DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Specification
The title of the invention is not descriptive. A new title is required that is clearly indicative of the invention to which the claims are directed.
The following title is suggested: Conference Record Processing with Machine Learning Model to Obtain Action Items
The abstract of the disclosure is objected to because it is more than one hundred and fifty words. Applicants’ abstract is 225 words. 37 CFR 1.72 states that an abstract should preferably not exceed 150 words. A corrected abstract of the disclosure is required and must be presented on a separate sheet, apart from any other text. See MPEP §608.01(b).
The disclosure is objected to because of the following informalities:
The Specification does not include any paragraph numbering or line numbering to facilitate entry of amendments under 37 CFR 1.121, which requires replacement of entire paragraphs. 37 CFR 1.52(b)(6) suggests paragraph numbering in the Specification. Applicants are requested to submit a Substitute Specification with paragraph numbering and/or line numbering.
On page 5, line 28, “precipitation product” does not appear to be a proper idiomatic translation.
Appropriate correction is required.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claim 12 is rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter.
The claim does not fall within at least one of the four statutory categories of patent eligible subject matter because it can be construed as a non-patent-eligible ‘signal claim’. MPEP §2106.03 I. states that transitory forms of signal transmission, i.e., ‘signals per se’, of a propagating electrical or electromagnetic signal or carrier wave do not represent one of the four statutory categories of invention under 35 U.S.C. §101. See Nuijten, 500 F.3d 1346, 84 USPQ2d 1495 (Fed. Cir. 2007). Independent claim 12 sets forth “a computer-readable storage medium” which can be broadly construed as a ‘signal claim’. The USPTO takes the position that claims directed to a computer-readable medium can be broadly construed to include non-patent-eligible interpretations. Here, a computer-readable storage medium could be construed as transitory storage of a signal on an optical fiber or on a wire. The Specification does not provide any express disclaimer that a computer-readable storage medium should not be construed to include transitory embodiments. Applicants can overcome this rejection by amending the claim language to set forth “a non-transitory computer-readable storage medium”.
Claim Rejections - 35 USC § 103
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 to 5, 11 to 12, 14 to 16, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Pruksachatkun et al. (U.S. Patent Publication 2021/0312128) in view of Huang et al. (U.S. Patent No. 11,470,279).
Concerning independent claims 1 and 11 to 12, Pruksachatkun et al. discloses a method, system, and computer-readable medium for extracting clinical follow-ups, comprising:
“a memory; a processor; and a computer program; where the computer program is stored in the memory and arranged for being executed by the processor to implement the following steps” – a method and system may be deployed through a machine that executes computer software embodied on a computer-readable medium by a processor (¶[0062] - ¶[0063]);
“obtaining a target sentence to be processed in a [conference] record” – a system may receive words or sentences from a medical record (“in a [conference] record”); a system may receive a sentence 204 from the medical record, and may identify if sentence 204, referred to as a focus sentence, is related to an important item; contextual data 202 may be sentences that appear around focus sentence 204 (¶[0025]: Figure 2); here, focus sentence 204 can be construed as “a target sentence to be processed”;
“at least encoding the target sentence by using a trained machine learning model to obtain a representation vector of the target sentence” – a system may receive contextual data 202 and focus sentence 204 and process the process contextual data 202 and the focus sentence 204 using a word embedding model 206; word embedding model 206 may be a trained machine learning model; word embedding model 206 may generate one vector embedding output for each word or a pair of words in the focus sentence; an embedding is a representation of a token in a vector space; system may include a sentence embedding model 208; sentence embedding model 208 may receive the output of the word embedding model 206 and determine sentence embeddings (¶[0026] - ¶[0029]: Figure 2);
