Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.
1.1Sampling between task blocks
What ifWhat if the system records physiology only during scheduled rest intervals between task blocks?
Claim language“during performance of a task”
A task controller starts a 60-second electrocardiogram recording when the operator enters a rest interval and closes the recording before the next block begins. The system computes heart rate and beat interval variability from the rest interval and uses the resulting recovery estimate to select a pacing cue before the next block.
1.2Feedback gated by signal quality
What ifWhat if the system passes raw physiological samples to the estimator and uses a separate signal quality gate to control feedback?
Claim language“conditioning the set of physiological signals to remove artifacts”
The estimator receives every digitized sample without correcting or discarding samples. A separate gate checks electrode contact impedance and converter saturation, suppresses stimulus delivery during poor contact or clipping, and enables delivery after five consecutive seconds of acceptable readings.
1.3State inference from measured values
What ifWhat if the state estimator uses scalar physiological measurements without generating physiological state embeddings?
Claim language“generate a set of physiological state embeddings”
A beat detector and a respiratory cycle detector calculate heart rate in beats per minute and breathing rate in breaths per minute. A decision tree compares each measurement with its resting baseline and outputs an autonomic arousal category for the feedback selector.
1.4Direct continuous vector classification
What ifWhat if a classifier operates directly on continuous physiological state embeddings without projecting them into a token space?
Claim language“projecting the set of physiological state embeddings into a token space”
An encoder emits one 32-coordinate floating-point vector per analysis window. A decision forest tests the coordinates in their original representation along with a separate task-phase field and returns class probabilities for cognitive load.
1.7Haptic feedback through a seat
What ifWhat if the operator feels the feedback through a vibration actuator fixed to the seat?
Claim language“a haptic feedback from a device worn by the operator”
A controller switches on a seat-mounted vibration transducer for two 150-millisecond pulses separated by 200 milliseconds. The seat pan and cushion transmit the pulses to the seated operator.
Section 2 · machine-generated
Variations
Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.
Other field of use
Feedback during musical rehearsal
A respiratory belt and a wrist motion sensor record breathing and bowing motion during string instrument rehearsal. After bandpass filtering, a temporal encoder converts synchronized signal windows into physiological state embeddings, which a projection layer maps into input tokens for a second model. The second model combines those tokens with the current score section and target tempo to estimate autonomic arousal. When arousal exceeds the player's calibration threshold, a wrist actuator delivers a pacing pulse at the next marked rest.
Component substitution
Wavelet suppression of recording artifacts
Discrete wavelet thresholding replaces independent component analysis in the electroencephalography conditioning stage. The conditioner decomposes each channel into wavelet bands and attenuates large transient coefficients in bands identified during a blink and jaw-clench calibration. The conditioner reconstructs the channels and passes them to an encoder whose output embeddings enter the second model as projected tokens. The second model combines the tokens with the current task phase to estimate attentional focus and selects a visual cue when focus falls below a configured threshold.
Parameter or scale variation
Overlapping windows for sustained arousal
An electrocardiography channel sampled at 250 hertz and a respiratory channel sampled at 25 hertz share a timestamp clock. After filtering, separate encoder branches process aligned 60-second windows every 10 seconds and concatenate their embeddings for projection into the second model's token space. The second model combines the projected tokens with the current task phase to produce an arousal score on a scale from 0 to 1. Three successive scores above 0.75 trigger an audio cue, after which a 60-second holdoff suppresses repeat cues.
Combination with existing systems
Feedback integrated with a driving simulator
A fixed-base driving simulator shares timestamps with a driver-facing eye tracker and a chest electrocardiography recorder. The conditioning stage masks missing gaze samples and filters electrocardiography before an encoder generates physiological embeddings and a projection layer maps them to model tokens. A second model uses the tokens with simulator telemetry for vehicle speed, road curvature, and scenario phase to estimate cognitive load. The simulator displays a dashboard cue when load exceeds a calibrated threshold during a straight-road segment and logs the cue with the scenario events.
Automated or data-driven variant
Automatic weighting of sensor channels
A quality classifier evaluates each filtered electrocardiography, respiratory, and motion window using sample gaps, clipping, and abrupt amplitude changes. The encoder generates a separate embedding for each modality, and the input stage scales each embedding by its quality score before projecting it into the second model's token space. The second model receives a missing-data marker for each rejected channel and combines the remaining tokens with the current task phase to estimate arousal. The feedback controller issues a pulse when arousal exceeds its threshold and at least two channels pass the quality check.
Cue selection from measured response
A respiratory belt and a wrist motion sensor feed a conditioning stage and a temporal encoder that generates physiological embeddings. A projection layer maps the embeddings into tokens, and a second model uses the tokens and current task phase to estimate arousal. A cue controller keeps a table of response scores for a visual breathing marker and a vibration pattern, using the change between the 30-second mean arousal estimates before and after each cue to update its score. On subsequent threshold crossings, the controller selects the cue with the largest measured average reduction and alternates cues until it has an initial score for each.
Specific version of a part
Direct synthesis of feedback
A voltage-controlled oscillator maps the arousal estimate continuously to a frequency between 220 and 440 hertz. An amplifier drives an earpiece with the oscillator output, and a 200-millisecond smoothing filter on the control voltage limits abrupt pitch changes.
