Dynamic guided physician-patient interaction engine

Abstract

Devices and associated methods relate to dynamically directed patient-provider interaction engines (PPISs). In one illustrative example, a PPIS can include, for example, an inference engine (IEN). The IEN can generate, for example, a context-specific body of knowledge (BOK). The PPIS can include, for example, a discrepancy engine (DEN). The DEN can dynamically identify discrepancies between one or more BOKs and a particular case, event, entity, and / or context-specific BOK. For example, in some embodiments, the DEN can generate a success score comparing a physician’s particular case and / or track record to a BOK (e.g., state-of-the-art practice, other physicians). For example, various embodiments can advantageously identify areas of success that contribute to the provider and corresponding actions and / or areas of improvement and corresponding suggested actions.

Description

[0001] Cross-references to related applications

[0002] This application claims the benefit of U.S. Provisional Application Serial No. 63 / 590,191, entitled “Dynamic Guided Physician-Patient Interaction Engine”, filed October 13, 2023, by Samuel Robert Browd et al.

[0003] The subject matter of this application may share the same inventors as the subject matter of the following applications and / or may be related to the subject matter of the following applications:

[0004] • U.S. Patent Application Serial No. 18 / 313,249, entitled “Dynamically Controlled Cerebrospinal Fluid Shunt”, filed on May 5, 2023 by Samuel Robert Browd et al.

[0005] • U.S. Patent Application Serial No. 63 / 365,407, entitled “Distributed Sensing and Control of Cerebrospinal Fluid,” filed on May 26, 2022, by Samuel Robert Browd et al.

[0006] • U.S. Application Serial No. 63 / 477,158, entitled “Central Nervous System Monitoring and Intervention”, filed on December 23, 2022 by Samuel Robert Browd et al.

[0007] • U.S. Application Serial No. 63 / 477,162, entitled “Cerebrospinal Fluid Polarization”, filed on December 23, 2022 by Samuel Robert Browd et al.

[0008] • U.S. Patent Application Serial No. 63 / 488,412, entitled “Dynamic Shunt Systems”, filed on March 3, 2023 by Samuel Robert Browd et al.

[0009] • U.S. Application Serial No. 63 / 364,253, filed May 5, 2022, by Samuel Robert Browd, entitled “Shunt Technology and the Potential for a Smart Shunt”; and

[0010] • U.S. Application Serial No. 63 / 590,313, entitled “Neurosurgical Devices and Methods,” filed on October 13, 2023, by Samuel Robert Browd et al.

[0011] This application incorporates the entire contents of the foregoing application and all applications claiming priority and / or benefits herein by reference. Technical Field

[0012] The various implementation schemes generally involve medical devices and / or methods. Background Technology

[0013] Artificial intelligence (AI) has significantly impacted the medical field, transforming all aspects of patient care and clinical procedures. One prominent application of AI is in diagnostic imaging and analysis. AI algorithms are already being used to process and interpret medical images. For example, AI-driven imaging tools can assist radiologists by identifying areas of interest, accelerating the diagnostic process and reducing the workload of healthcare professionals.

[0014] Virtual patient assistance leverages advanced technologies to enhance patient healthcare by providing accessible, real-time support and information. For example, these systems can use AI-powered chatbots and / or virtual assistants to answer patient inquiries, schedule appointments, provide medication reminders, and offer health advice based on the patient's medical history and symptoms. Patients can also use general-purpose search and AI tools to learn about medical topics on their own. Summary of the Invention

[0015] The apparatus and associated methods involve a dynamically guided patient-provider interaction engine (PPIS). In an exemplary example, the PPIS may, for example, include an inference engine (IEN). The IEN may, for example, generate a context-specific knowledge body (BOK). The PPIS may, for example, include a difference engine (DEN). The DEN may, for example, dynamically identify differences between one or more BOKs and specific case, event, entity, and / or context-specific BOKs. For example, in some embodiments, the DEN may generate success scores that compare physician-specific cases and / or records with BOKs (e.g., state-of-the-art practices, other physicians). For example, various embodiments may advantageously identify areas that contribute to the provider's success and corresponding actions and / or areas for improvement and corresponding recommended actions.

[0016] Various implementation schemes can achieve one or more advantages. For example, some implementation schemes can advantageously generate referral recommendations based on the provider’s immediate resources, such as available colleagues, thereby providing, for example, actionable recommendations.

[0017] Some implementation schemes can, for example, advantageously facilitate efficient communication channels between patients and healthcare providers, such as, by way of example and not limitation, analyzing patient search history and provider research. Exemplary implementation schemes can, for example, advantageously extract personalized, actionable insights from massive amounts of medical data, thereby improving both patient healthcare services and provider performance.

[0018] Details of the various embodiments are set forth in the accompanying drawings and the following description. Other features and advantages will be apparent from the description and drawings, as well as from the claims. Attached Figure Description

[0019] Figure 1 An exemplary patient-provider interaction system, PPIS, is described.

[0020] Figure 2 Such as Figure 1 The diagram shows an exemplary architectural implementation of PPIS.

[0021] Figure 3 An exemplary inference engine, such as one that can be used in PPIS, is described.

[0022] Figure 4 It describes an exemplary difference engine that can be used in PPIS.

[0023] Figure 5A and Figure 5B An exemplary provider interface, such as the Dynamically Guided Interactive Interface (DGI) generated by PPIS, is depicted.

[0024] Figure 6A An exemplary method, which may involve, for example, generating a knowledge body (BOK) from a patient, is described.

[0025] Figure 6B It describes what can be, for example, by a difference engine such as PPIS (e.g., as at least referenced) Figure 4 (Disclosed) Inspirational methods for applying and / or using context-specific BOKs.

[0026] Figure 7 The provider dashboard interface, such as that generated by PPIS, is described.

[0027] Figure 8 Describing such as at least reference Figure 7 The disclosed exemplary method for generating provider guidelines.

[0028] Figure 9 Describing such as at least reference Figures 3 to 4 The disclosed exemplary method for training the model.