“determining a probability value containing an action item in the target sentence according to the representation vector of the target sentence” – a system may output a multi-label score vector 210 that identifies if the sentence 204 is related to an important item and may also identify one or more categories of important items (¶[0025]: Figure 2); multi-label score vector 212 may be a confidence score relating to how likely the focus sentence relates to an important item or actionable item (“an action item in the target sentence according to the representation vector of the target sentence”) (¶[0030]: Figure 2); a method may further include identifying clinically actionable items in the written medical records based on the labeled sentences 310 (¶[0032]: Figure 3); word embeddings may be processed to identify if the focus sentence has actionable content based on a score vector 712 (¶[0052]: Figure 7); a system may receive a focus sentence 904 from the medical record and may output a multi-label score vector 910 that identifies if the focus sentence 904 is related to an important item and may also identify one or more categories of important items (¶[0055]: Figure 9); multi-label classifier 908 may be configured to determine a multi-label score vector 910 wherein each value of the score vector 910 identifies a score that provides a measure of how close the focus sentence 904 is to a category of important items that should be emphasized or extracted from a medical record (¶[0056]: Figure 9); here, a score is equivalent to “a probability value” and an actionable item or an important item is “an action item”;
“in response to determining, based on the probability value, that the target sentence contains the action item, obtaining related elements of the action items, the related elements being used for assisting a user in following up on to-do items and organizing a [conference] summary” – discharge summaries (“a [conference] summary”) describing a hospital stay contain crucial information and action items to share with patients and their future caregivers (¶[0017]); medical records including discharge summaries may be processed to identify important items; important items may include follow-up items such as medications, prescriptions, appointments, lab tests, and the like (“following up on to-do items”); important items may be identified and emphasized in the medical record and/or extracted from the medical record; identified important items may be presented to the physician or other relevant party (“obtaining related elements of the action items, the related elements being used for assisting a user in following up on to-do items and organizing a [conference] summary”) (¶[0018]); methods and systems described herein may automatically identify important items in the discharge summary; identified items may be emphasized within the text to allow the physician to quickly see and find the important items in the medical record; identified important items may be highlighted within the discharge summary text; a discharge summary includes important items categorized into three categories: medical-based follow-up 106, appointment-based follow-up 108, and lab-based follow-up 110 (¶[0021]: Figure 1); broadly, a summary is organized around important items to assist a user in following up on a to-do list that includes a medical follow-up, an appointment follow-up, and a lab follow-up.
Concerning independent claims 1 and 11 to 12, Pruksachatkun et al. clearly discloses all of the limitations with the exception of “a conference record”. Arguably, Pruksachatkun et al. discloses this limitation of “a conference record”, too, by determining action items in a medical record with a medical visit being construed as “a conference” between a doctor and a patient. However, even if “a conference record” is not disclosed by Pruksachatkun et al., Huang et al. teaches a similar way of extracting a conference summary from a transcript. (Abstract) Specifically, Huang et al. teaches obtaining a transcript of a conference, wherein the transcript includes strings with respective timestamps, inputting strings from the transcript to a machine learning model to obtain respective scores for the strings, and selecting a string for highlighting from the transcript based on respective scores of strings. (Column 1, Lines 29 to 39) A transcript of a conference is processed with natural language processing, e.g., implemented using a machine learning model with a vector representation of the transcript text. (Column 5, Lines 52 to 59) A classifier predicts whether a sentence includes an action item, and the selection of highlights is based in part on the presence of action items. (Column 6, Lines 11 to 14) A machine learning model (e.g., a PreSumm model) is trained and used to determine respective scores for strings of the transcript. The machine learning model may be used to extract a text summary from the transcript. Strings of the transcript are converted to sentence vectors and pairwise similarity metrics for the sentence vectors are used to determine respective scores for strings of the transcript. (Column 16, Line 58 to Column 17, Line 15: Figure 5: Step 506) Technique 900 includes detecting an action item phrase in a string from the transcript. A selected string may be selected based on presence of the action item phrase. (Column 22, Lines 65 to 67: Figure 9: Step 904) An objective is to address problems of automatically extracting a summary of a conference that avoids tedious and time consuming review of long recordings of audio and video content. (Column 5, Lines 30 to 43) It would have been obvious to one having ordinary skill in the art to summarize a conference record by extracting action items using machine learning and sentence vectors as taught by Huang et al. to extract actions items in clinical notes of Pruksachatkun et al. for a purpose of automatically extracting a summary of a conference that avoids tedious and time consuming review of long recordings of audio and video content.
