Device and computer program for determining a therapeutic stimulus
The device and method enhance psychological testing by integrating biometric data with psychological responses to provide adaptive, personalized therapeutic interventions, addressing the limitations of existing methods in reliability and accessibility.
Patent Information
- Application Number
- PCT/EP2026/067927
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-06-18
- Filing Date
- 2026-06-17
- Publication Date
- 2026-11-12
AI Technical Summary
Existing psychological testing methods lack reliability and accessibility, failing to provide comprehensive assessments of mental health conditions and inadequate support measures.
A device and method that integrates a processor circuit to receive psychological test responses and biometric data, determining a therapeutic stimulus to change the user's state, with adaptive feedback loops and machine learning for personalized interventions.
Enables precise, context-sensitive evaluations and tailored therapeutic stimuli to improve mental well-being through continuous optimization and individualized support.
Smart Images

Figure EP2026067927_12112026_PF_FP_ABST
Abstract
Description
[0001] DEVICE AND METHOD FOR DETERMINING A THERAPEUTIC STIMULUS
[0002] Technical field
[0003] The present disclosure relates to a device and a method for determining a therapeutic stimulus as described in the attached claims.
[0004] background
[0005] Mental health conditions can manifest in various forms and affect numerous areas of daily life. They can impact emotional well-being, performance, and social interactions. In many cases, mental health issues can go unnoticed for extended periods or are only inadequately recognized. However, addressing mental health problems early on can help identify problematic developments and initiate appropriate support measures.
[0006] Psychological tests can provide indications of certain mental states or impairments. Such tests can be used in various contexts, such as clinical diagnostics, prevention, or counseling and therapy. The results of these tests can serve as a basis for further assessments or interventions.
[0007] Therefore, there is a need to provide reliable and accessible means of conducting psychological testing.
[0008] The present device and method, according to the independent claims, take this need into account.
[0009] Summary
[0010] Exemplary embodiments relate to a device for determining a therapeutic stimulus. The device comprises a processor circuit. The processor circuit is configured to receive a user's response data to one or more psychological test questions. The response data was acquired via a user interface. The processor circuit is further configured to acquire the user's initial biometric data. The user's initial biometric data was acquired via at least one sensor unit during the input of the response data. The processor circuit is further configured to determine the user's state based on the response data and the user's initial biometric data. The processor circuit is also configured to determine a therapeutic stimulus based on the determined state. The therapeutic stimulus is designed to change the state.The processor circuit is also configured to provide the therapeutic stimulus to at least one output device.
[0011] The device can enable a differentiated recording of the user's psychological and / or physiological characteristics in order to comprehensively determine the user's state. By linking response data with accompanying biometric information, a more context-sensitive evaluation can be performed, leading to more precise assessments. The output of an individually tailored stimulus can help to provide immediate support and thus specifically influence the user's psychological state.
[0012] In one embodiment, the processor circuit can be further configured to acquire secondary biometric data from the user. This secondary biometric data can be acquired via at least one sensor unit after the therapeutic stimulus has been applied. The processor circuit can also be configured to determine an updated state of the user based on this secondary biometric data. Acquiring and evaluating this updated state after an intervention allows for the assessment of the intervention's effectiveness and the implementation of a feedback loop.
[0013] In one embodiment, the processor circuitry can be further designed to determine a difference between the user's current state and its updated state. Determining this difference can help to measure changes in user behavior or well-being.
[0014] In one embodiment, the processor circuit can be further designed to adapt the therapeutic stimulus based on the updated state. An adaptive stimulus can increase individual effectiveness and contribute to a continuous improvement in well-being. In another embodiment, the processor circuit can be further designed to acquire the user's first and / or second biometric data via the at least one sensor unit while the response data is being entered via the user interface. Acquisition during input can provide a direct correlation between behavior and physiological states.
[0015] In one embodiment, the first and / or second biometric data of the user can describe at least one biometric characteristic of the user that was captured by the at least one sensor unit. Relevant indicators can then be specifically analyzed using these biometric characteristics.
[0016] In one embodiment, the at least one biometric feature can include at least one of the following: pitch, volume, speech rate, intonation, prosodic features, speech pause patterns, articulation, pitch, heart rate, respiratory rate, skin conductance, body temperature, pupil diameter, eye movement, blink rate, facial expression, muscle activity, neural activity patterns, or posture. The biometric features allow for a comprehensive analysis of the patient's condition at multiple physiological levels.
[0017] In one embodiment, the processor circuitry can be further designed to select a subsequent test question based on the response data and / or the initial biometric data. An adaptive questioning strategy can increase the efficiency and relevance of the testing.
[0018] In one implementation example, the subsequent test question can be selected by a machine learning model. The use of machine learning methods can enable continuous optimization of the test logic.
[0019] In one embodiment, the user's condition can be a psychological state, a biometric state, or a psychobiometric state.
[0020] In one embodiment, determining the user's state can involve applying a rule-based procedure or a machine learning model. Various algorithmic methods can be flexibly employed to meet different application requirements. In one embodiment, determining the user's state can include: determining a preliminary state based on either the response data or the biometric data, and then adjusting the preliminary state using the other of the two data sources. A two-stage process can increase the robustness and accuracy of the state determination.
[0021] In one embodiment, determining the therapeutic stimulus can involve applying a rule-based procedure or a machine learning model. The stimulus selection can be dynamically adapted to complex individual data.
[0022] In one embodiment, the output device can be designed to generate at least one of the following therapeutic stimuli: a visual stimulus, an auditory stimulus, a haptic stimulus, a thermal stimulus, an electrical stimulus, a light-based stimulus, or an olfactory stimulus. The stimuli can address different sensory levels and be used in a targeted manner.
[0023] In one embodiment, the processor circuit can further be configured to evaluate usage data of the provided therapeutic stimuli and to adapt future selection decisions of the therapeutic stimulus for users with a similar condition based on the evaluated data. A learning component can contribute to the continuous optimization of stimulus selection.
[0024] In one embodiment, the device may also include a user interface.
[0025] In one embodiment, the device can further comprise at least one sensor unit.
[0026] In one embodiment, the device can further comprise at least one output device.
[0027] In one embodiment, the at least one sensor unit can comprise at least one of the following components: a microphone, a camera, a pulse sensor, a skin conductance sensor, a thermal sensor, an accelerometer, a gyroscope, a photoplethysmograph, an electrode array for recording bioelectrical signals, an infrared sensor, an eye-tracking system, a strain sensor, a radar-based vital sign sensor, an electroencephalogram, or a capacitive proximity sensor. The selection of sensors enables a broad detection base of physiological signals.
[0028] In one embodiment, the user interface can include one or more of the following components: a touchscreen, a microphone with speech recognition, a graphical display element, a control button, or a sensory input unit. Combining different input methods can enable flexible interaction with the system.
[0029] In one embodiment, the output device can comprise at least one of the following components: a screen, a loudspeaker, a vibration actuator, a stimulation module, a heating or cooling element, a light source, or an odor generation device. The output via different modalities can be specifically adapted to the individual condition of the user.
[0030] Exemplary embodiments relate to determining a therapeutic stimulus. The method comprises obtaining a user's response data to one or more psychological test questions, wherein the response data was captured via a user interface. The method further comprises obtaining initial biometric data of the user, wherein the initial biometric data of the user was captured by at least one sensor unit during the input of the response data. The method further comprises determining a user's state based on the response data and the initial biometric data of the user. The method further comprises determining a therapeutic stimulus based on the determined state, wherein the therapeutic stimulus is designed to change the state. The method further comprises providing the therapeutic stimulus to at least one output device.
[0031] Exemplary embodiments relate to a non-volatile, machine-readable medium on which a program is stored containing program code for carrying out the above-mentioned method when the program is executed on a processor or a programmable hardware component.
[0032] Exemplary embodiments relate to a computer program with program code for carrying out the above-mentioned method, when the program is executed on a processor or a programmable hardware component. Figure description
[0033] Some examples of devices and / or methods are explained in more detail below with reference to the accompanying figures. These show:
[0034] Fig. 1 schematically shows a device 100 for determining a therapeutic stimulus; and
[0035] Fig. 2 shows a schematic of a method 200 for determining a therapeutic stimulus.
[0036] Description
[0037] Some examples are now described in more detail with reference to the accompanying figures. However, other possible examples are not limited to the features of these detailed embodiments. These may include modifications of the features, as well as equivalents and alternatives to the features. Furthermore, the terminology used herein to describe certain examples should not be considered restrictive for other possible examples.
[0038] Identical or similar reference symbols throughout the description of the figures refer to identical or similar elements or features, which may be implemented in an identical or modified form, while providing the same or a similar function. Furthermore, the thickness of lines, layers, and / or areas in the figures may be exaggerated for clarity.
[0039] When two elements A and B are combined using "or," this is to be understood as revealing all possible combinations, i.e., only A, only B, and A and B, unless explicitly defined otherwise in a specific case. As an alternative formulation for the same combinations, "at least one of A and B" or "A and / or B" can be used. This applies equivalently to combinations of more than two elements.
[0040] When a singular form, e.g., "ein, eine" and "der, die, das," is used, and the use of only a single element is neither explicitly nor implicitly defined as mandatory, further examples may also use multiple elements to implement the same function. If a function is subsequently described as being implemented using multiple elements, further examples may implement the same function using a single element or a single processing entity.It is further understood that the terms "include", "comprehensive", "exhibit" and / or "exhibit" when used describe the presence of the specified features, integers, steps, operations, processes, elements, components and / or a group thereof, but do not exclude the presence or addition of one or more other features, integers, steps, operations, processes, elements, components and / or a group thereof.
