11 shows a perspective view of the exterior of the smell delivery device of Figure 10. Referring to Figure 1, there is provided a smell delivery system 100 in accordance with a first embodiment of the present disclosure. The smell delivery system 100 is provided for use with a user 10, in particular while they are sleeping. In the first embodiment, the smell delivery system 100 comprises a sensor 102. The sensor 102 is configured to measure a signal 104 indicative of a parameter associated with a sleep state of the user 10. In the first embodiment, the parameter is a physiological parameter. Specifically, the signal 104 corresponds to the pulse rate of the user 10. Thus, the sensor 102 is configured to measure the pulse rate of the user 10. In other words, the sensor 102 is a pulse rate sensor. In other examples, the sensor 102 may be configured to measure a signal 104 indicative of a different parameter, such as a different physiological parameter or a physical parameter. In some examples, multiple sensors 102 may be used to measure different signals 104. In the first embodiment, the sensor 102 is part of a smartwatch. In other words, the pulse rate sensor 102 is built into a smartwatch. As a result, the sensor 102 is worn on the user, in particular on the user’s wrist. In this manner, the sensor 102 can be in contact with the user’s skin, allowing the pulse rate to be detected. In other examples, the sensor 102 may be worn on a headband, a wristband, or attached to clothing. In other examples, the sensor 102 may be attached directly to the user, such as via electrodes on the skin. The position of the sensor 102 and the manner of attachment may depend on the parameter measured and how it is measured. In other examples, the sensor 102 is not attached to the user, but may be arranged in the vicinity of the user. For example, a microphone may be used to measure sound associated with a user. The microphone may be arranged in another device, such as a smartphone, located within the room in which the user 10 is sleeping, but not necessarily on the user 10. In other examples, the sensor 102 is a standalone device. In some examples, the sensor 102 may be part of the control unit 106 and / or the smell delivery device 110. Thus, the sensor 102 could be built into the control unit 106. In other examples, the sensor 102 could be built into the smell delivery device 110. In yet other examples, the control unit 106 and the smell delivery device 110 could be combined, and the sensor 102 could be built into the combined device. The smell delivery system 100 also comprises a control unit 106. The control unit 106 is configured to receive sensor data 108 from the sensor 102. In other words, the sensor 102 is configured to provide the sensor data 108 to the control unit 106. The sensor data 108 is associated with the measurement of the parameter. For example, the sensor data 108 may comprise the raw signal 104. In other examples, the sensor data 108 may be processed data corresponding to the measurements of the parameter. For example, sensor data 108 may be converted from a signal 104 (such as a voltage) into a value corresponding to the parameter. In the first embodiment, the sensor data 108 corresponds to the pulse rate (in beats per minute), converted from the signal 104 (voltage). In the first embodiment, the sensor data 108 is transmitted wirelessly to the control unit 106. In this manner, the control unit 106 is physically separated from the sensor 102. In particular, the sensor data 108 is transmitted by Bluetooth™. In this manner, the control unit 106 is local to the sensor 102 and is arranged within a short distance to enable short range communication such as Bluetooth™. In the first embodiment, both the sensor 102 and the control unit 106 are provided in the same room (in which the user 10 is sleeping). In other examples, other wireless or wired technologies are possible. For example, the sensor data 108 may be transmitted by the sensor 102 to a remote location such as a cloud server, and the sensor data 108 may then be received by the control unit 106 over the internet such as by Wi-Fi™. In particular, the smartwatch may upload the sensor data 108 to a remote server, and the control unit 106 may pull this sensor data 108 from the remote server. In yet other examples, the control unit 106 may be arranged at a remote location such as a cloud server, and the sensor 102 may provide the sensor data 108 over the internet (either directly from the sensor 102, or via an intermediate device such as the smell delivery device 110 or another device such as a smartphone). In other examples, the sensor 102 may be part of the control unit 106. For example, the control unit 106 and the sensor 102 may be part of the same device, such as a smartwatch. In this case, the sensor data 108 may be transferred internally within the device from the sensor 102 to the control unit 106. The control unit 106 is configured to process the sensor data 108. In particular, the control unit 106 is configured to determine a sleep state of the user 10 based at least on the sensor data 108. In the first embodiment, the control unit 106 is configured to determine a sleep state of the user 10 based on the sensor data 108 which corresponds to the pulse rate of the user 10. In particular, if the pulse rate is below a threshold, the control unit 106 determines that the user 10 is asleep. Whereas, if the pulse rate is above a threshold, the control unit 106 determines that the user 10 is awake. In other examples, different thresholds or criteria may be used, and different sleep states may be determined. By monitoring the pulse rate, the control unit 106 determines when the user 10 falls asleep. Furthermore, since the pulse rate may be affected by different stages of sleep, the control unit 106 can determine when the user 10 enters different stages of sleep. For example, the control unit 106 may distinguish between light and deep sleep based on the pulse rate. Based on determining the sleep state of the user 10, the control unit 106 is configured to determine settings for delivery of an olfactory output 114. In particular, the smell delivery system 100 also comprises a smell delivery device 110. The smell delivery device 110 is configured to generate an olfactory output 114 for delivery to the user 10. The control unit 106 is configured to determine settings for delivery of the olfactory output 114 based at least on the sleep state of the user 10. For example, upon determining that the user 10 is in a particular sleep state, the control unit 106 can determine appropriate settings for delivery of the olfactory output 114 suited for that sleep state. For example, the control unit 106 can determine that a particular scent is desirable for this particular sleep state. The control unit 106 can then transmit instructions 112 to the smell delivery device 110 corresponding to those settings. The smell delivery device 110 is configured to receive the instructions 112. In the first embodiment, the instructions 112 are transmitted wirelessly to the smell delivery device 110. In this manner, the control unit 106 is physically separated from the smell delivery device 110. In particular, the instructions 112 are transmitted by Bluetooth™. In this manner, the control unit 106 is local to the smell delivery device 110 and is arranged within a short distance to enable short range communication such as Bluetooth™. In the first embodiment, both the smell delivery device 110 and the control unit 106 are provided in the same room (in which the user 10 is sleeping). In other examples, other wireless or wired technologies are possible. For example, the instructions 112 may be transmitted by the control unit 106 to a remote location such as a cloud server, and the instructions 112 may then be received by the smell