Synchronized chest sensor recordings
Three adhesive chest electrodes feed a differential electrocardiography amplifier sampled at 500 hertz. A thoracic stretch belt records respiration at 50 hertz, and a shared clock timestamps both streams during the task.
Mains interference notch filtering
A second-order digital notch filter centered at 60 hertz processes each electroencephalography channel sampled at 500 hertz. A 2-hertz notch bandwidth suppresses mains interference while retaining adjacent frequency content.
Affine projection into token space
A learned 256-by-64 weight matrix and a 256-coordinate bias map each 64-coordinate physiological state embedding to a 256-coordinate input embedding. The second model receives the projected vectors as input tokens alongside tokens representing task context.
Contextual cognitive load classification
An attention-based sequence model reads eight projected physiological tokens together with tokens indicating task phase and the operator's resting baseline category. A three-class output layer reports probabilities for low, moderate, and high cognitive load.
Section 3 · quoted from USPTO records
Other work
Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.
1.1obtaining a set of physiological signals from a set of physiological sensors associated with an operator during performance of a task, wherein the set of physiological signals includes at least one of: electroencephalography (EEG) signals, electrocardiography (ECG) signals, eye-tracking signals, respiratory signals, or biomechanical motion signals;
One or more sensors (such as force plates, 3D camera sensors, inertial sensors as described hereinbefore) measure the attempt by the subject 15 to move according to the predetermined motions. In particular, the 3D camera sensors determine corresponding subject 15 physical position in space and any associated movements.
B · US 10,258,259 B1 · ¶ 95
1.2conditioning the set of physiological signals to remove artifacts, wherein conditioning the set of physiological signals includes at least one of: independent component analysis for removing eye movement artifacts and muscle noise, bandpass filtering, or notch filtering;
The advantage of aforementioned solution is the possibility to integrate input sensors into a flexible thoracic harness and/or a special sensory suit retaining the possibility of mounting sensors using classic disposable sensors, e.g. This considerably facilitates the application of the device in practice, reduces the requirements for its maintenance, and enhances the quality of measured signals as for e.g. higher resistance against movement artifacts.
A · US 2016/0183797 A1 · ¶ 12
1.3processing the set of physiological signals through a first model to generate a set of physiological state embeddings;
1.4projecting the set of physiological state embeddings into a token space of a second model to produce a set of projected embeddings that can be interpreted by the second model;
1.5determining, using the second model operating on the set of projected embeddings and contextual information about the operator, a neurophysiological state of the operator, wherein the neurophysiological state includes at least one of: a cognitive load level, an attentional focus metric, an autonomic arousal level, or a motor preparation state;
Using 4-10 sensors and/or 12 the push button, subject's reactions are collected, and then stored, visualized and evaluated in 3 control unit for the purpose of assessing the psychophysiological state, or the status of mental resistance against the presented stimuli—the extent of the stress triggered and the reaction and/or the speed of the subject's reaction to the stimuli.
A · US 2016/0183797 A1 · ¶ 29
1.6selecting, based on the neurophysiological state, a biofeedback stimulus from a set of biofeedback stimuli; and
If the intelligent controller 20 determines that the biomechanical state measure (for example COG) has moved beyond a threshold, the intelligent controller 20 activates one or more sensory feedback elements that correspond to the segment. Thus, the subject 15 receives a sensory stimulus (for example vibrotactile and auditory), or feedback, when there is a variance between the location of the biomechanical state measure and the predetermined movement threshold.
B · US 10,258,259 B1 · ¶ 95
1.7delivering the biofeedback stimulus to the operator through at least one of: a haptic feedback from a device worn by the operator, an auditory transducer, or a visual display element.
Thus, the subject 15 receives a sensory stimulus (for example vibrotactile and auditory), or feedback, when there is a variance between the location of the biomechanical state measure and the predetermined movement threshold.
B · US 10,258,259 B1 · ¶ 95
An empty cell means only that no passage was reproduced there.
Statements in these documents
Current solutions in the commonly available setup similarly do not allow for real-time evaluation or classification of psychophysiological states online as the control software, which is primarily designed for diagnostics lacks appropriate design for such substantially different tasks as real-time evaluation.
A · US 2016/0183797 A1 · ¶ 4
The above disadvantages are overcome to a great extent by the Multifunction Biotelemetry Support System for Psychophysiology Monitoring measuring and further evaluating physiology variables in real time.
A · US 2016/0183797 A1 · ¶ 8
Such measurements are useful for the assessment of the presence and severity of disequilibrium, balance and motion disorders.
B · US 10,258,259 B1 · ¶ 4
Such measurements are also useful for the determination of the therapeutic application of vibrotactile, auditory, or visual feedback to a subject during the subject's attempt to perform a predetermined task.
B · US 10,258,259 B1 · ¶ 4
Cite aspatent.fyi, US 2026/0354766 A1, claim 1, first published 2026-11-19T03:31:31-05:00, https://patent.fyi/US20260354766/, SHA-256 9fed891cec157a263389b6157e322b3521841dd7e7fbbe22c4327fc084bb12b8