[0029] The same reference numerals in the various figures indicate the same elements. Detailed Implementation

[0030] To aid understanding, this document is organized as follows. First, to help introduce the discussion of various implementation schemes, refer to... Figure 1 This section introduces the Dynamic Patient-Provider Interaction System (PPIS). Secondly, this introduction leads to a reference... Figures 2 to 4 A description of an exemplary architecture for an example implementation of PPIS and its associated engine. Third, refer to... Figures 5A to 8 Fourth, refer to the exemplary interface and associated methods. Figure 9 The discussion then shifts to exemplary model training methods. Finally, this document discusses additional implementations, exemplary applications, and aspects related to the generation of dynamic patient-provider interaction guidance.

[0031] Figure 1 An exemplary patient-provider interaction system (PPIS 105) is described. PPIS 105 can, for example, advantageously coordinate knowledge independently accessed by patients and / or healthcare providers (e.g., physicians). PPIS 105 can, for example, advantageously facilitate targeted interpretation of knowledge accessed by patients by physicians.

[0032] In the depicted example, a patient accesses knowledge via one or more patient devices 125. For example, one or more patient devices 125 may submit queries (e.g., to one or more search engines). Actions of one or more patient devices 125 may, for example, be transmitted to a patient-doctor interaction engine (PPIE 110). PPIE 110 may, for example, receive questions submitted by the patient and / or results accessed by the patient. For example, PPIE 110 may (e.g., via inference engine 115) generate inferences from patient data. For example, PPIE 110 may infer patient intent, patient questions, patient concerns, and / or patient symptoms. For example, PPIE 110 may (e.g., at a specific point in time) operate inference engine 115 to determine the patient knowledge body (BOK) and / or (e.g., at a specific point in time) infer the patient's state of understanding.

[0033] PPIE 110 can, for example, generate suggestions and transmit them to one or more provider devices 120 of a care provider (e.g., a doctor, nurse, or consultant) who will be seeing the patient. For example, PPIE 110 can summarize data from searches performed by the patient and / or accessed by the patient for the doctor. For example, PPIE 110 can infer the most important data for the patient (e.g., based on subsequent searches, based on dwell time, based on social network activity, and / or messaging activity). For example, PPIE 110 can be operatively coupled to one or more models 130, such as, for example, machine learning (ML) models and / or artificial intelligence (AI) models (e.g., external, internal). For example, PPIE 110 can be operatively connected to one or more external large language models (LLMs), such as… Figure 1 The location of model 130 is depicted in the illustrative diagram. PPIE 110 can, for example, generate a summary of information accessed by the patient and / or conclusions the patient may have reached for the physician. PPIE 110 can, for example, generate a summary of potential symptoms and / or patient history (e.g., activities, events, family history, environmental history) based on patient data and / or additional data (e.g., third-party data sources). For example, difference engine 140 and / or inference engine 115 may include operations for generating questions to be “outsourced” to one or more models 130.

[0034] PPIE 110 can generate recommendations for physicians, for example, based on inferences from patient data. For example, PPIE 110 can generate recommended literature for physician review (e.g., from scientific literature, medical sources). For example, PPIE 110 can generate summaries of potential differences, treatments, and / or educational materials (e.g., from recommended literature). For example, PPIE 110 can generate literature recommendations that can help physicians answer questions (e.g., based on patient questions inferred from patient data).

[0035] Inference engine 115 can, for example, infer the provider's current BOK and / or understanding state. Difference engine 140 can, for example, generate differences between the current provider's BOK and / or understanding state and relevant BOKs (e.g., standard medical services, new research, patient details). Inference engine 115 can, for example, generate recommendations to the provider based on the identified differences.

[0036] In some implementations, for example, the results of patient and / or physician queries, inferences, and / or suggestions can be labeled (e.g., by the patient, by the physician) as valid or invalid. Labels can be used, for example, to train a model (e.g., the intersection of patient and physician labels can be used to train on valid and / or invalid results).

[0037] In some implementations, by way of example and not limitation, the labels may be binary (e.g., true / false, helpful / unhelpful, relevant / irrelevant). In some implementations, the labels may, for example, be rated (e.g., 1-5, 1-10, -3 to +3). In some implementations, the labels may, for example, include emotional responses (e.g., "like it", "sad"). In some examples, the labels may include free text.

[0038] Tags can be used, for example, to rate references and / or outputs (e.g., suggestions, inferences) and / or to train models. In some implementations, tags can be used, for example, to rate users (e.g., rating patient knowledge based on doctor tags, rating doctor communication skills based on patient tags). Tag-based training and / or ranking can be performed, for example, based on the source of the tags (e.g., doctor, patient). Training and / or ranking can be performed, for example, based on comparisons with known truths (e.g., "false labels" compared with known true facts). Training and / or ranking can be performed, for example, based on the experience level of the tag initiator.

[0039] In some examples, tags can be generated automatically. For instance, tags can be generated based on their source (e.g., news articles can be tagged as low reliability, while highly cited peer-reviewed articles in medical journals can be tagged as highly reliable). Tags can also be generated based on comparisons with known truth (e.g., news articles compared to recently published medical textbooks). Furthermore, tags can be generated based on the author's level of experience (e.g., publications from board-certified physicians can be ranked as reliable, publications from journalists with multiple accurate articles can be ranked as potentially reliable, and publications from individuals with no apparent medical training or experience can be ranked as potentially unreliable).

[0040] Some implementation schemes can, for example, advantageously enable users (e.g., patients, doctors) to quickly assess (e.g., conduct a preliminary assessment) and / or screen the quality and / or reliability of the presented information.

[0041] In some implementations, the query object 124 (and / or selected pre-determined query) received from one or more patient devices 125 may include one or more of the following:

[0042] How do I interact with my doctor?

[0043] • What is it? (e.g., disease, medical term, treatment)

[0044] What are the treatment options?

[0045] What costs are involved? (For example, in a specific treatment)

[0046] • What does it look like? (e.g., pathology, physiological structure)

[0047] • Who are the experts? (e.g., local, regional, national, international)

[0048] Who treats it?

[0049] • Does a social network exist? (e.g., a support network for a specific condition) → How do I join? (e.g., if a social network exists)

[0050] What are the side effects? (e.g., treatment options)

[0051] How do I explain this to my doctor?