delivery device 110 over the internet such as by Wi-Fi™. In yet other examples, the control unit 106 may be arranged at a remote location such as a cloud server, and the control unit 106 may provide the instructions 112 over the internet (either directly to the smell delivery device 110, or via an intermediate device such as the sensor 102 or another device such as a smartphone). In other examples, the control unit 106 may be part of the smell delivery device 110. For example, the control unit 106 and the smell delivery device 110 may be part of the same device. In this case, the instructions 112 may be transferred internally within the device from the control unit 106 to the smell delivery device 110. In some examples, the sensor 102 may also be part of the smell delivery device 110. For example, the smell delivery device 110 may include a sensor 102. Thus, the smell delivery device 110, the control unit 106, and the sensor 102 may all form a combined device. The smell delivery device 110 is configured to deliver an olfactory output 114 based on the instructions 112. Various olfactory outputs 114 may be possible depending on the instructions 112. In some examples, the same olfactory output 114 can be used to provide the same smell, but the delivery of that olfactory output 114 can be adjusted. For example, the intensity of the olfactory output 114 can be controlled (such as by adjusting the flow rate or concentration of the odour). In some examples, a combination of olfactory outputs 114 can be selected. In some examples, the timing of delivery may be controlled, for example changing the duration of delivery of pulses of olfactory output 114, changing the timing between pulses, or changing the total duration over which multiple pulses of the same olfactory output 114 are delivered. In the first embodiment, the smell delivery device 110 comprises an odour chamber containing a plurality of removable canisters each containing a substance configured to produce an olfactory output 114. The smell delivery device 110 also comprises an outlet connected to the odour chamber and configured to release the substance from the smell delivery device 110. The smell delivery device 110 also comprises a flow generator, in the form of a pump. The flow generator is configured to generate a flow of carrier gas, in the form of air, and provide this carrier gas to the odour chamber. Based on the instructions 112, the smell delivery device 110 is configured to apply settings to control delivery of the olfactory output 114. Specifically, the smell delivery device 110 is configured to select one or more removable canisters for providing the olfactory output 114. In the first embodiment, the smell delivery device 110 causes release of the selected substance by operating one or more valves to allow release of the substance from the removable canister and into the odour chamber. The smell delivery device 110 also controls the flow generator to provide the flow of carrier gas at the desired level to pick up the substance from the odour chamber and deliver the olfactory output 114 from the outlet towards the user 10. For example, the olfactory output 114 can be directed to the vicinity of the user 10 as they sleep. In this manner, the user 10 can receive a particular olfactory output 114 based on the sleep state of the user 10. For example, the user 10 can receive a particular scent based on their sleep state. In other examples, the intensity or duration of a scent can be based on their sleep state. By controlling delivery of the olfactory output 114, the sleep state can be influenced. In some examples, this can be used to improve sleep quality. For example, certain smell delivery adjustments may be used to affect the transition into different sleep states (such as between phases of sleep), or to trigger waking up at a desired time. The desired time may be determined by an alarm setting, or it may be determined to coincide with a sleep state which is desirable to wake up from (for example, because this will allow the user to feel more awake and refreshed). The delivery of the olfactory output 114 can also be selected based on various sleep events detected from the sensor data 108. For example, in response to the user 10 transitioning from deep sleep to awake, the olfactory output 114 can be adjusted to encourage the user 10 falling asleep again. The olfactory output 114 can be selected based on a database of stored settings. For example, the database may include default settings such as particular scents for particular sleep states or particular sleep events. As one example, a default setting may be to provide the smell of lavender when the user 10 awakes during the sleep session. The database may also include user-specific settings, including smells that the user prefers, either by user feedback, or by determining effectiveness of those previous olfactory outputs 114 based on historical data. Therefore, since the smell delivery system 100 receives sensor data 108, determines a sleep state based on the sensor data 108, and determines setting for delivery of the olfactory output 114 based on the sleep state, the smell delivery system 100 is a platform that provides an end-to-end solution. Referring to Figure 2, there is provided a smell delivery system 200 in accordance with a second embodiment of the present disclosure. The smell delivery system 200 of the second embodiment is similar to the first embodiment, except where described below. The same reference numerals are used to denote identical features. One or more features described in relation to the first embodiment may readily be applied to the second embodiment, and vice versa. The second embodiment is the same as the first embodiment, except that the smell delivery system 200 of the second embodiment additionally includes an environmental sensor 216 configured to measure a signal 218 indicative of an environmental parameter, and the environmental sensor 216 is configured to transmit environmental sensor data 220 to the control unit 106 based on the signal 218. In the same manner as the first embodiment, the smell delivery system 200 comprises a sensor 102 configured to measure a signal 104 indicative of a parameter associated with a sleep state of the user 10. The sensor 102 is configured to transmit sensor data 108 to a control unit 106, wherein the sensor data 108 corresponds to the measured signal 104 indicative of the parameter 104. The control unit 106 is configured to determine a sleep state of the user 10 based on the received sensor data 108, as in the first embodiment. The sensor 102 may be any sensor for measuring a parameter associated with the sleep state, such as the pulse rate sensor 102 of the first embodiment, or any other sensor as described herein. The sensor 102 and the control unit 106 may have any arrangement, such as the sensor 102 and the control unit 106 both being local devices communicating wirelessly, the control unit 106 being remote, or the sensor 102 and the control unit 106 being arranged on the same device. In the second embodiment, the smell delivery system 200 also comprises an environmental sensor 216. The environmental sensor 216 is configured to measure a signal 218 indicative of an environmental parameter in an environment of the user 10. In the second embodiment, the environmental parameter is humidity. Thus, the environmental sensor 216 is configured to measure the humidity of the environment of the user 10. In other words, the environmental sensor 216 is a humidity sensor. In other examples, the environmental sensor 216 may be configured to measure a signal 218 indicative of a different environmental parameter. In some examples, multiple environmental sensors 216 may be used to measure different signals 218. In the second embodiment, the environmental sensor 216 is also part of