[0041] Fig. 1 schematically shows the device 100 for determining a therapeutic stimulus. The device comprises a processor circuit 110. For example, the processor circuit 110 can be formed by a single dedicated processor, a single shared processor, or a plurality of individual processors, some or all of which can be used jointly, a microcontroller, an application-specific integrated circuit (ASIC), an integrated circuit (IC), a system-on-a-chip (SoC), a programmable logic element, or a field-programmable gate array (FPGA) with a microprocessor on which software for screen control runs according to the principles described herein. The processor can, for example, be a computer processor (CPU).The processor circuit 110 may be a CPU (Central Processing Unit), a graphics processing unit (GPU), a neuromorphic processor, and / or a tensor processor. The processor circuit 110 may also be connected to a memory, such as a read-only memory (ROM) for storing software, a random access memory (RAM), and / or a non-volatile memory. For example, the device 100 may include or be coupled to a memory configured to store instructions which, when executed by the processor circuit 110, cause the processor circuit 110 to perform the steps and procedures described herein for determining a therapeutic stimulus.
[0042] The processor circuit 110 is configured to receive a user's response data to one or more psychological test questions. The response data is captured via a user interface 120. In some examples, the user interface 120 is part of the device 100. In other examples, the user interface 120 is an external unit outside of the device 100. For example, the processor circuit 110 can receive the response data via a data interface of the device 100 that is directly coupled to the user interface, such as through an internal bus system within the device. In another example, the processor circuit 110 can retrieve the response data from a local or external memory of the device 100, where the response data was temporarily stored after capture.In other examples, the processor circuit 110 can receive the response data via a network interface if the user interface is implemented as part of an external end device, such as a web-based or mobile interface connected to the device via the internet or a local network. The connection can be wired or wireless via Wi-Fi or Bluetooth. For example, a user answers test questions on an app on a smart device. The response data is stored locally and then transmitted to a central processor circuit on the smart device for further processing.
[0043] For example, a psychological test question is a structured, methodically designed question that can be used to assess certain characteristics, states, or behaviors of a person that may be relevant in psychological diagnosis or analysis. In some examples, a psychological test question may be designed to provide an assessment of the user's affective, cognitive, or behavioral aspects. Psychological test questions can relate to various topics, including mood, motivation, stress, attention, self-image, or social interaction. The questions can be open-ended or closed-ended, with closed-ended questions often including response options in the form of scales (e.g., "true - false"), multiple choices, or dichotomous options ("yes / no").In other examples, several psychological test questions can be related to each other in terms of content or function, for instance, by being grouped thematically or building sequentially upon one another to allow for a more comprehensive assessment. The questions can be presented via text, audio, or graphics, provided the user interface supports this. An example might be a series of ten questions in which the user indicates, on a scale of 1 to 5, how often they have felt sad, tense, or lacking in energy over the past seven days. The test questions are designed so that, together, they can provide an indication of depressive symptoms.
[0044] For example, response data can be structured digital representations of user answers to psychological test questions, which can be stored and processed in such a way as to enable automatic evaluation by a processor circuit 110. In some examples, response data can contain numerical values, categorical codes, or free text entries, depending on the type of underlying psychological test question. For closed questions, the response data can, for example, include standardized scale values, such as integers in the range of 1 to 5 for Likert scales, or binary values such as 0 and 1 for yes / no answers. For open questions, the response data structure can also contain strings of text characters, which can then be further processed, for example, using text recognition, keyword analysis, sentiment analysis, or large-language models (LLM).From a technical perspective, response data can be in a standardized data format such as JSON, XML, or in the form of a relational data structure (e.g., as a database entry with fields such as question ID, answer value, and timestamp). The response data can also include additional metadata, such as the time of the response, the duration of the response, or the input method used. An example of response data could be a JSON object consisting of an entry like {"questionjd": "Q12", "answer": 4, "timestamp": "2025-05-27T10:35:00Z"}, representing a response to a four-point scale question entered by the user via a touchscreen.
[0045] In some examples, the user interface may include one or more of the following components: a touchscreen, a microphone with speech recognition, a graphical display element, a control button, a sensory input device, or a speaker. For example, answers to test questions via user interface 120 may be entered through direct interaction with one or more of these components. In some examples, a touchscreen may be used, allowing users to enter data by tapping answer fields, sliders, or selection menus. In other examples, input may be voice-activated, using a microphone with speech recognition to capture spoken answers and convert them into machine-readable data. An answer may also be selected or confirmed via a control button or a sensory input device, such as by touch, proximity, or gestures.In these examples, the loudspeaker serves only for acoustic output and not for input, but is functionally integrated as part of the user interface.
[0046] In some examples, the processor circuit 110 can also be configured to output one or more psychological test questions via the user interface. For example, the processor circuit 110 can transmit the psychological test questions to the user interface 120 via an internal or external interface of the device 100. In some examples, the output of the psychological test questions can be provided via graphical components of the user interface 120, such as text fields, interactive selection menus, sliders, or multiple-choice fields on a touchscreen or graphical display. In some examples, the user interface 120 can provide acoustic output, for example, via a loudspeaker or by means of a speech-based system with speech output.The output can be displayed locally via an integrated screen or transmitted via a network connection to an external device, such as a smartphone, tablet, or internet-enabled terminal. The psychological test questions can be displayed individually or in groups, statically or dynamically, sequentially or adaptively.
[0047] The processor circuit 110 is configured to obtain initial biometric data from the user. This initial biometric data was acquired via at least one sensor unit 130 during the input of the response data. In one example, the at least one sensor unit 130 is part of the device 100. In another example, the at least one sensor unit 130 is an external unit located outside of the device 100.
[0048] In some examples, the processor circuit 110 can obtain the user's first biometric data and / or second biometric data (see below) via a data interface of the device 100 that is directly coupled to the at least one sensor unit 130, for example, via an internal bus system within the device 100. In another example, the processor circuit 110 can retrieve the biometric data from a local buffer of the device 100 in which the sensor data was stored after acquisition. In further examples, the processor circuit 110 can obtain the biometric data via an external network interface if the at least one sensor unit is part of an external unit, for example, a wearable or an external sensor system, that is connected to the device via the internet, a local network, or a wireless connection such as Bluetooth or Wi-Fi.
[0049] The initial biometric data of the user is captured simultaneously with the input of response data via the user interface 120. The at least one sensor unit 130 can be configured to record biometric measurements during user interaction with the test questions. The processor circuit 110 can process these time-coupled data streams together or store them in a common data frame. Synchronization can be ensured by clocked sampling, shared timestamps, or a central control unit that coordinates both the user interface and the at least one sensor unit. Simultaneous acquisition ensures that the biometric data can be directly linked to specific response events.This allows for a more precise and context-sensitive assessment of the user's condition, as physiological reactions (such as an increase in heart rate or changes in speech behavior) can be clearly assigned to individual answers.
[0050] In some examples, the processor circuit 110 is further designed to acquire the user's first and / or second biometric data via at least one sensor unit 130 while the response data is being entered via the user interface. The processor circuit 110 can, for example, control the at least one sensor unit 130. This includes, for example, actively switching it on or off, specifying sampling intervals, or synchronizing the start time of data acquisition with the user interaction. Technically, this can be achieved via a control signal sent from the processor circuit 110 to the sensor unit 130 via an internal bus connection or a specified control line. Alternatively, the processor circuit 110 can define a trigger condition, such as activating data acquisition when a test question is opened or when the user enters an answer.In other examples, the processor circuit 110 can also configure cyclic or event-based acquisition modes by specifying sampling rates, recording duration or specific filter parameters to the sensor unit.
[0051] For example, the user's (first and / or second) biometric data describes at least one biometric characteristic of the user, which was captured by at least one sensor unit 130. For example, a biometric characteristic can be a measurable property of the user or of the user's human body that can be captured by one or more sensor units. In some examples, a biometric characteristic can be a measurable physiological parameter of the user. The biometric characteristic can be captured in the form of a numerical value, a categorized characteristic, or a time series, and can be enriched with metadata such as timestamps, capture frequency, or sensor identifiers. The biometric data can originate from several different sensor units. The biometric data can be in the form of time series, numerical values, or coded characteristic vectors.For example, the biometric data can be supplemented with timestamps or metadata to classify the recording conditions. The data can be temporarily stored locally or transferred directly to the processor circuit 110 in a structured form (e.g., as JSON objects, tabular measurement data, or binary sensor packets).
[0052] For example, at least one biometric characteristic includes at least one of the following: pitch, volume, speech rate, intonation, prosodic features, speech patterns, fluency, pause patterns, articulation, pitch, heart rate, respiratory rate, skin conductance, body temperature, pupil diameter, eye movement, blink rate, facial expression, muscle activity, neural activity patterns, or posture. Pitch describes the fundamental frequency of a vocal signal and can be measured in hertz (Hz). It can be captured with a microphone and used to assess emotional arousal or vocal tension. Volume refers to the amplitude of an acoustic signal, typically expressed in decibels (dB). Increased or decreased volume can indicate excitement, anxiety, or relaxation.Speech rate describes the number of spoken units, such as syllables or words, per unit of time and can provide indications of cognitive load, stress, or activation level. Speech melody captures the pitch variation throughout a spoken utterance. It serves to identify intonation, sentence type, or emotional tone of speech.