[0052] What do I need to tell my doctor?

[0053] What is my doctor saying?

[0054] Are there any other options?

[0055] What does this mean for me?

[0056] How can I get help?

[0057] • Which products / services are available?

[0058] • How can I live my daily life after this diagnosis / treatment?

[0059] In some implementations, the query object 119 (and / or selected pre-determined query) received from one or more provider devices 120 may include one or more of the following:

[0060] How do I explain this to the patient?

[0061] What are the latest treatment options?

[0062] What are the latest results?

[0063] • Are there any alternative options?

[0064] What kind of lifestyle changes might be helpful?

[0065] What are the side effects of this treatment?

[0066] What else should I tell the patient?

[0067] What are the patient's recent health-related searches and concerns?

[0068] • Based on patient research, what is the patient's understanding of their condition?

[0069] • How can I clarify any misinformation that a patient may be encountering?

[0070] • What specific symptoms or changes have occurred in the patient's condition since the last visit?

[0071] Do I have any other cases similar to this?

[0072] What were the outcomes of those other cases?

[0073] What are the similarities with those other cases?

[0074] What makes this case different from the others?

[0075] • How do other doctors treat these types of cases?

[0076] Figure 2 Such as Figure 1 The diagram illustrates an exemplary architectural implementation of the PPIS. In this example, PPIE 110 includes a processor 205 operatively coupled to a memory module 220. The processor 205 is operatively coupled to a communication module 210. The processor 205 is operatively coupled to one or more memory modules 225.

[0077] As depicted, PPIE 110 can be operatively coupled to one or more models 130. PPIE 110 can, for example, be operatively coupled to a data source 135 (e.g., an external data source). Data source 135 can, for example, include public repositories 260 and / or personal repositories 265. Public repositories can, for example, include medical literature, popular news and / or articles, search engines, device information, product offerings and / or scientific literature. By way of example and not limitation, personal repositories can include electronic health records, personal records, personal query history, personal preferences, and personal device information (e.g., medical devices, wearable devices).

[0078] In this example, PPIE 110 is operatively coupled to data repository 230 (e.g., locally coupled to PPIE 110, remotely coupled). Data repository 230 may include public repository 260 and / or personal repository 265. In this example, data repository 230 includes entity profiles 270. Entity profile 270 may include, for example, patient and / or provider profiles. Entity profile 270 may include, for example, organizational profiles. As depicted, data repository 230 includes associations 275. Associations 275 may include, for example, predetermined (e.g., historical) associations between entities (e.g., entity profiles) and / or (e.g., from personal and / or public repositories) data objects.

[0079] Data repository 230 may include, for example, BOK 280. BOK 280 may include, for example, pre-determined BOK data objects and / or related data objects (e.g., BOK summary, BOK difference, natural language summary such as BOK and / or difference) (e.g., already generated by inference engine 115).

[0080] In some implementations, a public repository may include, for example, anonymized electronic health record (EHR) data. In some implementations, a personal repository may include EHR records that can be associated with patient users and / or healthcare provider users.

[0081] The exemplary public repository 260 may include, as an example and not a limitation:

[0082] • Medical literature databases (e.g., PubMed)

[0083] • News database (e.g., medical service news, weather news, environmental news, scientific discovery news, business news, and news that may be relevant to users)

[0084] • Popular science articles and / or discussions (e.g., blogs, forums, popular science news websites)

[0085] • Treatment and / or treatment outcome database

[0086] • Clinical research (e.g., ongoing)

[0087] • Regulatory databases (e.g., the U.S. Food and Drug Administration)

[0088] •Product Supply

[0089] An exemplary personal repository 265 may be included, as an example and not a limitation:

[0090] • Non-anonymized EHR

[0091] • Wearable device data (e.g., data from fitness trackers, smartwatches, implantable and / or wearable medical device sensors and / or actuators).

[0092] • Home automation data / account

[0093] • Device account (e.g., smartphone account)

[0094] • Service Provider Account

[0095] • Personal search history

[0096] •Personal information in clinical studies

[0097] • Private sale offers (e.g., special discounts)

[0098] For example, the pricing engine 250 can be configured to present advertisements for device options (e.g., fitness trackers, scales, gym equipment, home medical devices) based on patient and / or healthcare provider queries and / or responses. The pricing engine 250 can also be configured to present advertisements for services (e.g., gym memberships, instructor memberships, consultation memberships, healthcare providers) based on patient and / or healthcare provider queries and / or responses. The pricing engine 250 can also be configured to present medications and / or supplements based on patient and / or healthcare provider queries and / or responses (e.g., prescriptions).

[0099] For example, PPIE 110 can apply one or more self-trained models (e.g., unsupervised classifiers, neural network models) to received queries, responses, and / or other data (e.g., EMR data, literature) to generate associations between patient attributes, product attributes, healthcare provider attributes, condition (e.g., disease, symptom) attributes, and / or treatment option attributes, as examples and not limitations. These associations can be determined dynamically, for example. In some implementations, associations can be stored. Confidence levels can be increased or decreased, for example, based on the number of associations made in the training data pool (e.g., associated with stored associations). In some implementations, data can be divided into training data and test data pools. In some implementations, training data can include received continuous data (e.g., “online,” real-time data).

[0100] PPIE 110 may include, for example, a chat engine 245. Chat engine 245 may advantageously connect doctors and patients. For example, chat engine 245 may allow patients to ask questions of doctors. Chat engine 245 may generate suggestions for doctors, for example, based on patient queries.

[0101] PPIE 110 may include, for example, a predictive analytics engine 235. For instance, predictive analytics engine 235 may generate predictions of treatment outcomes. The output may, for example, be incorporated into one or more BOKs in a BOK 280.

[0102] In some implementations, by way of example and not limitation, the predictive analytics engine 235 can predict synergistic effects. For example, the predictive analytics engine 235 can predict synergistic effects in the delivery sequence and / or timing of multiple therapies (e.g., based on circadian rhythms, events, or physiological attributes). Multiple therapies may include, for example, pharmacological therapies. Multiple therapies may include, for example, phototherapy. Multiple therapies may include, for example, cell therapy. Multiple therapies may include, for example, electrotherapy. Multiple therapies may include, for example, mechanical therapies (e.g., fluid delivery, filtration, adjustment of force / pressure application).

