Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103 is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
Priority
Acknowledgment is made of applicant's claim for domestic priority based on provision application 63569575 filed on 03/25/2024.
Claim Rejections - 35 USC § 103
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 103 that form the basis for the rejections under this section made in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-20 are rejected under 35 USC 103(a) as being unpatentable over Williams et al. (US 11094180 B1) in view of Shriberg et al. (US 2019/0385711 A1).
Regarding Claim 1, Williams discloses a method (Fig. 5), comprising:
receiving training data from at least one sensor enabled environment (SEE) related to behavioral, health, wellness, and safety (BHWS) events affecting at least one person under monitoring (PUM) (Fig. 5 for training a machine learning module to identify abnormalities in sensor data corresponding to conditions associated with individuals in home environments; Fig. 5, steps 502-504, Col 12, Rows 28-37, receive historical data detected by a plurality of home environment sensors indicating one or more conditions associated with individuals in each home environment; per Col 8, Rows 57-66, conditions of individual within home environment were determined based on atypical behaviors of individuals (per Col 6, Rows 53-54, atypical physical, mental, emotional, social behaviors) indicated by identified abnormalities or shifts in behavior patterns), wherein each PUM of the at least one PUM is associated with a corresponding SEE of the at least one SEE (Col 6, Rows 11-37, sensors corresponding to video / image / audio / activity / movement body temperature data associated with individual; e.g., medical / health monitor device use data associated with individual; exercise data associated with individual, etc.);
developing a language of wellness (LoW) semantics for a linguistic artificial intelligence or machine learning (AI/ML) model by aligning the BHWS events with a pattern framework indicative of behaviors of the at least one PUM in the corresponding SEE (Fig. 5, step 506, Col 12, Rows 38-55, use AI machine learning models to analyze historical data indicating conditions associated with individuals in each home environment to identify and recognize patterns; Col 13, Rows 28-39, machine learning utilize deep learning focused on pattern recognition including natural language processing and semantic analysis; i.e., learning semantics pattern of an individual in each home environment through analysis of historical data indicating conditions associated with individuals in said home environment);
training the linguistic AI/ML model based on occurrences of the BHWS events in the training data and the LoW semantics (Col 13, Rows 35-39, machine learning includes semantic analysis) such that the linguistic AI/ML model is configured to:
generate a predicted BHWS event based on a series of behaviors observed for a particular PUM in a particular SEE reported to the linguistic AI/ML model (Col 13, Rows 49-60, machine learning analysis of the historical data identifies one or more abnormalities or anomalies in the historical data and their corresponding conditions comprise a predict model to be used to analyze current sensor data to predict condition associated with an individual in the home environment); and
generate a predictive alert in response to identifying that the predicted BHWS event disobeys the LoW semantics (Col 13, Rows 55-60, use the prediction model to analyze current sensor data to predict a condition associated with an individual in the home environment based upon certain abnormal or anomalous patterns in current sensor data including pattern recognition based on natural language processing / semantic analysis per Col 13, Rows 28-39; see Col 13, Row 64 – Col 14, Row 4 in view Col 10, Rows 45-46, Col 11, Rows 26-28, and Col 11, Rows 56-61, steps 404-408 of Fig. 4, compare identified abnormalities or anomalies in the historical data and their corresponding conditions to abnormalities or anomalies in current sensor data to determine a condition associated with the identified abnormalities or anomalies of an individual in the home environment and generate a notification indicating the condition associated with the individual to a caregiver of the individual).
Williams does not disclose developing a language of wellness (LoW) syntax for a linguistic artificial intelligence or machine learning (AI/ML) model.
Shriberg discloses a health screening monitoring system (¶149) developing a language of wellness (LoW) syntax for a linguistic artificial intelligence or machine learning (AI/ML) model (¶403, model training includes training syntactic language model training 3802, semantic pattern model training 3804; ¶405, syntactic language model assesses a patient’s depression from syntactic characteristics of the patient’s speech; ¶406, semantic pattern model assesses a patient’s depression from positive and negative content of patient’s speech) by aligning BHWS events with a pattern framework indicative of behaviors of the at least one PUM in the corresponding SEE (¶345, syntactic model identifies situations where ASR outputs and speech complexities correspond to internal focus correlated with depression; ¶390, after modeling is completed, combine / fuse each model).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to develop a language of wellness “LoW” syntax for a linguistic artificial intelligence or machine learning AI / ML model by aligning BHWS events with a pattern framework indicative of behaviors of PUM in a responding SEE to train the linguistic AI / ML model based on the LoW syntax (Shriberg, ¶166, combine language, acoustic, and visual models to produce a composite model and using clinical data and patient data to train composite model) to generate a predictive alert in response to identifying a predicted BHWS event disobeys the LoW syntax (Shriberg, ¶295, estimate a health state of a patient using what the patient says, how the patient says it, contemporaneous facial expressions, eye expressions, and poses in combination; e.g., Shriberg, ¶345, identify when individual is depressed; Shriberg, ¶467, identify individual’s highly emotional, anxious, semantically urgent situation) in order to generate an alert / notification indicating critical condition associated with the individual (Williams, Col 11, Row 56 – Col 12, Row 2; compare Shriberg, ¶467).