a smartwatch. In other words, the humidity sensor 216 is built into the smartwatch. In this manner, the smartwatch comprises the sensor 102 and the environmental sensor 216. In other examples, the environmental sensor 216 may be separate from the sensor 102. For example, the environmental sensor 216 may be worn on a headband, a wristband, or attached to clothing. In other examples, the environmental sensor 216 may be attached directly to the user. In other examples, the environmental sensor 216 is not attached to the user, but may be arranged in the vicinity of the user. In particular, since the environmental sensor 216 is configured to measure the humidity (or other parameter) of the environment, it need not be arranged on the user 10. In some examples, the environmental sensor 216 may be part of the control unit 106 and / or the smell delivery device 110. Thus, the environmental sensor 216 could be built into the control unit 106. In other examples, the environmental sensor 216 could be built into the smell delivery device 110. In yet other examples, the control unit 106 and the smell delivery device 110 could be combined, and the environmental sensor 216 could be built into the combined device. The environmental sensor 216 is configured to provide environmental sensor data 220 to the control unit 106. In other words, the control unit 106 is configured to receive the environmental sensor data 220 from the environmental sensor 216. The environmental sensor data 220 is associated with the measurement of the environmental parameter. For example, the environmental sensor data 220 may comprise the raw signal 218. In other examples, the environmental sensor data 220 may be processed data corresponding to the measurements of the environmental parameter. In the second embodiment, the environmental sensor data 220 is transmitted wirelessly to the control unit 106. In this manner, the control unit 106 is physically separated from the environmental sensor 216. In particular, the environmental sensor data 220 is transmitted by Bluetooth™. In this manner, the control unit 106 is local to the environmental sensor 216 and is arranged within a short distance to enable short range communication such as Bluetooth™. In the second embodiment, both the environmental sensor 216 and the control unit 106 are provided in the same room (in which the user 10 is sleeping). In other examples, other wireless or wired technologies are possible. For example, the environmental sensor data 220 may be transmitted by the environmental sensor 216 to a remote location such as a cloud server, and the environmental sensor data 220 may then be received by the control unit 106 over the internet such as by Wi-Fi™. In particular, the smartwatch may upload the environmental sensor data 220 to a remote server, and the control unit 106 may pull this environmental sensor data 220 from the remote server. In yet other examples, the control unit 106 may be arranged at a remote location such as a cloud server, and the environmental sensor 216 may provide the environmental sensor data 220 over the internet (either directly from the environmental sensor 216, or via an intermediate device such as the smell delivery device 110 or another device such as a smartphone). In other examples, the environmental sensor216 may be part of the control unit 106. For example, the control unit 106 and the environmental sensor 216 (and possibly also the sensor 102) may be part of the same device, such as a smartwatch. In this case, the environmental sensor data 220 may be transferred internally within the device from the environmental sensor 216 to the control unit 106. As with the sensor data 108 in the first embodiment, the control unit 106 is configured to process the environmental sensor data 220. In particular, the control unit 106 is configured to determine an impact of the environmental parameter on the delivery of the olfactory output 114. Since the user perception of smell is affected by environmental parameters, it can be useful to compensate for this. For example, the ability of the user 10 to detect a smell, or the perceived intensity of a smell, can be influenced by environmental parameters such as humidity. Therefore, by considering the environmental conditions, adjustments to delivery of the olfactory output 114 can be made to ensure that the smell is delivered at the desired level. In the second embodiment, the control unit 106 is configured to determine that the humidity is high, and that this will have a negative impact on the delivery of the olfactory output 114. In particular, under high humidity, the user 10 has a decreased sense of smell compared to low humidity. Based on this, the control unit 106 is configured to determine the settings for delivery of the olfactory output 114 based at least on the impact of the environmental parameter on the delivery of the olfactory output 114. Therefore, the control unit 106 considers not just the sleep state in determining the settings, but further modifies this to compensate for environmental conditions. In the second embodiment, the control unit 106 is configured to determine settings corresponding to a decrease in the intensity of the olfactory output 114 based on the determined impact of the environmental settings. The control unit 106 can then transmit instructions 112 to the smell delivery device 110, as in the first embodiment. In some examples, the environmental sensor 216 may also be part of the smell delivery device 110. For example, the smell delivery device 110 may include an environmental sensor 216. Thus, the smell delivery device 110, the control unit 106, and the environmental sensor 216 (and possibly also the sensor 102) may all form a combined device. Referring to Figure 3, there is provided a smell delivery system 300 in accordance with a third embodiment of the present disclosure. The smell delivery system 300 of the third embodiment is similar to the first and second embodiments, except where described below. The same reference numerals are used to denote identical features. One or more features described in relation to the first and second embodiments may readily be applied to the third embodiment, and vice versa. The third embodiment is the same as the first embodiment, except that the smell delivery system 300 of the third embodiment additionally includes user input 322. The user input 322 is provided by the user 10 to the control unit 106. The user input 322 corresponds to feedback from the user 10 indicating sleep quality. In particular, after a sleep session, the user 10 can provide user input 322 to indicate their perception of sleep quality. The user input 322 can indicate how long the user slept, how well they thought they slept, and to flag any particular sleep events, such as when they awoke during the sleep session. The user input 322 can also provide feedback on smell delivery. In particular, the user input 322 can indicate that the user liked or disliked a particular olfactory output 114. In the third embodiment, the user 10 is able to provide the user input 322 from a device, such as a smartphone. The user input 322 is then transmitted from the device to the control unit 106. For example, the user input 322 may be transmitted wirelessly, such as over Bluetooth™ if the control unit 106 is local to the user 10, or over the internet such as by Wi-Fi™. If the control unit 106 is remote, such as on a cloud server, the user input 322 may be transmitted over the internet to the control unit 106. In other examples, the user input 322 may be provided directly. For example, the control unit 106 may comprise a user interface (such as a keyboard, touchscreen, or voice input) allowing the user 10 to directly provide the user input 322 to the control unit 106. Alternatively, the smell delivery device 110 may comprise a user interface