[0103] PPIE 110 may include, for example, a notification engine 240. Notification engine 240 may generate notifications and / or suggestions to other caregivers (e.g., other doctors), for example, based on communications between the patient and the doctor. For example, a patient may contact a doctor due to a heart problem. The patient may also mention, for example, difficulty breathing. A notification to the patient's pulmonologist may be generated. Suggested literature and / or actions may be generated, for example. A summary of possible symptoms may be generated. In some implementations, for example, a medication prescribed by a doctor may be used to generate notifications to other doctors and / or pharmacists (e.g., based on potential interactions and / or side effects) based on scientific literature.

[0104] In some implementations, for example, patient information collected by PPIE 110 and / or (e.g., physician actions and / or responses provided to patients) may be stored in a repository. For example, anonymized data may be stored in a public repository 260. Personally identifiable data may be stored in one or more personal repositories in a personal repository 265.

[0105] In some implementations, for example, the PPIE 110 can be connected to one or more devices. Example devices include those disclosed at least with reference to the following:

[0106] • U.S. Patent Application Serial No. 18 / 313,249, filed on May 5, 2023, by Samuel Robert Browd et al., entitled "Dynamically Controlled Cerebrospinal Fluid Shunt". Figures 1 to 4 (For example, including diversion equipment and related equipment);

[0107] • U.S. Patent Application Serial No. 63 / 365,407, filed on May 26, 2022, by Samuel Robert Browd et al., entitled "Distributed Sensing and Control of Cerebrospinal Fluid". Figures 1 to 9 (For example, including brain shunt devices, CSF ion exchange devices, and phototherapy devices).

[0108] • U.S. Patent Application Serial No. 63 / 477,158, entitled "Central Nervous System Monitoring and Intervention," filed on December 23, 2022, by Samuel Robert Browd et al. Figure 1 (For example, "shunt 105" and associated devices);

[0109] • U.S. Patent Application Serial No. 63 / 477,162, filed on December 23, 2022, entitled “Cerebrospinal Fluid Polarization”, with paragraphs [0003-0040] (e.g., cerebrospinal fluid (CSF) polarization system (CSFPS) and associated device).

[0110] • U.S. Patent Application Serial No. 63 / 488,412, entitled "Dynamic Shunt Systems," filed on March 3, 2023, by Samuel Robert Browd et al. Figures 1 to 3 2 (e.g., intelligent shunts, non-invasive shunt diagnostic devices / systems, implantable devices); and

[0111] • PCT application serial number PCT / US2024 / 051288 filed on October 14, 2024, entitled “Neurosurgical Devices and Methods” (e.g., ventricular implants and / or spinal implants, such as “stent 505”, “capsule 605”, “structural reinforcement module 210”, “shunt 705”, and associated devices).

[0112] The entire contents of these applications are incorporated herein by reference. PPIE 110 may, for example, generate inferences and / or recommendations to physicians based on patient cognitive data (e.g., patient queries) and / or patient device data and / or events. Device data may, for example, be stored in personal and / or public (e.g., after the data has been de-identified) repositories.

[0113] In some implementations, PPIE 110 may, for example, generate recommendations for the patient. For instance, PPIE 110 (e.g., inference engine 115) may recommend (e.g., based on inferences about user preferences, such as via messaging history) that fresh carrots are available and that they are good for the user's condition (e.g., based on medical literature). In some implementations, PPIE 110 may, for example, advise the patient to ask their doctor about changes in values ​​detected by sensors (e.g., in implanted devices and / or monitoring devices such as those discussed above).

[0114] One or more storage modules 225 may include, for example, one or more data agents (e.g., chat agents). As depicted, PPIE 110 includes a patient agent 145 and a provider agent 150. Agents may be personalized, for example, for specific contexts (e.g., type of operation, patient category, specific provider specialty, such as neurosurgeon and / or radiculectomy specialist) and / or entities (e.g., hospital, provider, patient).

[0115] Figure 3An exemplary inference engine, such as one that can be used in PPIS, is depicted. In this example, inference engine 115 receives one or more data sources 135. The depicted exemplary data source 135 includes associated data objects 350 (e.g., such as those from association 275). Data source 135 may include, for example, EHR data 305 (e.g., personally identifiable data from personal repository 265, aggregated and / or anonymized data from public repository 260). Data source 135 may include, for example, a document source 310 (e.g., from public repository 260).

[0116] Data source 135 may include, for example, suggestible entity 315. Suggestible entity 315 may include, for example, people (e.g., patients, providers). Suggestible entity 315 may include, for example, organizations. Suggestible entity 315 may include, for example, events. Suggestible entity 315 may include, for example, products. Suggestible entity 315 may include, for example, services. In some implementations, suggestible entity 315 may include objects corresponding to entities that can be suggested by, for example, inference engine 115.

[0117] For example, data source 135 may include (e.g., historical predictions 320 generated by predictive analytics engine 235) and / or (e.g., historical notifications 325 generated by notification engine 240). Data source 135 may include, for example, historical patient input 330 and / or historical provider input 340 (e.g., input to chat engine 245). Data source 135 may include, for example, BOK 280. For example, data source 135 may include (e.g., historical patient-oriented outputs 340 from patient agent 145) and / or (e.g., historical provider-oriented outputs 345 from provider agent 150).

[0118] Various data sources 135 can be selected, for example, based on the associated data objects 350 between data objects in the source. As an illustrative example, historical inputs and / or outputs can be selected, not limited to, based on patients, suggestible entities, literature, and / or BOK.

[0119] Inference engine 115 may operate on one or more data sources 135 to generate one or more outputs. Inference engine 115 may include, for example, one or more models (e.g., machine learning models). Inference engine 115 may be operatively coupled to one or more models (e.g., one or more models 130). Models may include, for example, LLM, classifiers, deep learning models, general purpose trainers (GPT), and / or graph (e.g., graph artificial intelligence (AI)) models.

[0120] As depicted, the output of inference engine 115 may include, for example, BOK 280. BOK 280 may include, for example, a context-specific BOK. Context-specific BOKs may be generated, for example, based on a specific entity, EHR data, event data, and / or other combinations of specific selections from data source 135.