[0053] Prosodic features encompass overarching temporal and tonal structures of speech, such as rhythm, stress, and intonation patterns, which allow inferences about speech manner and emotional nuances. Speech patterns describe characteristic combinations and sequences of linguistic features within an utterance. These include, for example, word choice, sentence structure, pause distribution, and prosodic characteristics. Speech patterns can be captured with a microphone and subsequently analyzed by feature extraction from the acoustic and / or transcribed speech signal. They allow inferences about the user's speech manner, cognitive strategies, or emotional tendencies and can, for example, provide clues about uncertainty, haste, or thoughtfulness. Speech fluency describes the continuity, fluency, and rhythmic structure of spoken language over a period of time.It encompasses parameters such as pause length, speaking rate, volume modulation, and transitions between words or sentence parts. Speech fluency can be captured through continuous speech recording using a microphone and evaluated through time-series analysis of the speech signals. Anomalies in speech fluency, such as frequent interruptions, abrupt transitions, or monotonous sequences, can indicate cognitive strain, emotional states, or speech-motor abnormalities. The speech pause pattern describes the distribution, duration, and frequency of pauses within or between utterances. Conspicuous pause structures can correlate with uncertainty, thoughtfulness, or tension. Articulation describes the physically perceptible form of sound production during speech, particularly with regard to clarity, precision, or distinctness of pronunciation.Unclear or unusually emphasized articulation can indicate emotional or cognitive abnormalities. Pitch refers to the typical pitch range of the voice, that is, the frequency range in which a speaker prefers to speak. A change in pitch can indicate a change in affective or physiological state.
[0054] Heart rate describes the number of heartbeats per minute (bpm) and can be measured using a pulse sensor, an ECG, or a photoplethysmograph. It serves as an established indicator of activation and stress. Respiratory rate refers to the number of breaths per minute. It can be measured using pressure sensors, chest straps, infrared sensors, or radar-based vital sign technology and is a key vital sign. Skin conductance describes the electrical conductivity of the skin surface, typically measured in microsiemens (pS). It can be influenced by electrodermal activity and can serve as a measure of sympathetic arousal. Body temperature refers to the temperature of the skin or core body at a specific measurement point. It can be determined using infrared sensors or thermocouples and can be used as an indicator of autonomic processes.
[0055] Pupil diameter describes the size of the pupil and can be measured in millimeters. It changes depending on lighting conditions, cognitive effort, or emotional arousal. It can be measured using infrared-based eye-tracking systems. Eye movement encompasses the movement of gaze direction in the horizontal and vertical planes. It can be recorded by video-based eye-tracking systems and provides information about visual attention, orientation, and stimulus processing. The blink rate describes the number of blinks per minute. It can be measured with infrared cameras or optical sensors and is a sensitive indicator of fatigue, concentration, or emotional states.
[0056] Facial expression describes the configuration of the facial muscles, particularly those around the eyes, mouth, and forehead. It can be captured using image processing, allowing for the classification of defined expression patterns (e.g., smiling, frowning). Muscle activity describes the electrical or mechanical activity of individual muscle groups and can be recorded using electromyography (EMG) or strain sensors. It serves to assess tension, movement, or physical reaction.
[0057] Neural activity patterns describe the temporal and spectral structure of electrical activity in the brain, as recorded, for example, by an electroencephalogram (EEG). They encompass characteristic features such as amplitudes, frequency bands (e.g., alpha, beta, theta, or delta waves), and their spatial distribution across different brain regions. Such patterns can provide clues to cognitive states such as attention, alertness, fatigue, or emotional arousal and serve to assess the user's neurophysiological condition. Posture describes the spatial orientation and position of the body or individual body parts at rest or in motion. It can be recorded using inertial measurement units, accelerometers, or image-based methods and provides information about postural patterns, muscle tension, or readiness to move.
[0058] An example of biometric data could be a dataset containing heart rate (in bpm), skin conductance (in pS) and voice pitch (in Hz) at a fixed interval - for example { "time": "10:35:42", "hr": 82, "eda": 1.7, "fO": 192}.
[0059] In some examples, the at least one sensor unit 130 comprises at least one of the following components: a microphone, a camera, a pulse sensor, a skin conductance sensor, a thermal sensor, an accelerometer, a gyroscope, a photoplethysmograph, an electrode array for recording bioelectrical signals, an infrared sensor, an eye-tracking system, a strain sensor, a radar-based vital signs sensor, an electroencephalogram, or a capacitive proximity sensor. For example, the microphone may be configured to record acoustic characteristics of the user, in particular measuring pitch, volume, speaking rate, intonation, or articulation. The camera may be configured to detect visual biometric features, in particular recognizing facial expression, posture, eye movement, or pupil diameter.The pulse sensor can be used to measure heart rate. The skin conductance sensor can detect the electrical conductivity of the skin and thus provide indications of sympathetic activation. The thermosensor can determine body temperature at a measurement point, particularly on the skin surface. The accelerometer can record user movement data, which can provide insights into posture or physical activity. The gyroscope can detect rotational movements and be used to analyze body or head posture. The photoplethysmograph can optically measure heart rate and its variation. The electrode array for recording bioelectrical signals can be used to detect muscle activity, particularly through electromyography, or to record electrical cardiac activity, particularly through electrocardiography.
[0060] The infrared sensor can be used to determine body temperature or measure pupil diameter, especially in low-light conditions. The eye-tracking system can continuously monitor eye movements, pupil diameter, and blink rate. The strain sensor can detect mechanical changes caused, for example, by muscle tension or breathing. The radar-based vital signs sensor can determine the user's respiratory and heart rates without physical contact. The capacitive proximity sensor can detect subtle changes in the user's position and provide indications of changes in posture or approach.
[0061] The electroencephalogram (EEG) uses an array of electrodes placed on the scalp to record changes in electrical voltage generated by neuronal activity in the brain's cortex. These signals are interpreted as patterns of neuronal activity and provide information about the user's mental state. The EEG can detect characteristic frequency patterns, such as alpha, beta, theta, or delta waves, each of which correlates with specific cognitive or emotional states. For example, an increase in alpha activity may indicate relaxation or inactivity, while increased beta activity may be associated with cognitive load or heightened attention. The EEG is therefore suitable for the objective recording of neurophysiological states and can be used as a biometric data source in real-time applications.
[0062] As described above, the processor circuit 110 can be configured to capture the user's initial biometric data while they answer one or more psychological test questions. This capture can occur synchronously with the input of the response data via the user interface, allowing each response to be linked to a corresponding segment of the initial biometric data. This temporal correlation makes it possible to determine the user's biometric state for the period during which a specific cognitive or emotional response to a particular test question occurs. This creates a coherent database upon which the determination of the user's state can be based, both conceptually and physiologically.
[0063] Simultaneous data acquisition can be achieved through coordinated control of the user interface 120 and the at least one sensor unit 130. In some examples, the processor circuit 11 can activate both components simultaneously or synchronize them via a common time grid. The assignment of the response data to the corresponding biometric data can be accomplished either through shared timestamps or through structured acquisition within a contiguous data frame. The evaluation of these synchronized data pairs can enable a particularly precise assessment of the patient's condition, as both conscious self-reporting and unconscious physiological reactions can be taken into account.
[0064] In other examples, the user interface 120 can at least partially overlap functionally or structurally with the at least one sensor unit 130. For example, a microphone can be both part of the user interface 120—for instance, for voice input or voice control—and also serve as a sensor unit 130 for capturing acoustic biometric characteristics such as pitch, volume, or speech rate. Similarly, a touchscreen can be used both for inputting response data and for deriving indirect biometric information, such as through the recognition of touch patterns, pressure response, or reaction times.
[0065] For example, the user answers psychological test questions via a built-in microphone. The acoustic signal input is semantically evaluated to answer the psychological test questions, and furthermore, the acoustic signal input is analyzed for acoustic features such as pitch and intonation. The resulting synchronized response and biometric data are then passed to the processor circuit 110 for further analysis to determine the user's state.
[0066] In another example, the initial biometric data can also be collected before and / or after the answers have been entered.
[0067] The processor circuit 110 is designed to determine a user's state based on response data and the user's initial biometric data. For example, the user's state can be a quantifiable, measurable, and technical parameter Z. finalThis is based on the response data and / or the user's initial biometric data. In other words, the user's state can be, for example, a quantifiable, measurable, and technical parameter that represents the user's condition based on physiological and / or psychological characteristics.
[0068] For example, the user's state can be a psychological state, a physiological (i.e., biometric) state, or a psychophysiological state. For instance, the user's psychological state can encompass a description of the user's current mental or emotional state. This can include states such as concentration, exhaustion, stress, relaxation, anxiety, anger, agitation, boredom, or attention. For example, a physiological state can be a user's state described by measurable bodily parameters. For instance, a psychophysiological state can be a specific measurable combined state of the user, derived jointly from psychological and physiological measurable characteristics.
[0069] The state can be described, for example, by one or more numerical values, which can be on a discrete or continuous scale. In some examples, the state might represent a normalized value between two defined extremes, such as to represent an activation, load, or stress level. Alternatively, the state can be described by a discrete scale, such as integer ratings like 1 to 10, or by ordinally interpretable categories. Furthermore, the user's state can be described by classification into one of several predefined classes, where the classification can be based on rules, decision trees, or machine learning models. In some examples, the classification can be based on probability values, with each potential class assigned a probability measure that expresses the confidence in the class assignment.Classification can be deterministic or probabilistic and based on a weighted evaluation of the input data. The state can thus be represented as either a scalar, continuously evaluated quantity or as a probabilistically assigned class membership. In some examples, the determination of the state can be adaptive and take past state values or reference data into account. The reference data used for this purpose can be user-specific, for example, from previous sessions of the same user, or population-based, by including comparative data from multiple users to enable a context- or group-specific classification of the current state.
[0070] In some examples, determining the user's state may involve applying a rule-based procedure or a machine learning model.