Concerning claim 2, Pruksachatkun et al. discloses “adding a first preset character to a head of the target sentence” for [CLS] (¶[0057]: Figure 10); “adding a second preset character to a tail of the target sentence” for [SEP] (¶[0057]: Figure 10); “the first preset character, each text unit in the target sentence, and the second preset character being elements in a first set respectively” including text units W7 and W8 for a first line in Figure 10 (¶[0057]: Figure 10);
“inputting a word embedding vector and position information corresponding to each element in the first set and identification information of the target sentence into the trained machine learning model to output a hidden state vector representation corresponding to each element in the first set” – word embeddings model 206 may be a trained machine learning model (¶[0026]: Figure 2); P0 . . . P14 in a third line is “position information corresponding to each element in the first set”; SA in a second line is “identification information of the target sentence”; XCLS . . . X14 is “a hidden state vector representation corresponding to each element in the first set” (¶[0057]: Figure 10);
“setting the hidden state vector representation of the first preset character as the representation vector of the target sentence” – contextual embedding of the separation token XSEP of the focus sentence 1016 may be further passed through a multi-label classifier 1014 to generate labels that categorize the focus sentence; some embodiments provide tokens including embeddings of the [CLS] token XCLS may be used as the sentence-level representation and used as input to the trained multi-label classifier 1014 (¶[0057]: Figure 10); Compare Figure 7 of the Specification.
Concerning claim 3, Pruksachatkun et al. discloses “encoding a previous sentence of the target sentence, the target sentence, and a next sentence of the target sentence by using the trained machine learning model to obtain the representation vector of the target sentence” – input data 1002, 1004, and 1006 includes a left context sentence 1002 (“a previous sentence of the target sentence”), a right context sentence 1004 (“a next sentence of the target sentence”), and a focus sentence 1006 (“the target sentence”), which are input into word embedding model 1008 (“encoding . . . to obtain the representation vector of the target sentence”). (¶[0057]: Figure 10) Compare Figure 8 of the Specification.
Concerning claim 4, Pruksachatkun et al. discloses:
“adding a first preset character to a head of the previous sentence, adding a second preset character to a tail of the next sentence, adding the second preset character between the previous sentence and the target sentence, and adding the second preset character between the target sentence and the next sentence, the first preset character, each text unit in the previous sentence, each text unit in the target sentence, each text unit in the next sentence and the second preset character being elements in a second set respectively” – [CLS] (“a first preset character”) is added to a head of left context sentence 1002; [SEP] (“a second preset character”) is added to the tail of right context sentence 1004; [SEP] (“the second preset character”) is added between left context sentence 1002 and target sentence 1006; [SEP] (“the second preset character”) is added between target sentence 1006 and right context sentence 1004; a first line includes ‘text units’ W’s of left context sentence 1002, W’s of target sentence 1006, and W’s of right context sentence 1004 (“elements in a second set”);
“inputting a word embedding vector and the position information corresponding to each element in the second set, identification information of the previous sentence, identification information of the target sentence and identification information of the next sentence into the trained machine learning model to output a hidden state vector representation corresponding to each element in the second set” – word embeddings model 206 may be a trained machine learning model (¶[0026]: Figure 2); P0 . . . P14 in a third line is “position information corresponding to each element in the second set”; SB in a second line is “identification information of the previous sentence”; SA in a second line is “identification information of the target sentence”; SB in a second line is “identification information of the next sentence”; XCLS . . . X14 are obtained by word embedding model 1008 and is “a hidden state vector representation corresponding to each element in the second set” (¶[0057]: Figure 10);
“setting the hidden state vector representation of the second preset character between the target sentence and the next sentence as the representation vector of the target sentence” – contextual embedding of the separation token XSEP of the focus sentence 1016 may be further passed through a multi-label classifier 1014 to generate labels that categorize the focus sentence; some embodiments provide tokens including embeddings of the [CLS] token XCLS may be used as the sentence-level representation and used as input to the trained multi-label classifier 1014 (¶[0057]: Figure 10); Compare Figure 8 of the Specification.