[0121] The output of inference engine 115 may include, for example, context inference 370. Inference 370 may include, for example, context-aware inferences (e.g., inference that the current event is surgery for a specific patient). Inference 370 may also include, for example, inferences specific to a given context (e.g., the patient does not understand specific concepts related to the upcoming surgery).

[0122] The output of inference engine 115 may include, for example, patient-oriented output. For instance, inference engine 115 may be implemented as a patient agent 145 and / or operationally communicating with that patient agent. The output of inference engine 115 may include, for example, provider-oriented output 360. For instance, inference engine 115 may be implemented as a provider agent 150 and / or operationally communicating with that provider agent. As shown, the output (e.g., any output in the output) may be historical input, such as that shown regarding 355 / / and provider-oriented output 360. In some implementations, for example, the output data may be training data (e.g., for new and / or updated models).

[0123] Figure 4 An exemplary diff engine, such as one that can be used in PPIS, is depicted. In this example, the diff engine 140 can receive input from one or more data sources 135. Data source 135 may include, for example, at least a reference... Figure 3 The disclosed exemplary source. Inputs may include, for example, BOK 280. Inputs may include, for example, inference 370.

[0124] Various inputs can be selected, for example, based on the association between data objects 350 in the source. As an illustrative example, historical inputs and / or outputs can be selected, without limitation, based on patients, providers, another suggestible entity, literature, inferences, and / or BOK.

[0125] The difference engine 140 can operate on one or more data sources in data source 135 to generate one or more outputs. In this example, the output of the difference engine 140 may include, for example, a difference object 410.

[0126] The difference engine 140 may include, for example, one or more models (e.g., machine learning models). The difference engine 140 may be operatively coupled to one or more models (e.g., one or more models 130). Models may include, for example, LLMs, classifiers, deep learning models, general-purpose trainers (GPTs), and / or graph (e.g., graph artificial intelligence (AI)) models. Models may be specifically trained and / or fine-tuned to generate differences between inputs. In some implementations, the difference engine 140 may include one or more models specifically trained to generate and / or select (e.g., pre-determined) queries that are structured to guide models (e.g., general-purpose models, context-specific models) to generate detailed and / or summary definitions (e.g., natural language, vectorized) of differences between two or more inputs.

[0127] As depicted, the output of the difference engine 140 may include, for example, a BOK 280. The BOK 280 may include, for example, a BOK updated (e.g., enhanced) with information about the difference.

[0128] In some implementations, PPIE 110 may include one or more inference engines 115 and / or difference engines 140. For example, the output of one engine may be the input to the same engine and / or one or more other engines. Engines may be daisy-chained, for example. In some implementations, engines may be dynamically linked, such that a particular engine may receive input and provide output to a different model, for example, based on inferences made (e.g., context-specific).

[0129] Figure 5A and Figure 5B An exemplary provider interface, such as a dynamically guided interactive interface (DGI) generated by PPIS, is depicted. For example, the exemplary interface can be generated by one or more steps of method 600 and / or method 601. In this example, the provider interface is generated in response to an event (e.g., a patient visit). This includes EHR data (e.g., patient vital signs information, event information, patient complaints). Based on the EHR data and patient query information, the patient's BOK is derived. The physician's BOK is inferred. Differences between the patient's and provider's BOKs are determined. For example, it can be determined that the physician is unaware that the patient has received a specific summary of migraines. The physician may not know that the patient has been searching for allergies related to migraines. The physician may not know that the patient has been searching for wildflowers in a specific location. The physician may not know that certain allergens are currently blooming in that specific location. PPIE 110 (e.g., provider agent 150) can generate a summary based on the inferred differences, enabling the provider to collaboratively narrow key differences between the patient's and physician's understanding states within the specific context of the patient's visit.

[0130] Figure 6A An exemplary method, such as generating a knowledge body (BOK) from a patient, is described. Method 600 can be performed, for example, by one or more components of PPIE 110. In this example method, a query is received from a user in step 605. The user can be, for example, a patient. The query can be received, for example, in chat engine 245 and / or by patient agent 145. In step 610, a discovery result is generated and presented to the user (e.g., via patient agent 145, such as via operational inference engine 115 and / or difference engine 140). In step 615, the query is stored in, for example, a personal repository 265. As shown, for example, some embodiments may include (e.g., in personal repository 265 and / or public repository 260) storing the discovery result and / or generating an association data object 350 between the query and (e.g., one or more data sources already stored in data source 135) the discovery result (e.g., stored in association 275). Such embodiments can advantageously reduce unnecessary data duplication, for example.

[0131] Figure 6B It describes what can be, for example, such as by PPIS (e.g., as at least referenced) Figure 4 The disclosed differential engine applies and / or uses context-specific BOKs as an exemplary method. In this example, method 601 includes determining events related to the user (e.g., patient) and provider in step 620. In step 625, query and / or discovery results associated with the user may be retrieved, for example.

[0132] In step 630, a patient BOK object may be generated, for example, based on an association with a data source (e.g., including public and / or personal repositories). In step 635, (e.g., as...) Figures 5A to 5B (As shown) a summary of the patient's BOK (e.g., a natural language summary) generated by the provider.

[0133] In this example, the provider BOK is determined (e.g., retrieved, generated) in step 640. For example, the provider BOK may be determined based on associations with public repositories and / or personal repositories. In step 645, recommendations may be generated, for example, based on the provider BOK.

[0134] As shown, in step 650 (e.g., by difference engine 140), a difference between the provider's BOK and the patient's BOK is determined. If no difference is identified in step 655, a summary of the patient's BOK can be presented to the provider, for example, in step 660, as shown. In this example, when a difference is identified, a highlight and / or summary of the difference can be generated (e.g., by a natural language model) and presented to the provider (e.g., in addition to the summary from step 660).