[0071] In some examples, determining the state of user Z can befinal furthermore, determining a preliminary state Z anwort or Z biometrisch based on either the response data or the biometric data. In some examples, determining the user's state may involve Zf inat an adjustment of the provisionally determined state using the other of the two data sources Z anwort or include biometrics. In some examples, determining the user's state can be Z^inai. e ' n combining the two preliminary determined states of the two data sources to answer or...
[0072]
[0073] include.
[0074] In some examples, the user's preliminary state can initially be determined based on the response data. The responses can be structured input, such as multiple-choice options, numerical rating scales (e.g., from 0 to 10), free-text answers, or speech-based input converted into structured text representations. A rule-based procedure can be used to determine the preliminary state score from the response data. In such a procedure, each answer to a specific question can be assigned to a defined rating scale. This assignment can be done using predefined rules, thresholds, or weighted decision logic. For example, in response to the question "How stressed do you currently feel?", the following options could be coded: "very stressed" with a value of 10, "moderately stressed" with a value of 5, and "not at all stressed" with a value of 0.For multiple questions, the assigned individual values can be weighted and calculated into a sum or average value. In a more advanced example, the weighting of the questions themselves can be adaptive, for instance, through question group logic or predefined weighting schemes depending on the question type or context. The result is a numerical value that represents an initial assessment of the psychological state. Furthermore, the assessment of a state can be based on standardized values from norm samples, so-called T-scores or z-scores, which are entered into the system based on a representative sample. These values can change over time based on machine learning. In another example, the preliminary state can be determined based on the response data using a machine learning model.A previously trained model, such as a decision tree, a logistic regression model, a random forest approach, a neural network, or a large-language model (LLM), can receive the input responses as a feature vector. The input can be structured through appropriate feature extraction and preprocessing—for example, converting categorical responses into one-hot vectors, extracting keywords from free text, or creating semantic vectors from spoken responses. The trained model can then calculate a continuous output value (e.g., stress level on a scale of 0–10) or a probability distribution across defined classes (e.g., "emotionally stable," "irritated," "overwhelmed"). This assignment can be performed in real time or as a batch-based analysis.
[0075] In some examples, a hybrid approach can also be used, combining rule-based pre-filtering with model-based evaluation. For instance, implausible or incomplete answers can be identified and excluded or corrected using rule-based logic before the remaining answer data is fed into a model.
[0076] In some examples, the user's preliminary state can be determined based on the initial biometric data Z. biometrischThe initial biometric data can comprise one or more measurement series, synchronously acquired via at least one sensor unit during the input of response data. The acquired signals can represent various physiological parameters of the user, such as pitch, volume, heart rate, skin conductance, or pupil diameter. The biometric data can be in the form of continuous time series, discrete individual measurements, or aggregated feature vectors.
[0077] Biometric data is available directly as numerical values and can therefore be used directly as a preliminary assessment of the user's state. Alternatively, preliminary state determination based on biometric data can be rule-based. Specific value ranges or patterns of biometric parameters can be assigned to predefined categories. For example, a resting heart rate above 100 bpm can be classified as "elevated," and elevated skin conductance can be interpreted as "activated." Rule combinations (e.g., "high heart rate + accelerated respiratory rate") can be used to detect states such as "physiological stress." This logic can be implemented using decision trees, combined threshold systems, or lookup tables. The result is either a discrete state (e.g., category: "tense") or a numerical score (e.g., arousal level = 7.8).
[0078] In another example, state determination can be performed using a machine learning model. Here, the biometric data is preprocessed—for example, by smoothing, segmentation, deriving time features (mean, standard deviation, peak-to-peak distance), or spectral analysis. The resulting feature vectors can then be fed to a trained model. Examples of such models include random forest classifiers, support vector machines, neural networks, or LLMs. The model can calculate either a continuous state value (e.g., physiological stress = 5.4) or a probability distribution across state classes (e.g., "physiological stress relaxed": 30%, "physiological stress tense": 65%, "physiological stress overexertion": 5%).
[0079] In some examples, an adaptive model component can also be used, where the system learns from historical user data which biometric feature combinations are associated with specific states. Furthermore, personalized thresholds or profiles can be applied, for example, through an initial calibration procedure at user startup. In some examples, a hybrid approach can also be used.
[0080] In some examples, the user's state Z finai by a combination of the two preliminary states Z anwort or Z biometrisch to be determined, whereby a first provisional state Z anwort from the response data and a second preliminary state z biometric aus The first biometric data was calculated as described above. The final state Z finalThis can be determined by a procedure in which the two preliminary states are processed as input values and combined to form a common result Z^inai. Different methods can be used to compensate for inconsistencies between the two sources or to integrate complementary information.
[0081] In one example, the final state Zf inai through a weighted fusion of both preliminary states Z anwort and Z biometrisch to be determined. In doing so, the state Z calculated from the response data is determined. anwort and the state Z calculated from the biometric data biometrisch combined with weighting factors a and ß:
[0082] final = a ' ^answer + ß ' ^biometric
[0083] The weighting factors can be determined statically or dynamically, for example by a lem-based metamodel that derives optimal weights from historical data.
[0084] In further examples, a trained model (e.g., a neural network or a decision tree) can be used that receives both states together as a feature vector and calculates a final class assignment or an aggregated numerical state from it. Such models can learn from training data which combinations of psychological assessment and physiological measurement typically belong together, and thus also take into account complex interactions between the two dimensions.
[0085] In some implementations, the final state can be further enhanced with a confidence level or an uncertainty measure that quantifies the consistency or divergence between response and biometric data. This can be achieved, for example, through probability weighting or a confidence score, which is used as an additional parameter to inform system behavior, such as for controlling subsequent stimuli.
[0086] In some examples, the preliminary state Z biometischThe user's state is determined based on initial biometric data. This initial biometric data can represent sensor-acquired signals that directly reflect the user's physical or physiological parameters, such as heart rate, skin conductance, or voice pitch. These raw signals can undergo preprocessing, such as filtering, segmentation, and feature extraction, to generate a feature vector. Based on this feature vector, a rule-based system or a trained model (e.g., a neural network or decision tree) can determine a preliminary numerical or categorical state, such as a stress level on a scale of 0-10 or a classification like "physiologically calm" or "physiologically overstimulated."
[0087] In a next step, this preliminary state Z biometrischThe response data can be validated, corrected, and / or adjusted through an evaluation. Technically, validation can be achieved by checking the consistency between the state values. For example, a biometrically derived stress score of 9 on a scale of 0-10 might be considered inconsistent if the explicit response data indicates a very low subjective stress score (e.g., 1 out of 10). Such a divergence can be detected using predefined thresholds or distance metrics (e.g., absolute difference > 5). A correction rule can then be applied, such as averaging the two states, weighted recalibration, or selecting a state with a higher confidence level (e.g., a high plausibility score in the response dataset).
[0088] For example, determining the user's state involves determining the preliminary state based on biometric data and further adjusting this preliminary state using response data. In other words, the preliminary state Z determined based on biometric data can be... biometrisch The data undergoes quality assurance and refinement by using additional user response data for validation and correction. This involves, for example, a technical "signal correction" or "compensation logic" to correct potential misinterpretations, such as those caused by nonspecific physiological patterns. The correction can be rule-based, for instance, by applying thresholds or confidence intervals, where deviations between Z and Z are considered. biometrisch and a state Z derived from the response data antwortThese errors are identified and addressed through defined adjustment operations (e.g., weighted averaging, heuristic override logic, decision tables). In more advanced examples, a correction model can be used that combines both sources as a feature vector and derives a corrected version, Z final. This model can be a trained machine learning model (e.g., a metaclassifier, regression model, or transformer) that learns from historical case scenarios how response data compensate for typical misinterpretations of biometric patterns. Thus, validation is used not only for logical assurance but also for technical quality improvement—especially in scenarios with ambiguities, such as a high heart rate during physical exertion versus psychological stress.The response data thus act as semantic anchors for refining the state model and increase the reliability and individualizability of the final parameter Zf. inc .
[0089] In other examples, correction can be achieved using a meta-model that has learned from historical data in which situations physiological parameters are systematically overestimated or misinterpreted—for example, during physical activity, where a high heart rate does not necessarily indicate stress. In such cases, a high stress level can be recognized from physiological signals by an opposing response evaluation, and the final state can be adjusted downwards or classified as "neutral."
[0090] Validation can also be performed probabilistically. For example, the probability of a biometrically determined state can be modified using a correction function derived from the response data. An initial probability value of 0.75 for the state "highly stressed" could be reduced to 0.25 by classifying the response as "not stressed" with a weight of -0.5.
[0091] In other examples, a threshold system with adaptive factors can be used, where different tolerances or correction profiles apply to specific user groups or contexts. These can be dynamically adjusted, for example through continuous user monitoring or through explicit calibration prior to use.
[0092] In some examples, the preliminary state Z antwortThe user's response data is determined based on the response data. This data can contain structured information, such as multiple-choice selections, scale ratings, or semantically analyzed verbal responses. This information can be translated into a psychological state using rule-based procedures or a trained model, for example, as a scale value for subjectively perceived stress, concentration ability, or emotional stability. This value represents an initial assessment based on the user's self-reported information.
[0093] In a subsequent step, the preliminary state Z can be antwortThe assessment can be verified and adjusted as needed by evaluating the simultaneously acquired biometric data. The biometric data can, for example, comprise one or more measurement series representing indicators of physiological activation or relaxation—such as heart rate, skin conductance, or pitch. These signals are pre-processed (e.g., by smoothing, averaging, or frequency analysis) and converted into feature vectors that can be compared with typical state profiles or evaluated by a model. If the state indicators derived from the biometric data deviate significantly from the preliminary state, the final state can be adjusted using a correction function. For instance, a subjectively reported high stress level can be put into perspective if the biometric signals indicate a physiological state of rest. This correction can be achieved through predefined rules, adaptive weighting, or machine learning models.In one example, a difference analysis between expected and measured heart rate can lead to a deduction from the preliminary stress score. Alternatively, in cases of significantly differing data, an uncertainty marker or a confidence interval can be appended to the final result.