Concerning claim 5, Pruksachatkun et al. discloses that a system may output a multi-label score vector 210 that identifies if the sentence 204 is related to an important item and may also identify one or more categories of important items (¶[0025]: Figure 2); multi-label score vector 212 may be a confidence score relating to how likely the focus sentence relates to an important item or actionable item (¶[0030]: Figure 2); a method may further include identifying clinically actionable items in written medical records based on the labeled sentences 310 (¶[0032]: Figure 3); word embeddings may be processed to identify if the focus sentence has actionable content based on a score vector 712 (¶[0052]: Figure 7). Here, a focus sentence is “a target sentence”, and determining actionable items in a focus sentence of a medical record is “recognizing at least one of a time word and an action word in the any sentence”. That is, Applicants’ claim language only requires recognizing one of “an action word” or “a time word”, and “an action word” is recognized in an action item by Pruksachatkun et al.
Concerning claim 14, Pruksachatkun et al. discloses extracting action items from discharge summaries which include medical-based follow-up 106, appointment-based follow-up 108, and lab-based follow-up 110. (¶[0021]: Figure 1) These action items to follow up, then, are “performed by a conference-related party after a conference.” That is, a patient is “conference-related party”.
Concerning claim 15, Pruksachatkun et al. discloses “determining the probability value containing the action item in the target sentence according to the representation vector of the target sentence” – system may output a multi-label score vector 210 that identifies if the sentence 204 is related to an important item and may also identify one or more categories of important items (¶[0025]: Figure 2); multi-label score vector 212 may be a confidence score relating to how likely the focus sentence relates to an important item or actionable item (¶[0030]: Figure 2);
“and position information of the target sentence of the conference record” – input may further include position data associated with each word and token P0 to P14 (¶[0057]: Figure 10). Specifically, position information P7 to P9 represents “position information of the target sentence”.
Concerning claim 16, Pruksachatkun et al. discloses “three training stages containing a first training stage, a second training stage and a third training stage” –training of models 510 may be based on training data 508 that includes medical records labeled with actionable content; training may include training of one or more of the multi-label classifier 502, sentence embedding model 504, and/or the word embedding model 506; word embedding model 506 may be initialized with a pretrained model; multi-label classifier 502 and the sentence embedding model 504 may be initialized with random parameters; training may result in a trained multi-label classifier 514, trained sentence embedding model 516, and a fine-tuned word embedding model 518 (¶[0038]: Figure 5). Here, three training stages may be construed as training three models 514, 516, and 518.
Concerning claim 18, Pruksachatkun et al. discloses that training of models 510 may be based on training data 508 that includes medical records labeled with actionable content (“wherein sample data used in the second training stage is the conference record or a conference text”). (¶[0038]: Figure 5) User interactions with medical records that include identified important items may be tracked and used as training data. (¶[0041]: Figure 5) Training data may be manually annotated discharge summaries. (¶[0042]: Figure 5)
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Pruksachatkun et al. (U.S. Patent Publication 2021/0312128) in view of Huang et al. (U.S. Patent No. 11,470,279) as applied to claims 1 and 5 above, and further in view of Shah et al. (U.S. Patent Publication 2023/0161964).
Pruksachatkun et al. discloses actionable items based upon words (“recognizing at least one of the time word and the action word in the any sentence”), but does not disclose “in response to the any sentence not containing any sensitive word and/or a length of the any sentence meeting a preset condition, recognizing at least one of the time word and the action word in the any sentence.” However, Shah et al. teaches extracting topics from text content, and generating a set of sentences based on a document further comprises filtering the set of sentences based on a pre-configured length. (¶[0006]) A method can filter sentences based on the length of the sentence. Sentences that have words less than four, or greater than one hundred words, can be discarded. Since this type of information can be obtained from a structured text and is not relevant to deep topic extraction, hence in some embodiments, the method can discard them. (¶[0033]) Shah et al., then, teaches that a sentence is discarded so that it is not used in extracting topics (“in response to the any sentence . . . meeting a preset condition”). An objective is only extract information relevant to topic extraction. It would have been obvious to one having ordinary skill in the art to only recognize action items in Pruksachatkun et al. for sentences having a length that meets a preset condition as taught by Shah et al. for a purpose of only extracting information relevant to topic extraction.