Regarding Claim 2, Williams does not disclose wherein the predicted BHWS event is predictively generated based on one or more digital twins associated with the particular PUM that simulate actions of the particular PUM within the particular SEE.
Shriberg discloses wherein the predicted BHWS event is predictively generated (¶183, apply models of runtime model server 504 to audiovisual signals of the patient to estimate a patient’s health state) based on one or more digital twins associated with the particular PUM that simulate actions of the particular PUM within the particular SEE (¶314, personalize a model or set of models for a particular individual based upon their past history and label data known for the individual; i.e., a personalized model models / simulates past actions of an individual and therefore a digital twin of the individual; ¶315, during analysis, apply the personalized model for their screening or monitoring).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to predict BHWS event based on one or more digital twins associated with the particular PUM that simulate / model actions of the particular PUM within the particular SEE in order to produce more accurate and granular results (Shriberg, ¶314); i.e., more accurate and granular predictions / estimates of patient health state.
Regarding Claim 3, Williams does not disclose wherein linguistic AI/ML model is configured to operate in conjunction with a physics engine associated with the particular PUM to identify when a current BHWS event or the predicted BHWS event is outside of a physical capability of the particular PUM to perform.
Shriberg discloses wherein linguistic AI/ML model is configured to operate in conjunction with a physics engine associated with the particular PUM (¶301, based on the results of the model server (i.e., patient health estimate or predictions), interaction engine dynamically adapts the course of interaction to achieve the system’s goals, rules engine determines what, when, and who to send communication to, process engine with clinical and operational protocol logic to pass messages through a communication chain for serial completion of tasks, and routing engine send any messages to the user’s platform of choice) to identify when a current BHWS event or the predicted BHWS event is outside of a physical capability of the particular PUM to perform (¶467, speaker speech is determined to be highly emotional and anxious while semantically describing a highly urgent situation (e.g., car accident with severe injuries or frantic speech and screaming with semantic content describing violence in a school hallway), interactive screening or monitoring server logic triggers immediate notification of law enforcement).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to configure the linguistic AI/ML model to operate in conjunction with a physics engine associated with the particular PUM to identify when a current BHWS event or the predicted BHWS event is outside of a physical capability of the particular PUM to perform in order to implement alert rules by interactive screening or monitoring server logic (Shriberg, ¶467) implementing rules / conditional logic determining what, when, and who to send communications to (Shriberg, ¶301).
Regarding Claim 4, Williams discloses wherein the training data include tokenized representations of various BHWS events (Col 12, Rows 38-58, historical data detected by sensors for training machine learning model includes voice recognition / word recognition on text or voice speech data (i.e., tokenization) as input / prompt to the prediction model to make predictions), which include encrypted sensor data from the at least one SEE (Col 17, Rows 23-32, Sensing Peripheral heuristic Evidence, Reinforcement, and Engagement System (“SPHERE”) for autonomous and passive monitoring and measuring spheres of information, behaviors, data points to look for routines and changes to those routines; per Col 17, Rows 44-53, producing custom time period over time period digital snapshots of collected data to easily understand changes in norms, routines, and behaviors and keep all this personal data safe using blockchain encryption) and an unencrypted data label identifying a type of BWHS event associated with the encrypted sensor data (Col 13, Rows 10-11, iterative training the neural network / machine learning model using labeled training samples; i.e. encrypted historical data collected by the SPHERE system).
Regarding Claim 5, Williams as modified by Shriberg discloses wherein the linguistic AI/ML model is one of:
a large language model (LLM);
retrieval augmented generation (RAG) model;
private large language model (PLLM); and
a specialized language model (Shriberg, ¶16, NLP model such as a statistical language model; as applied to Williams, Col 13, Rows 28-39, train a natural language processing / semantic analysis machine learning model as prediction model).