[0094] In further examples, the final state Z finai The results are calculated using an integrative model that utilizes both response data and physiological data as input. Response data plays a prioritized role, while biometric data validates, weights, or dynamically adjusts the assessment. This ensures that subjective misjudgments are identified and corrected, or consciously considered, for example, in cases of suppressed emotional states that are not reflected in self-assessment but are evident in body language or physiology.
[0095] The processor circuit 110 is configured to determine a therapeutic stimulus based on the specified state. The therapeutic stimulus is designed to change the state. For example, the therapeutic stimulus can be a technically generated and specifically applicable stimulus that actively influences a quantified state Z. final The therapeutic stimulus is used by the user. It can be designed to bring about a targeted change in a specific condition of the user. final in a desired direction Z Richung and / or towards a desired target value Z ziei to change.
[0096] In some examples, the therapeutic stimulus can be based on the specific condition Zf. inal to be determined. The processor circuit 110 can contain a target model that defines a target state Z. ziei describes which depends on the detected state Z finaland is given within a defined context. The target state Z ziet can be explicitly described as a numerical target value or implicitly by a target direction AZ.
[0097] The change in condition brought about by the therapeutic stimulus can technically consist of the effect triggered by the stimulus causing a change in the measured state in the direction of the target. For example, a detected condition Zf inal = 7.8 on a stress scale with a target value Z ziel = 3.0 are compared. The processor circuit then determines, based on the difference AZ = Z ziei ~ Zfinai, a suitable stimulus from a stimulus set whose effect is correlated with a reduction in state. In some examples, the therapeutic stimulus is defined by a function W-. Zf inai-» AZ is modeled, which is defined by empirical data, models, or rule sets. Each stimulus can be assigned an expected state change AZ, which is taken into account during selection. The selection can be made using a decision tree, a learning-based model, or heuristic rules.
[0098] In one example, determining the therapeutic stimulus might involve applying a rule-based procedure or a machine learning model. The goal of this determination is to generate a stimulus that induces the specific state Z. finalThe user's state is altered towards a desired target value or state range. The therapeutic stimulus acts, for example, as a therapeutic intervention to improve the user's assessed state. In some examples, the rule-based procedure can be based on a set of predefined decision rules. For example, each rule can describe an association between a specific state value or class and a specific therapeutic stimulus. In some examples, the same therapeutic stimulus can be used to improve different states. The rules can be based on empirically validated relationships, therapeutic guidelines, or experience-based heuristics. For example, a rule might stipulate that if a stress level is classified as elevated (e.g., Zf), the therapeutic stimulus should be applied. inai> 7) An auditory stimulus in the form of relaxing music or nature sounds with a specific volume, frequency bandwidth, and duration is activated. In another example, a visually stimulating stimulus, such as high-contrast light pulses or motivational text displays, can be provided when the user is in a "low activation" state. The rules can be deterministic or dynamically adapted, for example, by including additional parameters such as time of day, previous stimulus effects, or user history. The technical implementation can be achieved through rule-based control logic, a decision tree structure, or a dynamically expandable rule set. User feedback on the stimulus can also be used to evaluate its success and incorporated into future decision trees.In one example, a user in a state of stress, with an elevated heart rate, who responds to calming music with a lower heart rate, might be exposed to calming music earlier in future states. Conversely, a user who, in the same initial state, responds to calming music with a still-elevated heart rate, is more likely to be exposed to other stimulus classes in future states (e.g., haptic stimulus, visual stimulus). In other examples, determining the therapeutic stimulus can be based on a machine learning model. In this case, the user's specific state is... zThe finai, along with optional additional information—such as the current usage situation, previous states, or user profile parameters—is fed as an input vector to a trained model. The model can then predict the therapeutic stimulus as a continuous vector of control parameters (e.g., intensity, duration, stimulus modality) or as a class assignment to a stimulus type. In some examples, the model might be a decision tree, a random forest model, a support vector machine, a neural network, or a deep learning model such as a recurrent network, an LLM, or a transformer-based model.
[0099] The model can be based on training data where the successfully applied stimulus is documented for different combinations of state and context. Through the training phase, the model learns which stimuli lead to a desired state change in specific situations. The model can be applied in real time to select an appropriate stimulus even in previously unknown constellations. In some examples, the learning process can be continuous, allowing the model to be constantly adapted during use through user feedback or state tracking.
[0100] In other examples, a hybrid approach can be used, in which a rule-based system first makes a preselection of possible stimuli, e.g., by excluding contraindicated stimuli or by prioritizing proven measures, and then a learning-based model makes the final stimulus decision. Alternatively, the rule-based system can be used for simple, frequent states, while complex or multidimensional state profiles are handled by the machine learning model.
[0101] In some examples, the therapeutic stimulus may be visual, such as light signals, colors, image sequences, or graphic elements displayed on a screen or projection-based device. In other examples, the therapeutic stimulus may be auditory, such as music, nature sounds, speech feedback, or frequency patterns played through loudspeakers or headphones. Further examples include haptic stimulation, such as vibration or pressure pulses; thermal stimulation, such as local heating or cooling of a surface; or electrical stimulation, such as low-frequency electrical signals used to stimulate peripheral nerves. In other implementations, the stimulus may be light-based, such as controlled pulsed light sequences (e.g.,...Blue light for activation or reddish light for relaxation), or an olfactory stimulus, for example through the controlled release of aromas via a fragrance module.
[0102] These stimuli can be used individually or in combination, and their intensity, duration, frequency, modality and sequence can be adapted so that they can be used in a targeted manner to influence the measured or estimated state.
[0103] In one example, the processor circuit 110 can search a stimulus database containing stimuli with known effect profiles, such as "low music frequency lowers heart rate by 10 bpm" or "positive speech content reduces stress rating by 2 points." In another example, a trained model can generate a suitable stimulus profile tailored to individual response histories.
[0104] The processor circuit 110 is configured to provide the therapeutic stimulus to at least one output device 140. For example, the output device 140 can be part of the device 100. In this case, the output device 140 is functionally coupled to the processor circuit 110, for example, via an internal connection, such as a bus system. The provision of the therapeutic stimulus by the processor circuit 110 to the output device 140 can be carried out directly via one or more integrated control lines within the device 100. The control can be software-based, implemented using dedicated signal lines, or via a central control logic. In some other examples, the output device 140 can be an external unit outside of the device 100.In this case, the processor circuit 110 can transmit the therapeutic stimulus to the external output device 140 via a data interface. The connection can be wired or wireless, for example, via USB, Bluetooth, WLAN, or a local network protocol. In one example, the output device 140 could be an externally connected speaker, a vibration module, a portable light module, or a mobile device such as a smartphone or tablet, which transmits the stimulus—for example, in the form of auditory, visual, or haptic signals—to the user. In other examples, the output device 140 could be implemented as part of a distributed system in which the processor circuit 110 addresses and coordinates multiple stimulus transmitters via a network.In one example, the output device is designed to generate at least one of the following therapeutic stimuli: a visual stimulus, an auditory stimulus, a haptic stimulus, a thermal stimulus, an electrical stimulus, a light-based stimulus, or an olfactory stimulus.
[0105] In some examples, output device 140 may include at least one of the following components: a screen, a loudspeaker, a vibration actuator, a stimulation module, a heating or cooling element, a light source, or an odor-generating device. The screen can be used as a visual output unit and is designed to generate a visual stimulus. For example, the screen may display calming images, color gradients, or symbolic cues to elicit a targeted emotional response or to redirect cognitive attention. The loudspeaker is designed to output an auditory stimulus. Acoustic signals, such as soothing sounds, music, voice prompts, or guided breathing instructions, may be played through the loudspeaker and may have a relaxing or energizing effect on the user.
[0106] The vibration actuator is designed to provide a haptic stimulus. This can be achieved through targeted vibrations at specific points on the body (e.g., the wrist, in the seating area, or in a mobile unit) to provide tactile feedback or generate rhythmic-stimulating stimuli. The stimulation module can be designed, for example, to generate an electrical stimulus. Here, small, controlled electrical impulses can be applied via skin contact surfaces, for instance, to enhance sensory feedback or to stimulate specific nerve points. The heating or cooling element is designed to generate a thermal stimulus. Such an element can selectively generate heat or cold to provide thermosomatosensory feedback, for example, to create comfort or wake-up effects.
[0107] The light source can be used to generate a light-based stimulus. This includes, for example, LEDs that illuminate in specific colors or pulse patterns to promote visual activation, calming, or focus. The odor generation device is designed to generate an olfactory stimulus. A specific fragrance (e.g., lavender for relaxation or citrus for activation) can be released to exert a psychophysiological influence on the user's state via the sense of smell. These components can be used individually or in combination to tailor the therapeutic stimulus to the user's measured condition. The selection and parameterization of the stimuli can be dynamically controlled by the processor circuit 110.
[0108] In some examples, the output device 140 can functionally or structurally overlap, at least partially, with the user interface 120 and / or the at least one sensor unit 130. For example, a screen can simultaneously serve as a user interface 120 for acquiring response data and as an output device 140 to display visual stimuli, such as colors, images, or interactive content. In another example, a vibrating input button can function both as a haptic input element of the user interface 120 and as a vibration actuator of the output device 140. Similarly, an LED display can simultaneously function as part of the sensor unit 130 for ambient light measurement and as a light source for the output device 140 for targeted visual stimulation.These overlaps enable a compact and integrated design in which components are used multiple times, which can increase the efficiency of the device 100 while simultaneously expanding its range of functions.
