DETAILED ACTION
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 .
Claim Objections
Claim 15 is objected to because of the following informalities:
"where" – Appears to be misspelled and should be "wherein" for consistency with patent claim language conventions. Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 6, 12, 14, and 15 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
In relation to claim 6: "the processed clinical data" lacks antecedent basis—no prior introduction of "processed" clinical data (Claim 1 mentions "clinical data" but not "processed"). This could render the claim indefinite, as it assumes processing without basis;
in relation to claim 12, similar issue as Claim 6—"the processed clinical data" lacks antecedent (no prior "processed" reference in Claim 7);
in relation to claim 14, “patients with insufficient interaction with the treatment device": "the treatment device" (singular) lacks clear antecedent basis—Claim 13 uses "treatment devices" (plural) for the population; this shift to singular could be ambiguous or indefinite and "the training module": lacks antecedent basis—no prior introduction of "a training module" (Claim 13 mentions "training the machine learning model" but not a "module"); and
in relation to claim 15, "an application executed on the mobile device": "the mobile device" (singular) lacks antecedent basis—prior reference is to plural "mobile devices"; this could be indefinite; and "the application interfacing with the treatment device": "the treatment device" (singular) has similar issue as in Claim 14 (plural in Claim 13).
Claim Rejections - 35 USC § 103
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 (i.e., changing from AIA to pre-AIA ) 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.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth 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 and 7 are rejected under 35 U.S.C. 103 as being unpatentable over Barrett et al. (US 2020/0058403A1; hereinafter “Barrett (A)”) in view of Pedersen et al. (“Predicting Dropouts from an Electronic Health Platform for Lifestyle Interventions: Analysis of Methods and Predictors”; hereinafter “Pedersen (C)”).
In relation to claim 1, claim 1 recites a system to predict churn in relation to the use of a treatment device by a patient.
Barrett (A) discloses a comprehensive system for monitoring medicament devices, such as inhalers, to improve patient treatment. The system includes a communication interface to collect event data from device sensors, a storage device (database server) to store this data along with other patient information, and an analytics module that uses machine learning to analyze the data, identify triggers for adverse events, and issue notifications based on a threshold analysis. For example, Barrett (A) discloses:
"The asthma analytics system includes client computing devices 110, a medicament device sensor 120, a medicament device 160, an application server 130, database server 140, and a network 150." (Barrett, [0024])
"The medication event information is sent to the application server 130 for use in analysis, generation of asthma rescue event notifications, and in aggregate analyses of event data across multiple patients." (Barrett, [0041])
"The database server 140 stores patient and provider data related data such as profiles,
medication events, patient medical history (e.g., electronic medical records)." (Barrett, [0058])
Barrett (A) further discloses the use of machine learning and statistical analysis to predict
patient outcomes. The model is trained on population data and uses threshold analysis to
trigger notifications:
"In one embodiment, the model 560 is trained using a machine learning technique,
examples of which include but are not limited to linear, logistic, and other forms of
regression (e.g., elastic net), decision trees (e.g., random forest, gradient boosting),
support vector machines, classifiers (e.g. Naïve Bayes classifier), and fuzzy matching."
(Barrett, [0144])
"In one embodiment, the baseline risk threshold is set as a fraction (percentage) of the
total number of usage events over the specified period... To label a day as an
exacerbation day, the model 560 assigns a label if the number of rescue inhaler usage
events in a given day is greater than the threshold..." (Barrett, [0149])
"The notification module 580 generates a trigger identification notification including any
one or more of: the identified trigger, remaining unidentified triggers, information
characterizing the identified trigger, suggestions and..." (Barrett, [0157])
Barrett (A) thus teaches the core components of the claimed system: a communication
interface for event data collection, a storage device for patient data, and an analysis module that inputs this data, applies a machine learning model trained on population data, uses a threshold, and triggers an action. However, Barrett (A) does not explicitly disclose applying a machine learning model to determine the "likelihood of churn" over a predetermined period, nor does it explicitly describe a machine learning pipeline trained on population data specifically to determine triggers for "patient churn". Barrett's focus is on predicting "rescue events" rather than treatment discontinuation.