[0135] Figure 7 An exemplary provider dashboard interface, such as one generated by PPIS, is depicted. The exemplary interface 700 includes an attention dashboard 705. As depicted, the attention dashboard 705 includes identifiers of patients 710 who may require current attention. The attention dashboard 705 includes corresponding success scores (SS 715), such as case SS (CSS). SS can be generated, for example, by PPIE 110 (e.g., via inference engine 115 and / or difference engine 140). The display may include a “More Information” indicator 715. SS 715 may selectively reveal a summary 725, for example. For example, SS 715 and / or summary 725 may be generated based on the patient’s current physical state inferred (by inference engine 115), the patient’s target state inferred (by inference engine 115), and / or the differences inferred between states (e.g., by difference engine 140).

[0136] The exemplary interface 700 includes a patient status display 730. For example, as shown, the patient status display 730 may include indications of the current best case 735 and / or worst case 740. For example, a case may include SS and / or a summary, as shown, such as those disclosed at least with reference to the attention dashboard 705.

[0137] An exemplary interface 700 includes a case matching display 745. The case matching display 745 may include, for example, case matching 750. For example, case matching 750 may include current and / or potential cases that match the provider's skills. For example, the provider's optimal skill set may be inferred (e.g., by inference engine 115). Patient profiles matching such skills may be inferred (e.g., by inference engine 115). Similarity and / or difference summaries may be generated and displayed (e.g., by difference engine 140), as shown. Inference engine 115 and / or difference engine 140 may be further applied to identify and / or summarize patients (e.g., those with positive outcomes) with similar profiles (e.g., historical patients). Similarities and / or differences between cases may be summarized. Thus, some implementations can advantageously guide providers regarding whether cases are well-matched to their skills and help them identify factors that can increase case success.

[0138] An exemplary interface 700 includes a provider advantage display 760. The provider advantage display 760 may, for example, summarize the provider's most successful advantages (e.g., services, operations, procedures, patient populations). The provider advantage display 760 may be generated, for example, by inferring the provider's case, inferring the corresponding SS, and applying a difference engine 140 to determine the similarity and / or difference between the case and / or the provider's BOK.

[0139] An exemplary interface 700 includes a provider improvement display 765. The provider improvement display 765 may, for example, include indications of cases that should be referred. For example, the provider's least successful case types may be inferred (e.g., based on target outcomes, standard medical services, and / or differences between provider outcomes). Differences in the provider's handling of such cases compared to more successful providers may be determined (e.g., by the difference engine 140). A corresponding technical improvement display 770 may be generated. Available referral specialists (e.g., more successful providers) may be identified. The availability of referral specialists (e.g., absolute availability, current availability, such as availability during surgery) may be inferred, and an available specialist display 775 may be generated. In some implementations, corrective actions and / or skill improvements may be inferred, for example, based on differences shown in the corresponding technical improvement display 770. Media (e.g., training, videos, images) may be generated (e.g., for long-term improvement, for immediate correction, such as when unexpected events occur during surgery).

[0140] Figure 8 Describing such as at least reference Figure 7 The disclosed exemplary method for generating provider guidance. As depicted, method 800 may include retrieving provider cases in step 805. In this example, a counter variable (“i”) is initialized to 1 in step 810, and the i-th case is selected in step 815. The BOK associated with this case is generated, for example, based on association with data source 135 (e.g., public repository 260, personal repository 265). In step 820, a target case profile is generated based on the case BOK (e.g., by inference engine 115, such as using input from difference engine 140). In step 825, a difference may be generated between the expected case profile and the case BOK, for example (e.g., by difference engine 140). In step 830, a CSS is generated based on the difference from step 825 (e.g., by inference engine 115).

[0141] If the CSS is determined to be positive (e.g., a good result) in step 835, a summary of the successful contributing factors of the recommendations is generated in step 860 (e.g., by the difference engine 140 and / or the inference engine 115 based on the case BOK and other case BOKs and / or standard medical services). If not, recommended improvement actions are generated in step 845. Improvement actions may be generated, for example, based on the provider BOK, case BOK and / or other provider and / or case BOKs, SS and / or availability. In step 850, a counter variable is incremented, and the steps are repeated until a case is reviewed (step 855, where n is the number of cases retrieved and / or selected from the cases retrieved in step 805). For example, at least a portion of any display of the attention dashboard 705 may be generated through some or all of the steps 805 to 855.

[0142] Once n cases have been evaluated, step 860 generates a provider global success summary based on a positive CSS. For example, a case matching display 745 and / or a provider advantage display 760 may be generated at least in part as disclosed with reference to step 860.

[0143] In step 865 (e.g., based on negative CSS), provider global recommendations are generated. For example, provider-improved display 765 can be generated at least in part as disclosed with reference to at least step 865.

[0144] Figure 9 Describing such as at least reference Figures 3 to 4 The disclosed method is an exemplary method for training a model. Method 900 can be executed, for example, by a processor (e.g., processor 205) of a program that executes instructions to retrieve data from a data repository (e.g., data source 135). Method 900 includes receiving historical data (e.g., from data source 135, such as at least referenced data) at step 905. Figures 3 to 4 (As disclosed). At step 910, corresponding historical data from the same and / or other sources (e.g., associations, knowledge bodies, and / or other data sources, such as those disclosed elsewhere herein) are identified and retrieved.

[0145] At step 915, the retrieved data is divided into a first dataset for training and a second dataset for testing. At step 920, a model (e.g., a model of inference engine 115 and / or difference engine 140) is applied to the training data to generate a trained model (e.g., a neural network model, a classifier, a large language model). In step 925, the trained model is applied to the test data to generate test output (e.g., such as at least a reference). Figures 3 to 4(Disclosed). At decision point 930, the output is evaluated to determine whether the model has been successfully trained (e.g., by comparison with a predetermined training criterion). The predetermined training criterion may include, for example, a maximum error threshold. This criterion may include, for example, an expert review threshold (e.g., percentage of accuracy, percentage of correctness, binary accurate / inaccurate, complete / incomplete). This criterion may include, for example, a match with the actual result (e.g., percentage of a match). For example, if the difference between the actual output (test data) and the predicted output (test output) is within a predetermined range, the model may be considered successfully trained. If the difference is not within a predetermined range, the model may be considered unsuccessfully trained. At step 935, the processor may generate a signal requesting additional training data, and method 900 loops back to step 930. If it is determined at decision point 930 that the model has been successfully trained, the trained model may be stored in step 940 (e.g., in storage module 225), and method 900 may, for example, terminate.