[0109] Device 100 represents a technical solution for the automated, context-sensitive delivery of therapeutic stimuli through the structured integration of input, analysis, and output modules. Specifically, the parallel and coordinated processing of psychological response data and synchronously acquired biometric signals generates a measurable state parameter that objectively and quantifies the user's condition. This state determination is not based solely on individual input values, but rather on an algorithmically embedded combination of different signal sources, thus enabling a more robust and error-resistant state analysis. The systematic assignment of the state value to a therapeutic stimulus establishes a targeted technical link between analysis and intervention.
[0110] This provides an integrated architecture that allows measurable inputs to be transformed into situation-dependent stimulus control. The device 100 is thus able to automatically react to complex user states and take targeted measures to modulate these states through therapeutic stimuli. This allows for controlled intervention in a physically representable system (the user's state), whereby input data is processed and converted into controlled actuator signals on the output side. Feedback
[0111] In one example, the processor circuit 110 is further designed to obtain secondary biometric data from the user. This secondary biometric data was acquired via at least one sensor unit after the therapeutic stimulus was delivered. For instance, the processor circuit 110 is further designed to determine an updated user state based on this secondary biometric data. In some examples, the updated state can be determined based on the secondary biometric data and / or secondary response data, where the secondary biometric data was acquired during the input of the secondary response data.
[0112] The timing is designed so that the acquisition of the second biometric data occurs after the delivery of the therapeutic stimulus. The therapeutic stimulus may still be active, already completed, or continuing intermittently. For example, the second biometric data collection can take place immediately after the start or during the presentation of an auditory stimulus to enable real-time observation of physiological responses. In other examples, the acquisition of the second biometric data may only occur after the complete termination of the therapeutic stimulus, for instance, to capture subsequent responses or to analyze a resting phase. In still other examples, the acquisition of the second biometric data can take place during the ongoing or renewed response phase in which the user answers psychological test questions while biometric characteristics are simultaneously recorded.The timing of the second data collection can therefore be flexibly adapted to the stimulus, while always ensuring that the data technically reflect the effect of the therapeutic stimulus. The second set of biometric data thus serves as the basis for an objective evaluation of the effect of the therapeutic stimulus on the user's condition.
[0113] In some examples, the same sensor unit(s) used to acquire the first biometric data can be used to acquire the second. This enables a coherent measurement methodology, where, for example, a microphone, pulse sensor, or camera can continuously or intermittently acquire data before, during, and after stimulus delivery. In another example, different or additional sensors can be used to observe complementary physiological parameters or to capture specific effects of the therapeutic stimulus. For instance, heart rate might initially be recorded via a pulse sensor, while after stimulus delivery, pupillary response could be additionally recorded using a camera or skin temperature via a thermocouple.The choice of sensor elements can therefore be adapted statically or dynamically to the intended therapeutic effect and the underlying state model.
[0114] Based on the second set of biometric data collected after the therapeutic stimulus, an updated user state will be determined. This updated state describes the same technical parameter (or vector) as the previously determined state, such as a numerical stress score, an emotional class state, or a physiological activation index. The updated state can be determined using the same methodological procedures as the original state (see above), for example, rule-based logic or a machine learning model. It can also be derived using the same feature extraction, data preprocessing, and classification structure to ensure methodological consistency.
[0115] In some examples, the updated state can be additionally or alternatively supported by newly entered user response data, such as a repeat test question or a verbal response after the stimulus. Crucially, the updated state must describe the same state parameter as the state before the stimulus, thus enabling direct comparability and quantitative evaluation. For example, an initially measured stress score of 8.5 (on a scale of 0-10) can be influenced by an auditory stimulus, resulting in an updated state of 4.2. In another example, a state initially classified as "tense" can be reclassified as "relaxed" or "neutral" based on subsequent data. This comparability allows for an objective assessment of the therapeutic stimulus's effectiveness.
[0116] In some examples, the updated user state can be derived from a variety of possible input data sources. For instance, the updated state can use both second response data and second biometric data, with the user's second biometric data being captured during the input of the second response data. The crucial aspect is that the updated state describes the same state parameter as the previously determined state, thus ensuring consistent state assessment and evolution.
[0117] For example, the updated state can be determined solely based on the second set of biometric data, without requiring a new response phase. This can occur, for instance, in continuous sensory monitoring, where physiological measurements (such as heart rate, skin conductance, or speech characteristics) are automatically used for reassessment after the stimulus. In another example, the updated state can be based on a combination of newly acquired biometric data and re-entered response data, such as when the user, after the stimulus, provides another assessment of their well-being on a scale that is processed along with measured physiological parameters.
[0118] In other examples, the previous response data set can be reused to calculate the updated state—for instance, if no new input is received—and merged with the new biometric data. This flexibility allows for adaptive adjustment to different interaction scenarios while still ensuring methodologically consistent state estimation over time. In these cases, the processor circuit 110 can decide which data combination to use based on internal parameters or contextual factors.
[0119] In some examples, the processor circuit 110 can be further designed to determine a difference between the user's state and the updated state.
[0120] In some examples, the processor circuit 110 can be further configured to determine a difference between the user's initial state and their updated state. This difference technically represents a quantitative or categorical comparison that provides information about the effectiveness of the previously provided therapeutic stimulus. For example, the difference 1Z can be calculated by subtracting two numerical values if both the initial state Zf inc as well as the updated state Z akt as scalar quantities. In the case of vector quantities, these can be normalized accordingly. For example, the initial state Zf inal determined on a scale of 0 to 10 with a value of 7.8, and the updated state Z akt If 4.2 is the difference, then AZ = Z results. akt - Z finai= -3.6. In this case, a negative value indicates a reduction in the value – for example, successful stress reduction. The difference can also be expressed as a percentage change or as a normalized distance value, e.g., AZ. norm = |z .
[0121]
[0122] In other examples, particularly with discrete state classification (e.g., "tense," "neutral," "relaxed"), the difference can be modeled as a state transition. Here, the difference is represented by a transition path or transition between discrete classes, for example, as AZ. kiasse = ("tense" -> "neutral"). Such categorical differences can be processed using lookup tables, state graphs, or probabilistic transition matrices.
[0123] Technically, the difference can also be calculated by comparing probability distributions if the state and the updated state are given as a vector of class assignment probabilities. In this case, the difference can be calculated, for example, as the Kullback-Leibler divergence or as the Euclidean distance between the probability vectors. This allows for a more refined evaluation, even with uncertainties or overlaps in the class assignments.
[0124] The determined difference can then serve as a basis for adjusting the therapeutic stimulus. It can be used, for example, to evaluate the stimulus's effect, to control adaptive learning processes, to initiate a further therapeutic stimulus, or to dynamically adjust rules or target parameters. In some examples, a defined threshold for Z can be used. t hr eS h OThere exist thresholds above or below which a renewed intervention is automatically triggered or therapeutic success is documented. The difference can serve both as a direct measure of the change in state and as an input variable for subsequent processes in the system logic.
[0125] In some examples, the processor circuit 110 can be further designed to adapt the therapeutic stimulus based on the updated state. This adaptation refers, for example, to a targeted change in at least one parameter of an already provided therapeutic stimulus, whereby this change is technically based on the analysis of the newly determined state Z. aktThe therapeutic stimulus can be adjusted by reconfiguring its parameters. For example, the processor circuit 110 can dynamically adjust the intensity, duration, modality, frequency, rhythm, amplitude profile, or other stimulus parameters of the therapeutic stimulus. If, for instance, an auditory stimulus with a specific volume and frequency was played during the first stimulus run, the volume can be reduced, the frequency changed, or the duration adjusted after the updated state has been detected, in order to achieve targeted fine-tuning.
[0126] The adaptation can be based on rules, model-based decisions, or learning components. In a rule-based process, the updated state Z can be aktThey can be divided into defined categories, each with specific stimulus modifications. For example, a persistently high stress level after the first stimulus trial might lead to the activation of an additional stimulus channel in the next trial, such as a combination of auditory and visual stimuli. Alternatively, a perceived low arousal level might lead to an increase in the frequency of a vibration stimulus.
[0127] In a machine learning-based system, adaptation can be based on a trained model that has learned from historical user data which stimulus modifications were effective during specific state transitions. The model can be Z. akt Use as input and calculate a customized stimulus profile, either as a new parameter combination (e.g. [Intensity = 4, Duration = 10 s, Frequency = 300 Hz]) or as a new stimulus class (e.g. switching from visual to haptic).
[0128] The adjustment can be incremental or complete. In incremental cases, an existing stimulus is modified (e.g., volume is reduced), while in other cases, the previous stimulus is replaced with a new one if the updated condition requires a different therapeutic strategy. This may be necessary, in particular, in the case of unexpected changes in the condition (e.g., worsening instead of improvement).
[0129] In some examples, adjusting the therapeutic stimulus can be part of a closed control loop in which the effectiveness of the therapeutic measure is continuously monitored and automatically adjusted in real time or with a defined delay. In such a technical control loop, the user's current state forms the basis for selecting an initial stimulus. After this stimulus is delivered, an updated state is determined, for example, by capturing further biometric data. The difference between the initial and updated states can serve as a control variable or control deviation. The processor circuit 110 analyzes this difference and, based on this analysis, makes a targeted decision as to whether, how, and to what extent the therapeutic stimulus should be modified or replaced.A low Z value, for example, can lead to an intensification or prolongation of the stimulus, while a high degree of goal attainment leads to a reduction or deactivation. A change in stimulus modality—for example, from auditory to visual—can also be triggered by this regulatory mechanism.