Regarding Claim 6, Williams as modified by Shriberg discloses wherein the particular SEE is not included in the at least one SEE from which the training data are received (Shriberg, ¶315, during analysis (i.e., using the prediction model / composite model to estimate / predict patient health), a client is identified and employ a personalized model for screening or monitoring when able. If personalized model is not available or no metadata for identifying the client is available, use a generic, population wide model; i.e., when an identified client does not have a personalized model (i.e., no historical data / training data specific to the client was available to train such personalized model) or the client is not an identifiable client, use generic model), and the particular PUM is not included in the at least one PUM associated with the at least one SEE (Shriberg, ¶458, interactive screening or monitoring server logic captures ambient speech and determines whether the speech captured is spoken by a voice registered with the health screening or monitoring server; i.e., those voice not registered with the server are not analyzed to make predictions / estimates).
Regarding Claim 7, Williams discloses wherein the linguistic AI/ML model is further configured to identify an immediate danger state based on a currently or previously observed BHWS event affecting the particular PUM in the particular SEE and to generate an immediate alert based on the immediate danger state (Col 11, Row 65 – Col 12, Row 2, when condition associated with the individual is an emergency medical condition, request ambulance, police, and fire services).
Regarding Claim 8, Williams discloses wherein disobeying the LoW syntax includes disobeying an established spatial zone in the particular SEE, wherein the predicted BHWS is predicted to occur outside of the established spatial zone (Col 19, Rows 53-59 in view of Col 17, Rows 30-32, looking for routines and changes to those routines such as time spent outside of the house; see also Shriberg, ¶461, in police telephone call triage, location of the speaker and qualities of the speaker’s voice such as emotion, energy, and substantive content of the speaker’s speech is important; i.e., in a police emergency (e.g., ¶467, location of a car accident or location of a school violence), the location is unknown and thus outside of any established spatial zone; see also ¶467, in triggering condition of captured speech is particularly serious and urgent, interactive screening or monitoring server logic 502 reports the location of the speaker if it may be determined).
Regarding Claim 9, Williams discloses wherein disobeying the LoW syntax includes disobeying a time window, wherein the predicted BHWS is predicted to occur outside of the time window, wherein the time window is one of an absolute time window during a day or a relative time window from performance of a previous behavior by the PUM (Col 11, Rows 37-46, captured data previously indicated that the individual showered once per day but indicates that the individual currently showers only once per week, determine a hygiene condition associated with the individual).
Regarding Claim 10, Williams discloses wherein disobeying the LoW syntax includes disobeying an established order for performing a first behavior relative to a second behavior (Col 18, Rows 8-17, keeping track an individual’s daily routine such as sleep and wake time, breakfast, lunch, and dinner routines, movement around house / property routine; in view of Col 17, Rows 31-32, looking for routines and changes to those routines).
Regarding Claim 11, Williams discloses wherein the linguistic AI/ML model is further configured to: generate a reactive alert in response to identifying that the PUM has performed or is currently performing a behavior that disobeys the LoW syntax (Col 11, Rows 37-46, captured data previously indicated that the individual showered once per day but indicates that the individual currently showers only once per week, determine a hygiene condition associated with the individual; see also Shriberg, ¶467, in triggering condition of captured speech is particularly serious and urgent, interactive screening or monitoring server logic 502 reports the location of the speaker if it may be determined).