for receiving the user input 322, and the user input 322 may be transmitted to the control unit 106 (or provided internally if the control unit 106 is combined with the smell delivery device 110). The user input 322 may be used by the control unit 106 as feedback on the sleep session. This may provide additional context to assist in identifying sleep events or characterising sleep states. This may also provide feedback on how the user 10 perceived the olfactory output 114. If the user 10 indicated restful sleep, and / or specifically that they enjoyed the particular scent, then the control unit 106 can take this into account in deciding whether to provide that olfactory output 114 again. Similarly, if the user 10 indicates that the intensity of the olfactory output 114 was too high, then the control unit 106 can adjust delivery of the olfactory output 114 next time to reduce the intensity. In some examples, the user input 322 of the third embodiment may be provided in combination with the environmental sensor 216 of the second embodiment. Referring to Figure 4, there is provided a smell delivery system 400 in accordance with a fourth embodiment of the present disclosure. The smell delivery system 400 of the fourth embodiment is similar to the first to third embodiments, except where described below. The same reference numerals are used to denote identical features. One or more features described in relation to the first to third embodiments may readily be applied to the fourth embodiment, and vice versa. The fourth embodiment is the same as the first embodiment, except that the smell delivery system 400 of the fourth embodiment additionally includes historical database 424. The historical database 424 is used to store historical sleep data. In the fourth embodiment, the control unit 106 is configured to provide historical sleep data 426 to the historical database 424. In the fourth embodiment, the historical sleep data 426 comprises the sensor data 108. In particular, after the control unit 106 receives the sensor data 108 from the sensor 102, the control unit 106 forwards the sensor data to the historical database 424 as historical sleep data 426. The historical database 424 can therefore store the sensor data 108 over time. In some examples, the control unit 106 provides the historical sleep data 426 at regular intervals, such as whenever it receives the sensor data 108 from the sensor 102. In the fourth embodiment, the historical sleep data 426 also comprises the sleep state. After the control unit 106 determines the sleep state based on the sensor data 108, the sleep state can be included in the historical sleep data 426 and provided to the historical database 424. In some examples, the sleep state is associated with the sensor data 108 so that the historical database 426 can store the corresponding sensor data 108 with the sleep state over time. In other examples, the sleep state may be omitted, and the historical sleep data 426 provides the sensor data 108 instead. In other examples, the sensor data 108 may be omitted, and the historical sleep data 426 provides the sleep state instead. In the fourth embodiment, the historical sleep data 426 also comprises the current settings for olfactory output 114. In particular, the settings for delivery of the olfactory output 114 corresponding to the sensor data 108 can be provided by the control unit 106 as part of the historical sleep data 426. In some examples, the control unit 106 is configured to determine an impact of delivery of the olfactory output 114 on the sleep state. For example, the control unit 106 can determine an impact on the sensor data 108 as a result of the olfactory output 114. For example, the control unit 106 can determine whether the selection of a particular scent resulted in a decrease in heart rate. Alternatively, the control unit 106 can determine whether the selection of the particular scent resulted in a change of sleep state based on the sensor data. If the olfactory output 114 had the desired impact on the sensor data or sleep state, the control unit 106 can include that impact as part of the historical sleep data 426. This information can be provided and stored at the historical database 424. Thus, the impact of delivery of the olfactory output 114 may be determined when the sensor data 108 is received, and can be stored in the historical database 424. Alternatively, the impact may not be determined at that point. The control unit 106 is also configured to receive historical sleep data 428 from the historical database 424. In some examples, the control unit 106 is configured to receive the historical sleep data 428 upon request. In the fourth embodiment, the historical sleep data 428 indicates sleep data associated with the sleep state. For example, in response to the control unit 106 determining the sleep state based on the sensor data 108, the control unit 106 requests historical sleep data 428 corresponding to the sleep state. For example, the sensor data 108 may indicate that the heart rate was low and has now increased. The control unit 106 may determine that this indicates a change from deep sleep to awake. Based on this information, the control unit 106 can obtain historical sleep data 428 corresponding to similar situations. For example, the historical sleep data 428 may comprise the settings for delivery of the olfactory output 114 in previous situations where the heart rate increased. Based on subsequent sensor data 108 or sleep state information, it can be determined whether those settings were successful in achieving the desired outcome (such as encouraging the user to fall back asleep). This determination may be performed by the control unit 106 in realtime in response to receiving the historical sleep data 428. Alternatively, the determination may be performed by the control unit 106 as sensor data 108 is received, and that determination may be included in the historical sleep data 426 and stored in the historical database 424. Therefore, the control unit 106 may use historical sleep data 428 to determine whether particular settings were successful or not. Based on this information, the control unit 106 can select particular settings for delivery of the olfactory output 114. For example, if the historical sleep data 428 suggests that a particular scent had the desired effect in a similar situation, the control unit 106 may determine to provide the same scent this time. Providing the historical sleep data 426 to the historical database 424 may involve transmission to an external device, such as wireless transmission as disclosed herein. Alternatively, the historical database 424 may be provided as part of the control unit 106. The historical database 424 may be local, such as stored in memory on the control unit 106 or the smell delivery device 110, or it may be remote, such as in cloud storage. Thus, providing the historical sleep data 426 may involve storing in memory. Similarly, providing the historical sleep data 428 may involve transmission or internal communication. Thus, providing the historical sleep data 428 may involve retrieving from memory. In some examples, the historical sleep data of the fourth embodiment may be provided in combination with the environmental sensor 216 of the second embodiment and / or the user input 322 of the third embodiment. For example, the environmental sensor data 220 may be provided as part of the historical sleep data 426 and stored in the historical database 424. Referring to Figure 5, there is provided a smell delivery system 500 in accordance with a fifth embodiment of the present disclosure. The smell delivery system 500 of the fifth embodiment is similar to the first to fourth embodiments, except where described below. Similar reference numerals are used to denote similar features. One or more features