Claims 8 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Pruksachatkun et al. (U.S. Patent Publication 2021/0312128) in view of Huang et al. (U.S. Patent No. 11,470,279) as applied to claims 1 and 5 above, and further in view of Hirabayashi et al. (U.S. Patent Publication 2020/0143803).
Concerning claim 8, Huang et al. teaches that a transcript includes strings that are associated with respective timestamps and a conference summary is then generated using timestamps from the transcript associated with strings (e.g., sentences) that have been selected for inclusion in the conference summary. (Abstract) However, Huang et al. does not include “related elements of the action item comprise: time information of the action item, the time information comprises: a time word in the target sentence and a timestamp corresponding to the time word.” Hirabayashi et al. teaches importance degree determination means for recognizing details in a voice acquired by voice acquisition means. (Abstract) Specifically, the keyword database 80 includes a time word database 81, a purpose word database 82, an action word database 83, and an importance degree word database 84. Time word database 81 stores a word related to a time, which is referred to as a time word. The time word includes words capable of specifying a time such as “10 a.m.”, “15:30”, and “1 hour later”. A time word also includes words indicating time constraints such as “immediately”, “right now”, “as soon as possible (ASAP)”, “quickly”, “in the daytime”, “tonight”, and “until tomorrow”. Action word database 83 stores a word related to an action, which is referred to as an action word. Action words include words representing details of operations to be performed by store clerks in charge of commodity delivery such as “replenish” and “arrange”. Importance degree word database 84 stores a word related to a degree of importance, which is referred to as an importance degree word. The importance degree word includes words having urgency, such as “immediately”, “right now”, “as soon as possible (ASAP)”, and “quickly”, among words indicating time constraints. (¶[0055] - ¶[0058]: Figure 6) Hirabayashi et al., then, teaches “related elements of the action item comprise: time information of the action item, the time information comprises: a time word in the target sentence”. An objective is to accurately give an indication to an unspecified operator according to a degree of importance. (¶[0019] - ¶[0022]) It would have been obvious to one having ordinary skill in the art to recognize related elements of an action item including time information of an action item as taught by Hirabayashi et al. with strings of words associated with timestamps in a conference transcript of Huang et al. for a purpose of giving an accurate indication of a degree of importance to an operator.
Concerning claim 20, Pruksachatkun et al. discloses that each value of the score vector 212 identifies a score that provides a measure of whether focus sentence 204 belongs to a category of important items that should be emphasized or extracted from a medical record; multi-label score vector 212 may be a confidence score relating to how likely the focus sentence relates to an important item or actionable item; a threshold value for each element of the vector may be used to determine if the focus sentence should be classified as an important item (“determining that any sentence is the target sentence is to be processed”) (¶[0030]: Figure 2); word embeddings may be processed to identify if the focus sentence has actionable content based on a score vector 712 (“in response to any sentence containing . . . the action word”) (¶[0052]: Figure 7). Hirabayashi et al. teaches importance degree determination means for recognizing details in a voice acquired by voice acquisition means. (Abstract) Keyword database 80 includes a time word database 81, a purpose word database 82, an action word database 83, and an importance degree word database 84. (¶[0055] - ¶[0058]: Figure 6) Hirabayashi et al., then, teaches recognizing an importance degree based upon “both the time word and the action word”.
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Pruksachatkun et al. (U.S. Patent Publication 2021/0312128) in view of Huang et al. (U.S. Patent No. 11,470,279) as applied to claim 1 above, and further in view of Farzindar (U.S. Patent Publication 2008/0104506).