Pedersen (C) addresses this missing element. This reference teaches the prediction of "dropouts" from digital health interventions, a term synonymous with "churn" in this context. Pedersen (C) discloses the use of data mining and machine learning models trained on data from a population of users to identify individuals at high risk of discontinuing the intervention. The abstract of Pedersen (C) states in the Background and Conclusions subsections:
“Machine learning methods have been proven to predict dropouts in other settings but lack implementation in health care."
"Dropouts from eHealth lifestyle interventions can be predicted using various data mining methods. This can support health coaches in preventing attrition."
Based on the above teachings, it would have been obvious to one of ordinary skill in the art at the time of the invention to modify the system of Barrett (A) to incorporate the churn prediction methods taught by Pedersen (C). The motivation to combine these references is strong, as both address the well-known problem of patient non-adherence and discontinuation in digital health. Applying Pedersen's churn prediction model to Barrett's device monitoring system would be a logical step to enhance patient management by proactively identifying and intervening with patients at risk of ceasing their treatment, thereby improving the overall efficacy of the system.
In relation to claim 7, claim 7 recites method steps that correspond to the system elements of Claim 1. The same rejection rationale applies:
As discussed above, Barrett (A) discloses methods for:
Collecting patient data and rescue medication events (clinical and event data)
Monitoring medicament device operation via sensors and communication interfaces
Applying machine learning and statistical analysis
Determining severity/threshold levels; and
Triggering notifications based on analysis.
Barrett (A) does not disclose a “churn” prediction capability. However, Pedersen (C) discloses methods for predicting dropout/churn from eHealth platforms using data mining. Therefore, for an artisan skilled in the art, the implementation of the teachings of Pedersen (C) in the system disclosed by Barret (A) would have been considered an obvious alternative in the design of the algorithm of the system. The artisan would have been motivated to make the modification because such implementation would have improved patient retention in digital health interventions.
Claims 2-6, 8-12, and 13-16 are rejected under 35 U.S.C. 103 as being unpatentable over Barrett et al. (US 2020/0058403A1; hereinafter “Barrett (A)”) in view of Pendersen et al. (“Predicting Dropouts from an Electronic Health Platform for Lifestyle Interventions: Analysis of Methods and Predictors”; hereinafter “Pedersen (C)”), as discussed above, and in further view of Van Sickle (US 2020/0188613A1; hereinafter “Van Sickle (D)) and Greene et al. (A novel statistical method for assessing effective adherence to medication and calculating optimal drug dosages; hereinafter “Greene (B)”).
In relation to claims 2-6 and 8-12, the additional limitations of dependent claims 2-6 and 8-12 are also taught by the cited prior art:
Claim 2/8: The use of rescue and controller inhalers is explicitly disclosed in Barrett (A)
(e.g., [0096], [0098]) and is the subject of the study in Greene (B) (see Abstract; subsections titled “Objective” and “Conclusions”).
Claim 3/9: The integration of device sensors with mobile applications for data collection is taught by Barrett (A) (e.g., [0024], [0033]) and Van Sickle (D) (e.g., [0017], [0036], [0037], and [0038]).
Claim 4/10: While not explicitly taught, it would have been obvious to one of ordinary skill in the art to include customer service data in the training data, as it is a well-known source of patient feedback and behavior relevant to predicting treatment discontinuation or churn.
Claim 5/11: Barrett (A) discloses alerts [notifications] for patients and healthcare providers (e.g.,
[0054], [0157]). Extending this notification system to include manufacturers for product
improvement, quality control, and aggregate data analysis would be an obvious and
logical design choice for a person of ordinary skill in the art.