described in relation to the first to fourth embodiments may readily be applied to the fifth embodiment, and vice versa. The smell delivery system 500 comprises a smartwatch 502. The smartwatch 502 comprises a plurality of sensors similar to the sensor 102 of the first to fourth embodiments. In other examples, the sensors need not be provided in a smartwatch 502. For example, the sensor may be provided in the smell delivery device 510. The smartwatch 502 is configured to measure sleep data 504 from the user 510 while they sleep. In the fifth embodiment, the sleep data 504 comprises a plurality of sensor measurements including pulse rate to measure the pulse rate of the user 510 and an accelerometer to measure movement of the user 510. The sleep data 504 is measured in real-time by the smartwatch 502. In the fifth embodiment, the smell delivery system 500 also comprises a sleep app 506. The smartwatch 502 is configured to provide biofeedback 508 to the sleep app 506. The sleep app 506 is an application running on a cloud platform. The sleep app 506 is similar to the control unit 106 of the first to fourth embodiments. The biofeedback 508 corresponds to the measurements of the sensors as defined by the sleep data 504. In the fifth embodiment, the biofeedback 508 is provided by Wi-Fi™. In particular, the smartwatch 502 connects to the internet via Wi-Fi™, and provides the biofeedback 508 to the sleep app 506 over the internet. The smell delivery system 500 also comprises a smell delivery device 510. The smell delivery device 510 is similar to the smell delivery device of the first to fourth embodiments. The smell delivery device 510 also comprises an environmental sensor 516 similar to the environmental sensor 216 of the second embodiment. The smell delivery device 510 is configured to provide environmental feedback 520 to the sleep app 506 corresponding to measurements of the environmental sensor 516. The environmental feedback 520 may be provided over Wi-Fi™ in a similar manner to the biofeedback 508. In other examples, the environmental sensor 516 and the environmental feedback 520 may be omitted. The sleep app 506 also provides a sleep feedback request 530 to the user 510 or providing user input. The user 510 then provides sleep feedback 522 to the sleep app 506. The sleep feedback 522 is similar to the user input 322 of the third embodiment. The sleep feedback 522 may indicate user feedback corresponding to sleep quality or scent delivery. For example, the sleep feedback 522 may be provided via a device of the user 510 such as a smartphone, connected to the sleep app 506 over the internet. The sleep feedback request 530 may also be received at the user device from the sleep app 606. In some examples, the sleep feedback request 530 may be omitted, and the user 510 is able to provide sleep feedback 522 when desired. In other examples, the sleep feedback 522 may be omitted. The sleep app 506 is configured to process the biofeedback 508, the environmental feedback 520, and the sleep feedback 522 to determine settings for delivery of an olfactory output 514. For example, based on the real-time biofeedback 508, the sleep app 506 may determine that the user 510 has exited a period of deep sleep and is awake. The sleep app 506 may therefore determine to provide a particular scent to encourage the user 510 to fall back asleep. Based on predetermined settings, as well as sleep feedback 522 indicating impact of previous scents on sleep quality and user preferences, the sleep app 506 can determine settings for delivery of the olfactory output 514, such as selection of a particular scent for delivery at a particular intensity. Based on the environmental feedback 520, the sleep app 506 can also adjust the settings. For example, in response to determining that the humidity is high, the sleep app 506 may determine that perception of the olfactory output 514 would be reduced under those environmental conditions. The sleep app 506 may therefore increase the intensity of the olfactory output 514 to compensate for this. The sleep app 506 is then configured to provide a scent release request 512 to the smell delivery device 510. The scent release request 512 is similar to the instructions 112 of the first to fourth embodiments. In the fifth embodiment, the smartwatch 502 is also configured to provide sleep tracking results 532 to the user 510. The sleep tracking results 532 may be based solely on the sleep data 504. For example, the sleep tracking results 532 may indicate the sensor data over the sleep session. Alternatively, the sleep tracking results 532 may include information about sleep state, for example received from the sleep app 506 after processing. The sleep tracking results 532 may be displayed on a display of the smartwatch 502, for example. In other examples, the sleep tracking results 532 may be omitted. Referring to Figure 6, there is provided a smell delivery system 600 in accordance with a sixth embodiment of the present disclosure. The smell delivery system 600 of the sixth embodiment is similar to the first to fifth embodiments, except where described below. Similar reference numerals are used to denote similar features. One or more features described in relation to the first to fifth embodiments may readily be applied to the fifth embodiment, and vice versa. The sixth embodiment is the same as the fifth embodiment, except that the smartwatch 602 provides biofeedback 608a to the smell delivery device 610 via Bluetooth™, which then provides the biofeedback 608b to the sleep app 606 via Wi-Fi™, rather than the smartwatch 502 providing the biofeedback 508 directly to the sleep app 506 via Wi-Fi™. The smell delivery system 600 comprises a smartwatch 602 as in the fifth embodiment, but in other examples the sensor(s) may be provided other than in a smartwatch. The smartwatch 602 obtains real-time sleep data 604 corresponding to sensor measurements. The smartwatch 602 also provides sleep tracking results 632 for display to the user 610, although this may be omitted. The smartwatch 602 provides biofeedback 608a to the smell delivery device 610 by Bluetooth™. In other examples, other short-range communication technologies may be used. This avoids the need for the smartwatch 602 to be connected to the internet. The smell delivery device 610 comprises an environmental sensor 616 as in the fifth embodiment. The smell delivery device 610 then provides the biofeedback 608b along with the environmental feedback 620 to the sleep app 606, such as via the internet. In some examples, the environmental feedback 620 may be omitted. The sleep app 606 provides a sleep feedback request 630 to a user device, and the user 610 can respond with sleep feedback 622 as in the fifth embodiment. In some examples, the sleep feedback request 630, and optionally the sleep feedback 622, may be omitted. The sleep app 606 then provides a scent release request 612 as in the fifth embodiment. The smell delivery device 610 then releases the scent 614 as in the fifth embodiment. Referring to Figure 7, there is provided a graph 700. The graph 700 includes a sleep stage plot 734, a heart rate plot 736, and an accelerometer plot 738. The heart rate plot 736 and the accelerometer plot 738 correspond to sensor measurements provided to the control unit, such as the control unit 106 or sleep app 506, 606. For example, the control unit 106, such as in the first embodiment, may be configured to receive sensor data 108 from multiple sensors 102. In particular, the sensors 102 include a heart rate sensor and an accelerometer. The heart rate sensor measures the heart rate of the user 10, and the accelerometer measures the movement of the user. After the control unit 106 receives the sensor data 108, it can process and store the sensor data 108. The heart rate plot 736 shows the heart rate in beats