[0130] This control loop can be implemented in software using dynamic control logic, implemented as an adaptation module. This module can utilize both fixed rule sets and adaptive machine learning models to make stimulus decisions based on past responses, user profiles, or statistical correlations. For example, a stimulus strategy that has proven highly effective for multiple users with similar profiles can be preferentially applied to the current user.
[0131] The advantage of this closed-loop control system lies in its ability to dynamically respond to changes in the user's condition and to control stimulus delivery in a context-dependent, user-specific, and effectiveness-oriented manner. This feedback logic allows the Device 100 not only to deliver a single stimulus but also to react specifically to course, trends, and unexpected responses. This increases not only its robustness but also its therapeutic efficiency and enables more precise, individualized user intervention—especially in scenarios with complex psychophysiological dynamics.
[0132] Test questions
[0133] In one example, the processor circuitry can be further designed to select one or more subsequent test questions based on the response data and / or the initial biometric data. The answers to the one or more subsequent test questions can be part of the response data or the second set of response data. The selection can be adaptive, based on the response data previously provided by the user and / or the synchronously captured initial biometric data. The goal of the selection is to control the survey in such a way as to enable a more accurate, targeted, or efficient determination of the user's psychological or physiological state. The selection can be implemented using rule-based logic, where, for example, it is stipulated that certain answers to previous questions trigger the presentation of specific follow-up questions.For example, if a response indicates heightened stress levels, a more in-depth question about the emotional state or the cause of the stress may follow. Similarly, a specific reaction time or a noticeable biometric pattern (e.g., elevated heart rate) can trigger the selection of targeted questions.
[0134] In one example, one or more subsequent test questions can be selected by a machine learning model. This model could be, for instance, a decision tree, a neural network, or a transformer-based Large Language Model (LLM) that uses previous answers and / or biometric data to assess what further information is needed for a precise diagnosis. The model architecture can be trained to select context-relevant, diagnostically valuable, or psychometrically valid questions.
[0135] The Processor Circuit 110 can access a database of test questions classified according to psychological dimensions (e.g., stress, attention, motivation), response formats (multiple choice, scale, free text), and difficulty levels. By utilizing previous response data and biometric indicators, the system can select questions that are best suited, in terms of content, diagnostics, or form, to further specify the user's condition. In some cases, question weighting can be applied. The weighting of questions can adjust dynamically—for example, so that questions with higher diagnostic relevance are asked more frequently or that redundant questions are avoided.
[0136] The newly posed questions, along with their respective answer data, can be incorporated into the state determination and can influence the final state value Z. finaldirectly influence or modify existing assessments. In some examples, the selection process can be iterative: After each new answer, the processor circuit 110 can re-evaluate the intermediate state and decide whether further questions should be asked or whether the current state diagnosis is already sufficiently stable and valid.
[0137] The advantage of this dynamic question selection lies in the fact that the psychological testing can be adapted to the individual reactions and changes in the user's state. This enables personalized, context-adaptive, and efficient diagnostics that would not be achievable with conventional static testing methods. Furthermore, it reduces unnecessary questions, minimizing user workload and potentially improving response quality. The system architecture thus allows for precise, adaptive control of the diagnostic dialogue with high technical relevance. In one example, processor circuit 110 is also configured to evaluate usage data of the provided therapeutic stimuli. Processor circuit 110 is further configured to adapt future selection decisions for the therapeutic stimulus for users with similar conditions based on the evaluated data.The collected usage data can relate to both the technical characteristics of the provided stimuli and their temporal and contextual use. For example, the processor circuit 110 can record, for each individual therapeutic stimulus, the time of delivery, the stimulus modality, the parameterization (e.g., intensity, frequency, duration), and the preceding state. inal Store and analyze the user's data and the subsequently determined updated state of the user.
[0138] Analyzing this usage data can make it possible to create effect profiles for individual stimuli. Such effect profiles can, for example, model the change in state (AZ) typically expected from a specific stimulus in a given initial state. These profiles can be based on heuristic rules or trained predictive models. Based on these profiles, the processor circuit 110 can determine which stimuli have historically resulted in a significant improvement in the condition of users with similar initial states.
[0139] Based on this evaluated data, the processor circuit 110 can be designed to selectively adapt future therapeutic stimulus selection decisions. For example, the processor circuit 110 can employ a clustering algorithm to group similar user states and identify the most effective stimuli within these groups. In another example, a learning-based decision model can be used that, based on historical data, predicts a suitable stimulus for a given state with a high degree of probability. The stimulus can then be provided in real time, adapted to the user's state and their class-specific response history.
[0140] In some examples, the processor circuit 110 can be further configured to systematically record not only the applied therapeutic stimulus, but also the change in the user's state caused by this stimulus. For this purpose, the processor circuit 110 can record the state Z. final before the stimulus is provided and the updated state Z akt The difference between the initial state and the updated state (Z) is compared after stimulus delivery. From this difference (ZA), the effectiveness of the respective therapeutic stimulus in the specific application context can be derived. For example, a user with a baseline score of ZJinal = 8.5 on a stress scale and an updated state of ZB can be compared. aktIf, after application of an auditory stimulus (e.g., calming music), an improvement in condition (AZ = -4.5) is documented, this change can be recorded as a positive effect entry in a database along with the specific stimulus profile (e.g., stimulus type, duration, frequency range). In cases where no significant change in condition occurs or the condition even worsens, this can also be recorded, for example, to avoid ineffective or contraindicated stimuli in similar condition categories in the future.
[0141] Based on this differential data, the processor circuit 110 can learn which stimuli actually cause a change in state in which contexts, state classes, or user profiles. For example, a weighted model can be created that considers not only the frequency but also the effect size (AZ) and, if applicable, the confidence level. This allows the system to prioritize, in future cases, those stimuli that have demonstrably led to an improvement in the user's state with a high probability.
[0142] In further developments, the change in state AZ can be used in a learning recommendation system, for example through reinforcement learning, in which positive changes are incorporated into the model adaptation as a reward signal. This enables continuous optimization of stimulus selection at both the individual and population-wide levels.
[0143] This form of learning-based stimulus selection expands the system into a self-optimizing feedback mechanism that, through continuous user feedback, not only improves acute stimuli but also increases long-term therapy efficiency through data-based individualization.
[0144] Further details and aspects are mentioned in connection with the examples described below. The example shown in Fig. 1 may include one or more optional additional features corresponding to one or more aspects mentioned in connection with the proposed concept or the examples described below (e.g., Fig. 2).
[0145] Fig. 2 shows a schematic representation of a method 200 for determining a therapeutic stimulus. The method is carried out, for example, by the device 100 as described in Fig. 1. The method 200 comprises obtaining 210 response data from a user to one or more psychological test questions, wherein the response data were acquired via a user interface. The method 200 further comprises obtaining 220 initial biometric data from the user, wherein the initial biometric data were acquired via at least one sensor unit during the input of the response data. The method 200 further comprises determining 230 a state of the user based on the response data and the initial biometric data. The method 200 further comprises determining 240 a therapeutic stimulus based on the determined state, wherein the therapeutic stimulus is designed to influence the state.The procedure 200 further includes the provision 250 of the therapeutic stimulus to at least one output device.
[0146] Further details and aspects are mentioned in connection with the examples described above. The example shown in Fig. 2 may include one or more optional additional features corresponding to one or more aspects mentioned in connection with the proposed concept or an example described above (e.g., Fig. 1).
[0147] Further embodiments relate to a non-volatile machine-readable medium on which a program is stored, comprising program code which, when executed on a processor or programmable hardware component, performs a method such as Method 200.
[0148] Further embodiments relate to a computer program with program code for carrying out a method, such as method 200, when the program is executed on a processor or a programmable hardware component.
[0149] The following are some examples of the proposed concept:
[0150] An example (e.g., Example 1) relates to a device (100) for determining a therapeutic stimulus, wherein the device (100) comprises a processor circuit (110) configured to receive response data from a user to one or more psychological test questions, wherein the response data were acquired via a user interface; to receive initial biometric data from the user, wherein the initial biometric data from the user were acquired via at least one sensor unit during the input of the response data; to determine a state of the user based on the response data and the initial biometric data of the user; to determine a therapeutic stimulus based on the determined state, wherein the therapeutic stimulus is designed to change the state; and to provide the therapeutic stimulus to at least one output device.
[0151] An example (e.g., Example 2) relates to a device according to claim 1, wherein the processor circuit is further configured to obtain second biometric data of the user, wherein the second biometric data of the user were acquired via at least one sensor unit after the provision of the therapeutic stimulus, and to determine an updated state of the user based on the second biometric data of the user.
[0152] An example (e.g., Example 3) relates to a device according to claim 2, wherein the processor circuit is further configured to determine a difference between the user's state and the updated state.
[0153] An example (e.g., Example 4) relates to a device according to one of claims 2 or 3, wherein the processor circuit is further configured to adapt the therapeutic stimulus based on the updated state.
[0154] An example (e.g., Example 5) relates to a device according to one of claims 1 to 4, wherein the processor circuit is further configured to capture the first and / or second biometric data of the user via the at least one sensor unit during the input of the response data via the user interface.
[0155] An example (e.g., Example 6) relates to a device according to any one of claims 1 to 5, wherein the first and / or second biometric data of the user describe at least one biometric feature of the user that was detected by the at least one sensor unit.