Huang et al. teaches generating conference summaries with elements of action items, but does not specifically provide “wherein the conference summary comprises: . . . a conference issue, a conference conclusion, and a question discussed in the conference” However, Farzindar teaches producing a document summary of a court judgment that presents the arguments of each party relating to each issue, ‘issues’, which identifies the questions of law addressed by the court, ‘judicial analysis’, which state the reasoning and jurisprudence used by the judge to arrive to his conclusion, and ‘conclusion’, which expresses the final decision of the court. (¶[0052]) An objective is to produce a document summary from a document that reduces human time to provide manual summaries which may include a risk that a legal expert may misinterpret a judgment. (¶[0004]) It would have been obvious to generate a conference summary with elements of action items in Huang et al. to include a conference issue, a conference conclusion, and a question discussed in the conference as taught by Farzindar for a purpose of reducing human time and risk of misinterpretation to produce a document summary.
Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Pruksachatkun et al. (U.S. Patent Publication 2021/0312128) in view of Huang et al. (U.S. Patent No. 11,470,279) as applied to claims 1 and 16 above, and further in view of Jin et al. (U.S. Patent Publication 2021/0216880).
Pruksachatkun et al. discloses that a model may be pretrained on general language sources including Wikipedia articles. (¶[0024] and ¶[0038]: Figure 5) However, Pruksachatkun et al. does not disclose “wherein the training sample data used in the first training step is a Chinese written text.” Here, Applicants’ Specification appears to include a limitation of “written” text to contrast with spoken text. Generally, Chinese text would be an obvious alternative language on which to train a model because it is one of the most common of the world’s languages. Specifically, Jin et al. teaches a method for knowledge extraction based on TextCNN that firstly obtains a character vector dictionary and a word vector dictionary. A first training text is Chinese Wikipedia. (¶[0094]) An objective is to perform knowledge extraction in a manner that is not cumbersome and does not consume a great quantity of resources. (¶[0004]) It would have been obvious to one having ordinary skill in the art to pretrain a model from articles in Wikipedia in Pruksachatkun et al. with written text of Wikipedia in Chinese as taught by Jin et al. for a purpose of performing knowledge extraction in a manner that is not cumbersome and does not require a great quantity of resources.
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Pruksachatkun et al. (U.S. Patent Publication 2021/0312128) in view of Huang et al. (U.S. Patent No. 11,470,279) as applied to claims 1 and 16 above, and further in view of Brinker et al. (U.S. Patent Publication 2008/0010226).
Pruksachatkun et al. discloses training a multi-label classifier of actionable items. (¶[0024] and ¶[0038]: Figure 5) However, Pruksachatkun et al. does not disclose “wherein sample data used in the third training stage is annotation data containing a sentence containing the action item and a sentence that does not contain the action item, the sentence containing the action item being marked as a positive example, and the sentence that does not contain the action item being marked as a negative example.” Here, training with annotation data for an actionable item marked as a positive example or a negative example is equivalent to training a binary classifier. Specifically, Brinker et al. teaches multilabel classification (MLC) that associates with an instance x of a predefined set of class labels into relevant (e.g., positive) and irrelevant (e.g., negative) labels. (¶[0005]) Binary relevance learning (BR) trains a separate binary model for each label using all examples Px as positive examples and all examples Nx as negative examples. (¶[0008]) Multilabel ranking (MLR) is understood as learning a model that associates with a query input x of a label set L into relevant (e.g., positive) and non-relevant (e.g., negative) labels. (¶[0025]) An objective is to provide a ranking among the states that can used to specify the order in which an operator should solve problems or process a list of action items. (¶[0020]) It would have been obvious to one having ordinary skill in the art to train a model to classify action items in Pruksachatkun et al. with positive examples and negative examples as taught by Brinker et al. for a purpose of providing a ranking that can be used to specify an order in which an operator should solve problems or process a list of actions.
Allowable Subject Matter
Claims 7 and 9 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
The prior art made of record and not relied upon is considered pertinent to Applicants’ disclosure.
Siohan et al. discloses related prior art directed to extracting action items.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MARTIN LERNER whose telephone number is (571) 272-7608. The examiner can normally be reached Monday-Thursday 8:30 AM-6:00 PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Richemond Dorvil can be reached at (571) 272-7602. 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.
/MARTIN LERNER/Primary Examiner
Art Unit 2658 July 6, 2026
Prosecution Insights