per minute over time. The accelerometer plot 638 shows the acceleration (in X, Y, and Z axis) over time. As described herein, the control unit 106 can also determine a sleep state based on the sensor data 108. For example, the control unit 106 can determine a sleep stage. The sleep stage plot 734 shows the sleep stages (including awake, REM, light sleep, and deep sleep) over time. Values and changes in the sensor data 108 can indicate particular sleep stages and changes between sleep stages. Sleep events, such as the user waking up, can also be indicated from the sensor data 108. The control unit 106 can use a classification algorithm on the sensor data. For example, the classification algorithm may be a random forest. This can predict and label the sleep stage based on the sensor data. The sensor data may be taken over a particular time period, such as every 30 seconds (which is a standard epoch length in sleep research). The sleep state can be determined for each epoch. In particular, as can be seen from the graph 700, periods of low heart rate as shown in the heart rate plot 736 can be indicative of deep sleep. In addition, periods of low movement as shown in the accelerometer plot 738 can be indictive of deep sleep. In combination, the two sets of sensor data 108 can allow more accurate determination of sleep state. For example, spikes in heart rate and movement can indicate changes in sleep state. For example, at approximately 03:00 am, the user enters a period of deep sleep, which corresponds to a low heart rate and low movement. This is followed by a set of spikes in heart rate and movement, which translates to the user waking up, entering a brief period of light sleep, and waking up again. After about 04:30 am, the heart rate and movement drops again and the user enters another period of deep sleep. Alternatively, changes in one parameter but not another may indicate a particular change. For example, before about 04:30 am, the heart rate increases, but there is little change in the movement. This indicates the user coming out of deep sleep, but not awaking. Therefore, the combination of different sensors, especially at least one sensor measuring a physiological parameter and at least one sensor measuring a physical parameter, can be particularly useful in determining a specific sleep stage. The control unit 106 may therefore operate to determine sleep state based on data from one or more different sensors, not limited to the heart rate and accelerometer shown in Figure 7. Referring to Figure 8, there is provided a method 800. The method 800 may be performed by the smell delivery system according to any of the first to sixth embodiments. The method 800 begins by starting and loading a sleep smell application. If a stop button is pressed, the method 800 ends. If the stop button is not pressed, the method continues. If the method continues, environmental and biofeedback data is collected from sensors. Algorithms are then used to determine a sleep state. In particular, undesirable sleep events may be detected. For example, if the biofeedback data suggests that the heart rate has increased and the user has awoken, an undesirable sleep event may be detected. If an undesirable sleep event is not detected, then sleep can continue as before without adjustments in smell delivery. The method can return to check whether the stop button has pressed. If the stop button has not been pressed, the method can repeat and collect more sensor data. If undesirable sleep events are detected, the data can be transferred to an engine. Algorithms are used to determine scent selection and delivery instructions. Based on the undesirable sleep events, the algorithms can determine settings for delivery of the olfactory output. For example, this can be based on a database of known settings for certain undesirable sleep events, which may be populated with pre-set values. This data may also be tailored to a user based on indicated preferences or historical data. For example, if particular settings have resolved undesirable sleep events in the past, those settings may be more likely to be selected again. After scent selection and determining delivery instructions, delivery of the olfactory output is triggered. This results in the smell delivery device delivering the olfactory output. After delivery of the olfactory output, the sensors collect environmental and biofeedback data. This data is transferred to the engine. This can be used for storing historical data, for example to determine an impact of the settings on sensor data, for example indicating whether the selected settings resolved the undesirable sleep event or not. If the event process has not finished, the method returns to collection of sensor data until the process has finished. If the event process has finished, the data is transferred to the engine. The method then returns to check if the stop button has been pressed. If the stop button has been pressed, then the method concludes. If the stop button has not been pressed, then the method repeats, by collecting more sensor data and detecting undesirable sleep events. Referring to Figure 9, there is provided a method 900. The method 900 may be performed by the smell delivery system according to any of the first to sixth embodiments. The method 900 is the same as the method 800, except that it further includes user input. Specifically, after the events process finishes, the method involves asking a question to the user. For example, the question may be related to feedback on the user’s sleep. In response to asking the question, the user provides an answer. For example, the answer may comprise user feedback indicating a quality of the sleep. Referring to Figure 10, a smell delivery device 1010 is provided, for use with embodiments of the present disclosure. For example, the smell delivery device 1010 may be used as the smell delivery device 110 of the first to fourth embodiments, the smell delivery device 510 of the fifth embodiment, or the smell delivery device 610 of the sixth embodiment. The smell delivery device 1010 has an inlet 1040 which provides air into an inlet channel. The air passes through a first filter 1042a and along pipework to an airflow generator 1044. The airflow is then passed through a second filter 1042b and to a manifold 1048 including an array of outlet valves 1050. The outlet valves 1050 are connected to an array of removable canisters 1052 which comprise substances to produce an olfactory output. The airflow generator 1044 can therefore be used to draw air into the smell delivery device 1010, which is cleaned by the filters 1042a, 1042b, and provided to the canisters 1052 for collecting the substance. A fan 1046 is used for cooling the components. In some examples, one or more of the filters 1042a, 1042b may be omitted. In some examples, the fan 1046 may be omitted. Referring to Figure 11, the smell delivery device 1010 of Figure 10 is shown with an outer casing 1054 applied. The canisters 1052 are connected to an outlet 1058. The outlet 1058 includes an array of outlet apertures 1060. This allows the air supplied to the canisters 1052 to be output at the outlet apertures 1060, carrying the one or more substances from the canisters 1052 for output at the outlet apertures 1060. The olfactory output can then be provided to the user. Embodiments of the present disclosure may include a smell delivery device, such as the smell delivery device 1010. The smell delivery device 1010 may be used to deliver the olfactory output. The settings for delivery of the olfactory output may involve adjustments to selection of particular canisters 1052 to deliver a particular olfactory output, for example by controlling outlet valves 5 1050. In other examples, the settings may involve adjustments to the flow rate provided by the airflow generator 1044.