[0146] In some implementations, such as those depicted, the method may include, for example, dynamic training and / or updating of the model. For instance, method 900 may include a decision point 945 at which (e.g., periodically) model update (e.g., retraining) triggers are monitored. When a trigger is detected (e.g., accuracy drift, availability of additional data, reaching a time period), method 900 may return to step 905.

[0147] Although various embodiments have been described with reference to the accompanying drawings, other embodiments are possible.

[0148] For example, in some implementations, PPIS 105 can be configured to operate as a patient-doctor communication system. This system may, for example, include a patient data agent and a doctor data agent. The patient data agent may, for example, be configured to analyze patients' search history and inquiries. The doctor data agent may, for example, be configured to analyze doctors' search history and treatment patterns. The two agents may, for example, be configured to communicate directly to extract information and create an efficient language for patient-doctor communication.

[0149] In some implementations, PPIS 105 can be configured to operate as a personalized treatment recommendation system. For example, PPIE 110 can be operatively coupled to a database of patient treatment and outcomes. The system can, for example, be configured to analyze current patient data and compare it to similar past cases. The system can, for example, be configured to suggest treatment options based on the physician's past success with similar patients.

[0150] In some implementations, PPIS 105 can be configured, for example, to operate as a surgical assistance system. PPIE 110 can, for example, include a computer vision device and / or engine configured to analyze surgical procedures in real time. The system can, for example, be configured to provide surgeons with contextual information and recommendations based on a large number of sources of similar surgeries and outcomes.

[0151] In some implementations, PPIS 105 can be configured, for example, to operate as a post-treatment monitoring system. PPIE 110 can, for example, include a patient monitoring module. This module can, for example, be configured to track online searches and inquiries from patients after treatment. The system can, for example, be configured to alert the physician if the patient's behavior indicates a potential problem or dissatisfaction with the treatment.

[0152] In some implementations, PPIS 105 can be configured to operate as a physician performance analysis system. PPIE 110 may, for example, include modules configured to analyze physician performance across a variety of procedures. The system may, for example, be configured to identify areas of physician expertise and areas that may require improvement. The system may, for example, be configured to provide personalized career development recommendations.

[0153] In some implementations, PPIS 105 can be configured to operate as a patient understanding verification system. PPIE 110 may, for example, include an interaction module for the patient. This module may be configured, for example, to inquire about the patient's understanding of their condition and treatment plan. The system may be configured, for example, to provide additional explanations or to notify the physician if the patient's understanding is insufficient. For example, personalized consent procedures and / or documents may be generated (e.g., by inference engines 115 and / or 140 and / or based on the output from inference engines 115 and / or 140).

[0154] In some implementations, PPIS 105 can be configured to operate as a treatment deviation analysis system. PPIE 110 may, for example, include a module for comparing planned treatment with actual treatment performed. This module may be configured, for example, to analyze deviations, record the causes of changes, and communicate these changes to the patient in an understandable manner.

[0155] In some implementations, PPIS 105 can be configured to operate as a continuous healthcare communication system. PPIE 110 may include, for example, a patient-facing application. This application may be configured, for example, to allow patients to ask questions and express concerns after treatment. The system may be configured, for example, to provide AI-generated responses or escalate questions to a physician if necessary.

[0156] Although an exemplary system has been described with reference to the accompanying drawings, other implementations may be deployed in other industrial, scientific, medical, commercial, and / or residential applications.

[0157] In various implementations, some bypass circuitry implementations can be controlled in response to signals from analog or digital components, which can be discrete, integrated, or a combination of both. Some implementations may include programmed devices, programmable devices, or some combination thereof (e.g., PLA, PLD, ASIC, microcontroller, microprocessor), and may include one or more data repositories (e.g., memory cells, registers, blocks, pages) to provide single-level or multi-level digital data storage capabilities, and may be volatile, non-volatile, or some combination thereof. Some control functions may be implemented in hardware, software, firmware, or any combination thereof.

[0158] A computer program product may contain a set of instructions that, when executed by a processor device, cause the processor to perform a specified function. These functions may be performed in conjunction with a controlled device that is operatively in communication with the processor. A computer program product that may include software may be stored in a data repository tangibly embedded in a storage medium (such as an electronic storage device, magnetic storage device, or rotating storage device) and may be fixed or removable (e.g., hard disk, floppy disk, thumb drive, CD, DVD).

[0159] Although an example of a portable system has been described with reference to the above figures, other implementations can be deployed in other processing applications, such as desktop and network environments.

[0160] Temporary auxiliary energy input can be received, for example, from a rechargeable or disposable battery, enabling use in portable or remote applications. Some implementations can operate in conjunction with other DC voltage sources, such as a 9V (nominal) battery. Alternating current (AC) input (which can be provided, for example, from a 50 Hz / 60 Hz power port or from a portable generator) can be received via a rectifier and appropriate scaling. Providing AC input (e.g., sine wave, square wave, triangle wave) can include a line frequency transformer to provide voltage boost, voltage reduction, and / or isolation.

[0161] While specific architectural features have been described, other features can be incorporated to improve performance. For example, caching (e.g., L1, L2, ...) techniques can be used. Random access memory may be included, for example, to provide temporary storage and / or to load stored executable code or parameter information for use during runtime operation. Other hardware and software may be provided to perform operations, such as network or other communication using one or more protocols, wireless (e.g., infrared) communication, stored operating energy and power (e.g., batteries), switching and / or linear power supply circuitry, software maintenance (e.g., self-testing, upgrades), etc. One or more communication interfaces may be provided to support data storage and related operations.

[0162] Some systems can be implemented as computer systems that can be used with various implementations. For example, various implementations may include digital circuit systems, analog circuit systems, computer hardware, firmware, software, or combinations thereof. Apparatus may be implemented in a computer program product tangibly embodied in an information carrier (e.g., a machine-readable storage device) for execution by a programmable processor; and methods may be executed by a programmable processor of a program executing instructions to perform the functions of various implementations by manipulating input data and generating output. Various implementations may advantageously be implemented in one or more computer programs that can be executed on a programmable system including at least one programmable processor coupled to receive data and instructions from a data storage system, at least one input device, and / or at least one output device, and to send data and instructions to the data storage system, at least one input device, and / or at least one output device. A computer program is a set of instructions that can be used directly or indirectly in a computer to perform a specific activity or produce a specific result. Computer programs can be written in any form of programming language, including compiled or interpreted languages, and can be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for a computing environment.