Claim 6/12: Greene (B) explicitly teaches quantifying the impact of various input factors
on predicted outcomes, stating that their method results in a "48% reduction in odds of
an exacerbation per standard deviation increase in adherence score (Odds Ratio = 0.52,
p = 0.002)" (Greene (B), p. 10 of 17; section titled “Predictive Validity of new Adherence model”). It would be obvious to include such a quantifiable impact score in the prediction endpoint for better interpretability and clinical utility.
Based on the above teachings, the limitations disclosed in claims 2-6 and 8-12 were well-known in the art at the time the invention was filed, and therefore, the use or implementation of these enhancements in the invention would have been considered obvious alternatives in the design of the system.
In relation to claim 13, Claim 13 recites a pipeline method for training a machine learning model to predict patient churn. Barrett (A) does not disclose such pipeline method. However, Greene (B) provides a detailed methodology for training a predictive statistical model based on data from a large population of patients using inhaler devices. Greene (B) discloses:
Collecting data inputs from a population of patients, including event data (dose timing
from the INCA™ device) and clinical data (patient outcomes) [see Abstract; Objective];
Curating this data into datasets for training and validation [e.g. page 8 of 17; “In total…]
Training a model that weights various input factors to determine their impact on
predicting adverse events, such as exacerbations [e.g. page 9 of 17; second line].
Greene (B) provides a detailed methodology for training a predictive statistical model
based on data from a large population of patients using inhaler devices. Greene (B) discloses collecting data inputs from a population of patients, including event data (dose timing) and clinical data (patient outcomes), curating this data into datasets for training and validation, and training a model that weights various input factors to determine their impact on predicting adverse events. For example, Greene (B) states:
"The parameters of the model are optimised against patient outcome data using maximum likelihood methods. The model is fitted and validated by secondary analysis of two independent datasets from two remote-monitoring studies of adherence... Training data came from a cohort of asthma patients (~ 47,000 samples from 218 patients)." (Greene, Abstract; Methods, p.1)
"The model was optimised using pre-existing adherence data and measures of clinical outcomes obtained from a clinical trial of the INCATM device in an asthma cohort... The training set consisted of up to three months of adherence and clinical data from 218 otherwise healthy patients..." (Greene, pp. 7-8, section titled “Datasets”)
Greene (B) does not, however, explicitly apply this training pipeline to the prediction of churn. As established previously, Pedersen (C) teaches the use of machine learning models trained on population data to predict patient dropout or churn. Therefore, based on the above teachings, for an artisan skilled in the art, it would have been obvious to apply the detailed training pipeline methodology of Greene (B) to the problem of churn prediction as taught by Pedersen (C). The artisan would have been motivated to combine the teachings because such combination would have leveraged a proven data-driven modeling approach from the field of inhaler adherence to address the related and critical problem of patient treatment discontinuation.
In relation to claims 14-16, the additional limitations of dependent claims 14-16 are also taught by the prior art or represent obvious variations:
In relation to claim 14, as discussed above, data refinement and conversion are standard necessary steps in any machine learning pipeline, as implicitly taught by Greene (B).
In relation to claim 15, as discussed above, the inclusion of mobile device and application data is taught by Barrett (A) and Van Sickle (D), and including customer service data is an obvious extension of this well-known inclusion capability.
In relation to claim 16, as discussed above, the execution of a stored, trained model for individual prediction is the fundamental purpose and an inherent final step of such a system, as disclosed by Barrett (A).
Based on the above teachings, the limitations disclosed in claims 14-16 were well-known in the art at the time the invention was filed, and therefore, the use or implementation of these enhancements in the invention would have been considered obvious alternatives in the design of the system.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MANUEL A MENDEZ whose telephone number is (571)272-4962. The examiner can normally be reached Mon-Fri 7:00 AM-5:00 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Bhisma Mehta can be reached at 571-272-3383. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
Respectfully submitted,
/MANUEL A MENDEZ/ Primary Examiner, Art Unit 3783