Claims

1. A smell delivery system, comprising:a smell delivery device for delivering an olfactory output, the smell delivery device comprising:an odour chamber for containing a substance configured to produce an olfactory output; andan outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device; and a control unit, comprising:an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user;a processor configured to:determine a sleep state of the user based at least on the sensor data; anddetermine settings for delivery of an olfactory output based at least on the sleep state of the user; andan output interface configured to provide instructions to the smell delivery device corresponding to the settings;wherein the smell delivery device is configured to deliver the olfactory output based on the instructions.

2. The smell delivery system according to claim 1, wherein the parameter associated with a sleep state of the user comprises a physiological parameter associated with a sleep state of the user.

3. The smell delivery system according to claim 2, wherein the physiological parameter comprises: a heart rate of the user, a blood pressure of the user, a blood volume of the user, a blood oxygen saturation of the user, brain activity of the user, a body temperature of the user, sweating of the user, body composition of the user, and / or breathing of the user.

4. The smell delivery system according to any preceding claim, wherein the parameter associated with a sleep state of the user comprises a physical parameter associated with a sleep state of the user.

5. The smell delivery system according to claim 4, wherein the physical parameter comprises: a position of the user and / or movement of the user.

6. The smell delivery system according to any preceding claim, wherein the sleep state comprises a sleep stage.

7. The smell delivery system according to any preceding claim, wherein the processor is configured to determine the settings for delivery of the olfactory output to facilitate maintaining the sleep state or to facilitate transitioning the sleep state.

8. The smell delivery system according to any preceding claim, wherein the processor is further configured to determine a predicted subsequent sleep state, and wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the predicted subsequent sleep state.

9. The smell delivery system according to any preceding claim, wherein the processor is further configured to determine a progress of the sleep state, and wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the progress of the sleep state.

10. The smell delivery system according to any preceding claim, wherein the input interface is further configured to receive subsequent sensor data from the at least one sensor, and wherein the processor is further configured to determine an impact of the settings for delivery of the olfactory output on the sleep state based at least on the subsequent sensor data.

11. The smell delivery system according to claim 10, wherein the processor is further configured to determine subsequent settings for delivery of the olfactory output based at least on the impact of the settings on the sleep state.

12. The smell delivery system according to any preceding claim, wherein the processor is further configured to determine the settings for delivery of the olfactory output based at least on an impact of the previous settings on the sleep state.

13. The smell delivery system according to claim 12, wherein the processor is configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state, optionally wherein the processor is configured to determine the impactof the previous settings for delivery of the olfactory output on the sleep state based at least on historical sleep data.

14. The smell delivery system according to any preceding claim, wherein the input interface is further configured to receive a user input from the user indicating user feedback data associated with sleep, optionally wherein the processor is further configured to determine the settings for delivery of the olfactory output based at least on an impact of the previous settings on the sleep state and wherein the processor is configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state based at least on the user feedback data.

15. The smell delivery system according to any preceding claim, wherein the input interface is further configured to receive environmental sensor data from at least one environmental sensor configured to measure a signal indicative of an environmental parameter in an environment of the user.

16. The smell delivery system according to claim 15, wherein the processor is further configured to determine an impact of the environmental parameter on the delivery of the olfactory output, based at least on the environmental sensor data.

17. The smell delivery system according to claim 16, wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the impact of the environmental parameter on the delivery of the olfactory output.

18. The smell delivery system according to any preceding claim, wherein the settings for delivery of the olfactory output correspond to one or more of: initiating delivery of an olfactory output, ceasing delivery of an olfactory output, selecting a particular olfactory output or combination of olfactory outputs for delivery, selecting an intensity of an olfactory output for delivery, and / or selecting a timing of an olfactory output for delivery.

19. The smell delivery system according to any preceding claim, wherein the smell delivery device comprises a flow generator configured to provide a flow of carrier gas to the odour chamber for releasing the substance, and wherein the settings for delivery of the olfactory output correspond to adjusting a flow rate of the flow generator.

20. The smell delivery system of any preceding claim, wherein the control unit is integral with the smell delivery device.

21. The smell delivery system of any preceding claim, wherein the control unit is separate from the smell delivery device.

22. The smell delivery system of any preceding claim, further comprising the at least one sensor, optionally wherein the sensor is integral with the smell delivery device and / or the control unit.

23. A method of delivering an olfactory output, comprising:receiving sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user;determining a sleep state of the user based at least on the sensor data;determining settings for delivery of an olfactory output based at least on the sleep state of the user; andproviding instructions to a smell delivery device corresponding to the settings.

24. The method of claim 23, further comprising delivering, by the smell delivery device, the olfactory output based on the instructions.

25. A computer program, computer program product, or computer-readable medium comprising instructions which, when executed by a processor, cause the processor to perform the method of claim 23 or 24.AMENDMENTS TO THE CLAIMS HAVE BEEN FILED AS FOLLOWS:31 07 25Claims:

1. A smell delivery system, comprising:a smell delivery device for delivering an olfactory output, the smell delivery device comprising:an odour chamber for containing a substance configured to produce an olfactory output; andan outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device; and a control unit, comprising:an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user;a processor configured to:determine a sleep state of the user based at least on the sensor data, wherein the processor is configured to determine multiple sleep states from the sleep stages: REM, light sleep and deep sleep; anddetermine settings for delivery of an olfactory output based at least on the sleep state of the user, wherein determining the settings includes using a database having settings for particular sleep scents for particular sleep stages which obtain pre-determined effects on the user’s sleep; andan output interface configured to provide instructions to the smell delivery device corresponding to the settings, wherein the smell delivery device is configured to deliver the olfactory output based on the instructions, andwherein the input interface is further configured to receive subsequent sensor data from the at least one sensor, and wherein the processor is further configured to determine an impact of the settings for delivery of the olfactory output on the sleep state based at least on the subsequent sensor data.