Regarding Claim 12, Williams discloses a method, comprising:
deploying a linguistic artificial intelligence or machine learning (AI/ML) model, the linguistic AI/ML model being configured to process prompts according to a language of wellness (LoW) semantics (Col 4, Rows 44-51, using a machine learning module for training / developing a predictive model identifying a potential condition associated with an individual in a home environment using abnormalities or anomalies in the sensor data associated with the home environment; Col 13, Rows 28-39, machine learning / deep learning of pattern recognition includes natural language processing and semantic analysis; Col 13, Rows 55-56, using the predictive model to analyze current sensor data; i.e., sensor data includes natural language inputs or prompts to the predictive model to perform natural language processing and semantic analysis);
receiving a prompt from a computing device located remotely from where the linguistic AI/ML model is deployed (Fig. 1, Col 5, Rows 50-55, sensor data sensor 102 associated with smartphone in home environment 104; Col 6, Rows 49-51, server 106 in Fig. 1 performs the analysis), the prompt including at least one tokenized behavioral, health, wellness, and safety (BHWS) event occurring in a particular Sensor Enabled Environment (SEE) associated with the computing device and with a particular person under monitoring (PUM) (Col 6, Rows 49-51, server 106 performs the analysis; Col 13, Row 55 – Col 14, Row 5, use predictive model to analyze sensor data per Fig. 4 step 404 by comparing one or more abnormalities or anomalies in current sensor data to the identified abnormalities or anomalies in the historical data and their corresponding conditions in order to predict a condition associated with an individual in the home environment base on abnormal / anomalous patterns in current sensor data; per Col 12, Rows 55-58 and Col 19, Rows 53-59, detecting activity and movement of the individual includes when detecting / sensing phone text and call, use voice recognition / word recognition on text or voice speech data (i.e., tokenization) as input / prompt to the prediction model to make predictions);
generating a predicted BHWS event based on the prompt (Col 11, Rows 26-32, determine a condition associated with an individual in the home environment based upon identified abnormalities or anomalies corresponding to atypical behaviors or shifts in behavior patterns); and
transmitting the predicted BHWS event to the computing device as an output responsive to the prompt (Col 7, Row 66 – Col 8, Row 2 and 11, Rows 56-60, generate a notification indicating condition associated with the individual for display to a caregiver of the individual by transmitting the notification to a caregiver device 108; in view of Col 8, Rows 3-8, caregiver device 108 is the smart phone with the sensor 102 in the home environment 104 of Fig. 1).
Williams does not disclose the linguistic AI/ML model being configured to process prompts according to a language of wellness “LoW” syntax.
Shriberg discloses a health screening monitoring system (¶149) developing a language of wellness (LoW) syntax for a linguistic artificial intelligence or machine learning (AI/ML) model to process prompts according to a language of wellness syntax (¶403, model training includes training syntactic language model training 3802, semantic pattern model training 3804; ¶405, syntactic language model assesses a patient’s depression from syntactic characteristics of the patient’s speech; ¶406, semantic pattern model assesses a patient’s depression from positive and negative content of patient’s speech; ¶345, syntactic model identifies situations where ASR outputs and speech complexities correspond to internal focus correlated with depression).
It would’ve been obvious to one ordinarily skilled in the art before the effective filing date of the invention to develop / configure a language of wellness “LoW” syntax for a linguistic artificial intelligence or machine learning AI / ML model (Shriberg, ¶166, combine language, acoustic, and visual models to produce a composite model and using clinical data and patient data to train composite model) to process prompts according to the language of wellness syntax in order to generate an alert / notification indicating critical condition associated with the individual (Williams, Col 11, Row 56 – Col 12, Row 2; compare Shriberg, ¶467) such as speech syntax indicating depression (Shriberg, ¶345).
Regarding Claim 13, Williams as modified by Shriberg discloses wherein the linguistic AI/ML model is one of:
a large language model (LLM);
retrieval augmented generation (RAG) model;
private large language model (PLLM); and
a specialized language model (Shriberg, ¶16, NLP model such as a statistical language model; as applied to Williams, Col 13, Rows 28-39, train a natural language processing / semantic analysis machine learning model as prediction model).
Regarding Claim 14, Williams as modified by Shriberg discloses wherein the particular SEE is not included among training SEE from which training data used to train the linguistic AI/ML model are received, and the particular PUM is not included among training PUM associated with the training SEE (Shriberg, ¶315, during analysis (i.e., using the prediction model / composite model to estimate / predict patient health), a client is identified and employ a personalized model for screening or monitoring when able. If personalized model is not available or no metadata for identifying the client is available, use a generic, population wide model; i.e., when an identified client does not have a personalized model (i.e., no historical data / training data specific to the client was available to train such personalized model), use generic model).
Regarding Claim 15, Williams discloses wherein the linguistic AI/ML model is further configured to identify an immediate danger state based on a currently or previously observed BHWS event affecting the particular PUM in the particular SEE and to generate an immediate alert based on the immediate danger state (Col 11, Row 65 – Col 12, Row 2, when condition associated with the individual is an emergency medical condition, request ambulance, police, and fire services).