[0156] An example (e.g., Example 7) relates to the device according to claim 6, wherein the at least one biometric feature comprises at least one of the following features: pitch, volume, speech rate, intonation, prosodic features, speech pause patterns, articulation, pitch, heart rate, respiratory rate, skin conductance, body temperature, pupil diameter, eye movement, blink rate, facial expression, muscle activity, neural activity patterns, or posture. An example (e.g., Example 8) relates to a device according to any one of claims 1 to 7, wherein the processor circuit is further configured to select a subsequent test question based on the response data and / or the initial biometric data.
[0157] An example (e.g., Example 9) relates to a device according to claim 8, wherein the subsequent test question is selected by a machine learning model.
[0158] An example (e.g., Example 10) relates to a device according to any one of claims 1 to 9, wherein the user's condition is a psychological state, a physiological state, or a psychophysiological state of the user.
[0159] An example (e.g., Example 11) relates to the device according to any one of claims 1 to 10, wherein determining the user's state includes applying a rule-based method or a machine learning model.
[0160] An example (e.g., Example 12) relates to the device according to any one of claims 1 to 11, wherein determining the user's state comprises: determining a preliminary state based on either the response data or the biometric data and adjusting the preliminary determined state using the other of the two data sources.
[0161] An example (e.g., Example 13) relates to the device according to any one of claims 1 to 12, wherein determining the therapeutic stimulus comprises applying a rule-based procedure or a machine learning model.
[0162] An example (e.g., Example 14) relates to the device according to any one of claims 1 to 13, wherein the output device is designed to generate at least one of the following therapeutic stimuli: a visual stimulus, an auditory stimulus, a haptic stimulus, a thermal stimulus, an electrical stimulus, a light-based stimulus, or an olfactory stimulus.
[0163] An example (e.g., Example 15) relates to the device according to any one of claims 1 to 14, wherein the processor circuit is further configured to evaluate usage data of the provided therapeutic stimuli and to adapt future selection decisions of the therapeutic stimulus for users with a similar condition based on the evaluated data. An example (e.g., Example 16) relates to a device according to any one of claims 1 to 15, further comprising the user interface.
[0164] An example (e.g., Example 17) relates to a device according to any one of claims 1 to 16, further comprising at least one sensor unit.
[0165] An example (e.g., Example 18) relates to a device according to any one of claims 1 to 17, further comprising at least one output device.
[0166] An example (e.g., Example 19) relates to the device according to any one of claims 1 to 18, wherein the at least one sensor unit comprises at least one of the following components: a microphone, a camera, a pulse sensor, a skin conductance sensor, a thermosensor, an accelerometer, a gyroscope, a photoplethysmograph, an electrode array for recording bioelectrical signals, an infrared sensor, an eye-tracking system, a strain sensor, a radar-based vital parameter sensor, an electroencephalogram, or a capacitive proximity sensor.
[0167] An example (e.g., Example 20) relates to the device according to any one of claims 1 to 19, wherein the user interface comprises one or more of the following components: a touchscreen, a microphone with speech recognition, a graphic display element, a control button, a sensory input unit or a loudspeaker.
[0168] An example (e.g., Example 21) relates to the device according to any one of claims 1 to 20, wherein the output device comprises at least one of the following components: a screen, a loudspeaker, a vibration actuator, a stimulation module, a heating or cooling element, a light source or an odor-generating device.
[0169] An example (e.g., Example 22) relates to a procedure (200) for determining a therapeutic stimulus, comprising: obtaining (210) response data from a user to one or more psychological test questions, wherein the response data were collected via a user interface; obtaining (220) initial biometric data from the user, wherein the initial biometric data from the user were collected via at least one sensor unit during the input of the response data; determining (230) a state of the user based on the response data and the initial biometric data from the user; determining (240) a therapeutic stimulus based on the determined state, wherein the therapeutic stimulus is designed to influence the state; and providing (250) the therapeutic stimulus to at least one output device.
[0170] An example (e.g., Example 23) relates to a non-volatile, machine-readable medium on which a program is stored containing program code for performing one of the methods according to claim 22 when the program is executed on a processor or a programmable hardware component.
[0171] An example (e.g., Example 24) relates to a computer program with program code for performing one of the methods according to claim 22 when the program is executed on a processor or a programmable hardware component.
[0172] The aspects and features described in connection with one of the previous examples can also be combined with one or more of the further examples to replace an identical or similar feature of that further example or to additionally introduce the feature into the further example.
[0173] Examples can also include a (computer) program with program code for executing one or more of the above procedures, or refer to such a program when executed on a computer, processor, or other programmable hardware component. Steps, operations, or processes of various procedures described above can therefore also be executed by programmed computers, processors, or other programmable hardware components. Examples can also include program storage devices, such as digital data storage media, that are machine-, processor-, or computer-readable and encode or contain machine-executable, processor-executable, or computer-executable programs and instructions. The program storage devices can, for example,Digital storage devices include or may include magnetic storage media such as magnetic disks and magnetic tapes, hard disk drives, or optically readable digital data storage media. Further examples may also include computers, processors, control units, field-programmable logic arrays (PLAs), field-programmable gate arrays (PGAs), graphics processing units (GPUs), application-specific integrated circuits (ASICs), integrated circuits (ICs), or system-on-a-chip (SoCs) programmed to perform the steps of the procedures described above.
[0174] It is further understood that the disclosure of several steps, processes, operations, or functions disclosed in the description or claims should not be interpreted as necessarily occurring in the described sequence, unless explicitly stated in a specific case or required for technical reasons. Therefore, the preceding description does not restrict the execution of multiple steps or functions to a specific sequence. Furthermore, in other examples, a single step, function, process, or operation may include and / or be broken down into multiple sub-steps, functions, processes, or operations.
[0175] If certain aspects described in the preceding sections relate to a device or system, these aspects should also be understood as a description of the corresponding procedure. For example, a block, device, or functional aspect of the device or system may correspond to a feature, such as a process step, of the corresponding procedure. Similarly, aspects described in relation to a procedure should also be understood as a description of a corresponding block, element, property, or functional feature of that device or system.
[0176] The following claims are hereby included in the detailed description, each claim being a separate example. It should also be noted that—although a dependent claim may refer to a specific combination with one or more other claims—other examples may include a combination of the dependent claim with the subject matter of any other dependent or independent claim. Such combinations are hereby explicitly proposed unless it is stated in a specific case that a particular combination is not intended. Furthermore, features of a claim are also to be included for each other independent claim, even if that claim is not directly defined as dependent on that other independent claim.
Claims
Patent claims 1. Device (100) for determining a therapeutic stimulus, wherein the device (100) comprises a processor circuit (110) configured to: Obtaining response data from a user to one or more psychological test questions, where the response data was collected via a user interface; Obtaining initial biometric data of the user, wherein the initial biometric data of the user was acquired via at least one sensor unit during the input of the response data; Determining a user's state based on response data and the user's initial biometric data; Determining a therapeutic stimulus based on the determined condition, wherein the therapeutic stimulus is designed to change the condition; and providing the therapeutic stimulus to at least one output device.
2. Device according to claim 1, wherein the processor circuit is further configured to obtain second biometric data of the user, wherein the second biometric data of the user were acquired via at least one sensor unit after the provision of the therapeutic stimulus; and Determining an updated user state based on the user's second biometric data.
3. Device according to claim 2, wherein the processor circuit is further configured to determine a difference between the state and the updated state of the user.
4. Device according to one of claims 2 or 3, wherein the processor circuit is further configured to adapt the therapeutic stimulus based on the updated state.
5. Device according to one of claims 1 to 4, wherein the processor circuit is further configured to capture the first and / or second biometric data of the user via the at least one sensor unit during the input of the response data via the user interface.
6. Device according to any one of claims 1 to 5, wherein the first and / or second biometric data of the user describe at least one biometric feature of the user that was detected by the at least one sensor unit.
7. The device according to claim 6, wherein the at least one biometric feature comprises at least one of the following features: pitch, volume, speech rate, intonation, prosodic features, speech pause patterns, articulation, pitch, heart rate, respiratory rate, skin conductance, body temperature, pupil diameter, eye movement, blink rate, facial expression, muscle activity, neural activity patterns, or posture.
8. Device according to any one of claims 1 to 7, wherein the processor circuit is further configured to select a subsequent test question based on the response data and / or the first biometric data.
9. Device according to claim 8, wherein the subsequent test question is selected by a machine learning model.
10. Device according to any one of claims 1 to 9, wherein the condition of the user is a psychological condition, a physiological condition or a psychophysiological condition of the user.
11. The device according to any one of claims 1 to 10, wherein determining the user's state comprises applying a rule-based method or a machine learning model.
12. The device according to any one of claims 1 to 11, wherein determining the user's state comprises: determining a preliminary state based on either the response data or the biometric data; and an adjustment of the provisionally determined state using the other of the two data sources.
13. The device according to any one of claims 1 to 12, wherein the output device is designed to generate at least one of the following therapeutic stimuli: a visual stimulus, an auditory stimulus, a haptic stimulus, a thermal stimulus, an electrical stimulus, a light-based stimulus or an olfactory stimulus.
14. The device according to any one of claims 1 to 13, wherein the processor circuit is further configured to evaluate usage data of the provided therapeutic stimuli; and Adapting future therapeutic stimulus selection decisions for users with a similar condition based on the evaluated data.
15. Computer program with program code to perform the following steps when the program is executed on a processor or programmable hardware component: Obtained (210) of response data from a user to one or more psychological test questions, wherein the response data were collected via a user interface; Obtained (220) of initial biometric data of the user, wherein the initial biometric data of the user were acquired via at least one sensor unit during the input of the response data; Determine (230) a user's state based on the response data and the user's initial biometric data; Determining (240) a therapeutic stimulus based on the determined condition, wherein the therapeutic stimulus is designed to influence the condition; and Providing (250) the therapeutic stimulus to at least one output device.