2. A smell delivery system, comprising:a smell delivery device for delivering an olfactory output, the smell delivery device comprising:an odour chamber for containing a substance configured to produce an olfactory output; and31 07 25an outlet connected to the odour chamber and configured to release the substance for delivering the olfactory output from the smell delivery device; and a control unit, comprising:an input interface configured to receive sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user;a processor configured to:determine a sleep state of the user based at least on the sensor data, wherein the processor is configured to determine multiple sleep states from the sleep stages: REM, light sleep and deep sleep; anddetermine settings for delivery of an olfactory output based at least on the sleep state of the user and an impact of previous settings on the sleep state, wherein determining the settings includes using a database having settings for particular sleep scents for particular sleep stages which obtain pre-determined effects on the user’s sleep;an output interface configured to provide instructions to the smell delivery device corresponding to the settings, wherein the smell delivery device is configured to deliver the olfactory output based on the instructions.

3. A smell delivery system according to claim 1 wherein the settings in the database are tailored to a user based on the user’s indicated preferences or historical data.

4. A smell delivery system according to claim 1 or 2 wherein the system includes an engine having algorithms which the processor uses to determine the settings for delivery of the olfactory output.

5. A smell delivery system according to claim 4 wherein the engine includes environmental and biofeedback data which are used by the engine to determine the settings for delivery of the olfactory output.

6. A smell delivery system according to any preceding claim wherein the processor is configured to determine settings which improve the sleep quality of the user’s sleep.

7. The smell delivery system according to any preceding claim, wherein the parameter associated with a sleep state of the user comprises a physiological parameter associated with a sleep state of the user.31 07 258. The smell delivery system according to claim 7, wherein the physiological parameter comprises: a heart rate of the user, a blood pressure of the user, a blood volume of the user, a blood oxygen saturation of the user, brain activity of the user, a body temperature of the user, sweating of the user, body composition of the user, and / or breathing of the user.

9. The smell delivery system according to any preceding claim, wherein the parameter associated with a sleep state of the user comprises a physical parameter associated with a sleep state of the user.

10. The smell delivery system according to claim 9, wherein the physical parameter comprises: a position of the user and / or movement of the user.

11. The smell delivery system according to any preceding claim, wherein the processor is configured to determine the settings for delivery of the olfactory output to facilitate maintaining the sleep state or to facilitate transitioning the sleep state.

12. The smell delivery system according to any preceding claim, wherein the processor is further configured to determine a predicted subsequent sleep state, and wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the predicted subsequent sleep state.

13. The smell delivery system according to any preceding claim, wherein the processor is further configured to determine a progress of the sleep state, and wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the progress of the sleep state.

14. The smell delivery system according to any preceding claim, wherein the processor is further configured to determine subsequent settings for delivery of the olfactory output based at least on the impact of the settings on the sleep state.

15. The smell delivery system according to any preceding claim, wherein the processor is configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state, and optionally wherein the processor is configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state based at least on historical sleep data.31 07 2516. The smell delivery system according to any preceding claim, wherein the input interface is further configured to receive a user input from the user indicating user feedback data associated with sleep, optionally wherein the processor is further configured to determine the settings for delivery of the olfactory output based at least on an impact of the previous settings on the sleep state and wherein the processor is configured to determine the impact of the previous settings for delivery of the olfactory output on the sleep state based at least on the user feedback data.

17. The smell delivery system according to any preceding claim, wherein the input interface is further configured to receive environmental sensor data from at least one environmental sensor configured to measure a signal indicative of an environmental parameter in an environment of the user, and optionally or preferably the processor is further configured to determine an impact of the environmental parameter on the delivery of the olfactory output, based at least on the environmental sensor data.

18. The smell delivery system according to claim 17, wherein the processor is configured to determine the settings for delivery of the olfactory output based at least on the impact of the environmental parameter on the delivery of the olfactory output.

19. The smell delivery system according to any preceding claim, wherein the settings for delivery of the olfactory output correspond to one or more of: initiating delivery of an olfactory output, ceasing delivery of an olfactory output, selecting a particular olfactory output or combination of olfactory outputs for delivery, selecting an intensity of an olfactory output for delivery, and / or selecting a timing of an olfactory output for delivery.

20. The smell delivery system according to any preceding claim, wherein the smell delivery device comprises a flow generator configured to provide a flow of carrier gas to the odour chamber for releasing the substance, and wherein the settings for delivery of the olfactory output correspond to adjusting a flow rate of the flow generator.

21. The smell delivery system of any preceding claim, wherein:the control unit is integral with the smell delivery device or the control unit is separate from the smell delivery device, and / orthe system further comprises the at least one sensor, optionally wherein the sensor is integral with the smell delivery device and / or the control unit.31 07 2522. A method of delivering an olfactory output, comprising:receiving sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user;determining a sleep state of the user based at least on the sensor data, wherein determining the sleep state includes determining multiple sleep states from the sleep stages: REM, light sleep and deep sleep;determining settings for delivery of an olfactory output based at least on the sleep state of the user and an impact of previous settings on the sleep state, wherein determining the settings includes using a database having settings for particular sleep scents for particular sleep stages which obtain pre-determined effects on the user’s sleep; andproviding instructions to a smell delivery device corresponding to the settings.

23. A method of delivering an olfactory output, comprising:receiving sensor data from at least one sensor configured to measure a signal indicative of a parameter associated with a sleep state of a user;determining a sleep state of the user based at least on the sensor data, wherein determining the sleep state includes determining multiple sleep states from the sleep stages: REM, light sleep and deep sleep;determining settings for delivery of an olfactory output based at least on the sleep state of the user, wherein determining the settings includes using a database having settings for particular sleep scents for particular sleep stages which obtain predetermined effects on the user’s sleep;providing instructions to a smell delivery device corresponding to the settings; andreceiving subsequent sensor data from the at least one sensor, and determining an impact of the settings for delivery of the olfactory output on the sleep state based at least on the subsequent sensor data.

24. The method of claim 23, further comprising delivering, by the smell delivery device, the olfactory output based on the instructions.

25. A computer program, computer program product, or computer-readable medium comprising instructions which, when executed by a processor, cause the processor to perform the method of claim 23 or 24.

Images