Regarding Claim 16, Williams discloses in response to identifying that the predicted BHWS event disobeys the LoW syntax: generating a predictive alert (Col 13, Rows 55-60, use the prediction model to analyze current sensor data to predict a condition associated with an individual in the home environment based upon certain abnormal or anomalous patterns in current sensor data including pattern recognition based on natural language processing / semantic analysis per Col 13, Rows 28-39; see Col 13, Row 64 – Col 14, Row 4 in view Col 10, Rows 45-46, Col 11, Rows 26-28, and Col 11, Rows 56-61, steps 404-408 of Fig. 4, compare identified abnormalities or anomalies in the historical data and their corresponding conditions to abnormalities or anomalies in current sensor data to determine a condition associated with the identified abnormalities or anomalies of an individual in the home environment and generate a notification indicating the condition associated with the individual to a caregiver of the individual); and
transmitting the predictive alert to a stakeholder associated with the particular SEE or the particular PUM (Col 7, Row 66 – Col 8, Row 2 and 11, Rows 56-60, generate a notification indicating condition associated with the individual for display to a caregiver of the individual by transmitting the notification to a caregiver device 108).
Regarding Claim 17, Williams discloses wherein disobeying the LoW syntax includes at least one of:
disobeying an established spatial zone in the particular SEE, wherein the predicted BHWS is predicted to occur outside of the established spatial zone (Col 19, Rows 53-59 in view of Col 17, Rows 30-32, looking for routines and changes to those routines such as time spent outside of the house; see also Shriberg, ¶461, in police telephone call triage, location of the speaker and qualities of the speaker’s voice such as emotion, energy, and substantive content of the speaker’s speech is important; i.e., in a police emergency (e.g., ¶467, location of a car accident or location of a school violence), the location is unknown and thus outside of any established spatial zone; see also ¶467, in triggering condition of captured speech is particularly serious and urgent, interactive screening or monitoring server logic 502 reports the location of the speaker if it may be determined);
disobeying the LoW syntax includes disobeying a time window, wherein the predicted BHWS is predicted to occur outside of the time window, wherein the time window is one of an absolute time window during a day or a relative time window from performance of a previous behavior by the PUM (Col 11, Rows 37-46, captured data previously indicated that the individual showered once per day but indicates that the individual currently showers only once per week, determine a hygiene condition associated with the individual); and
disobeying an established order for performing a first behavior relative to a second behavior (Col 18, Rows 8-17, keeping track an individual’s daily routine such as sleep and wake time, breakfast, lunch, and dinner routines, movement around house / property routine; in view of Col 17, Rows 31-32, looking for routines and changes to those routines).
Regarding Claim 18, Williams discloses wherein the linguistic AI/ML model is further configured to: generate a reactive alert in response to identifying that the PUM has performed or is currently performing a behavior that disobeys the LoW syntax (Col 11, Rows 37-46, captured data previously indicated that the individual showered once per day but indicates that the individual currently showers only once per week, determine a hygiene condition associated with the individual; see also Shriberg, ¶467, in triggering condition of captured speech is particularly serious and urgent, interactive screening or monitoring server logic 502 reports the location of the speaker if it may be determined).
Regarding Claim 19, Williams as modified by Shriberg discloses wherein the linguistic AI/ML model is provided for treatment or prophylaxis of a health condition indicated for the particular PUM in a health care plan (HCP) included or referenced in the prompt (Shriberg, ¶518, determine if patient is following treatment defined by patient’s treatment plan; ¶519, predict whether a patient adheres to a course of treatment or medication by analyzing voice based biomarkers; i.e., questioning the patient with whether the patient adheres to a treatment plan and analyzing voice biomarkers to determine if the patient is or is not following the treatment plan).
Regarding Claim 20, Williams discloses wherein the prompt includes a senor data stream from at least one sensor disposed in the SEE or a token stream from at least one tokenization service or model that processes sensor data from sensors in the SEE (Col 12, Rows 55-58 and Col 19, Rows 53-59, detecting activity and movement of the individual includes when detecting / sensing phone text and call, use voice recognition / word recognition on text or voice speech data (i.e., tokenization) as input / prompt to the prediction model to make predictions).
Conclusion
Prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
US 2025/0131206 A1 discloses a LLM based home assistant (¶31) that takes in multi-modality sensor information from a sensing hub (¶49) to generate user specific and context aware suggestions or actions (¶46) such as detecting user had possibly fell in bathroom and call caregiver or 911 for assistance (¶70).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to examiner Richard Z. Zhu whose telephone number is 571-270-1587 or examiner’s supervisor Hai Phan whose telephone number is 571-272-6338. Examiner Richard Zhu can normally be reached on M-Th, 0730:1700.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/RICHARD Z ZHU/Primary Examiner, Art Unit 2654 09/04/2026
Prosecution Insights