[0163] Suitable processors for executing instructions include, for example, both general-purpose microprocessors and special-purpose microprocessors, which can comprise a single processor or one of multiple processors in any type of computer. Typically, the processor receives instructions and data from read-only memory or random access memory, or both. The fundamental components of a computer are the processor for executing instructions and one or more memories for storing instructions and data. Typically, a computer will also include one or more mass storage devices for storing data files, or operatively coupled to communicate with one or more mass storage devices for storing data files; such devices include disks, such as internal hard disks and removable disks; magneto-optical disks; and optical disks. Storage devices suitable for tangibly embodying computer program instructions and data include all forms of non-volatile memory, for example, semiconductor memory devices such as EPROM, EEPROM, and flash memory devices; disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and memory may be supplemented by or incorporated into an ASIC (Application-Specific Integrated Circuit).

[0164] In some implementations, each system can be programmed with the same or similar information and / or initialized with substantially the same information stored in volatile and / or non-volatile memory. For example, when coupled to a suitable host device, such as a desktop computer or server, a data interface can be configured to perform automatic configuration, automatic download, and / or automatic update functions.

[0165] In some implementations, one or more user interface features can be customized to perform specific functions. Various implementations can be implemented in computer systems including graphical user interfaces and / or internet browsers. To provide interaction with the user, some implementations can be implemented on a computer having a display device. The display device can, for example, include an LED (light-emitting diode) display. In some implementations, the display device can, for example, include a CRT (cathode ray tube). In some implementations, the display device can include, for example, an LCD (liquid crystal display). The display device (e.g., a monitor) can be used, for example, to display information to the user. Some implementations can, for example, include a keyboard and / or pointing devices (e.g., a mouse, touchpad, trackball, joystick), through which the user can provide input to the computer.

[0166] In various implementations, the system can communicate using suitable communication methods, equipment, and techniques. For example, the system can communicate with compatible devices (e.g., devices capable of transmitting data to and / or from the system) using point-to-point communication, where messages are transmitted directly from the source to the receiver via a dedicated physical link (e.g., fiber optic link, point-to-point cabling, daisy chain). Components of the system can exchange information via analog or digital data communication of any form or medium, including packet-based messaging over a communication network. Examples of communication networks include, for example, LANs (Local Area Networks), WANs (Wide Area Networks), MANs (Metropolitan Area Networks), wireless and / or optical networks, computers and networks constituting the Internet, or some combination thereof. Other implementations can transmit messages by broadcasting to all or substantially all devices coupled together by the communication network, for example, by using omnidirectional radio frequency (RF) signals. Other implementations can transmit messages characterized by high directionality, such as RF signals transmitted using directional (i.e., narrow-beam) antennas or infrared signals that can optionally be used with focusing optics. Other implementations using appropriate interfaces and protocols are possible, such as, by way of example and not intended to be limiting, USB 2.0, Firewire, ATA / IDE, RS-232, RS-422, RS-485, 802.11 a / b / g, Wi-Fi, Ethernet, IrDA, FDDI (Fiber Distributed Data Interface), Token Ring networks, frequency division, time division, or code division based multiplexing techniques, or some combination thereof. Some implementations may optionally incorporate features such as error checking and correction (ECC) for data integrity or security measures such as encryption (e.g., WEP) and password protection.

[0167] In various implementations, the computer system may include Internet of Things (IoT) devices. IoT devices may include objects embedded with electronics, software, sensors, actuators, and network connectivity, enabling these objects to collect and exchange data. IoT devices can be used with wired or wireless devices by sending data to another device via an interface. IoT devices can collect useful data and then enable the data to flow automatically between other devices.

[0168] Various examples of modules can be implemented using circuit systems comprising a variety of electronic hardware. By way of example, and not limitation, the hardware may include transistors, resistors, capacitors, switches, integrated circuits, other modules, or some combination thereof. In various examples, a module may include analog logic, digital logic, discrete components, traces, and / or memory circuitry fabricated on a silicon substrate comprising various integrated circuits (e.g., FPGAs, ASICs), or some combination thereof. In some implementations, a module may involve software that executes pre-programmed instructions, is executed by a processor, or some combination thereof. For example, various modules may involve both hardware and software.

[0169] In one illustrative aspect, the system of claim 1 and / or any of the figures disclosed herein can be implemented as a computer-implemented method executed by at least one processor. In some embodiments, any method disclosed herein can be implemented as a computer-implemented method executed by at least one processor. By way of example and not limitation, any combination of the steps of method 600, method 601, method 800 and / or method 900 can be implemented as one or more computer-implemented methods.

[0170] In one exemplary aspect, a computer program product (CPP) may include a program of instructions tangibly embodied on a non-transitory computer-readable medium, wherein when the instructions are executed on a processor, the processor causes operations to be performed. Operations may include one or more steps (including skipped steps) of any method disclosed herein, and / or any steps reasonably necessary to implement the specifically described steps. For example, steps that at least refer to claim 1 and / or any of the accompanying drawings disclosed herein may be implemented in a CPP. By way of example and not limitation, any combination of steps of method 600, method 601, method 800, and / or method 900 may be implemented as one or more CPPs.

[0171] Many embodiments have been described. However, it will be understood that various modifications can be made. For example, advantageous results can be achieved if the steps of the disclosed technique are performed in a different order, or if the components of the disclosed system are combined in a different manner, or if the components are supplemented with other components. Therefore, other embodiments are contemplated within the scope of the appended claims.

Claims

1. A system comprising: A data repository, the data repository comprising a program of instructions; and At least one processor, operatively coupled to the data repository, such that when the at least one processor executes a program of the instructions, the at least one processor causes operations to be performed to generate context-specific interactive data objects, the operations including: In response to events related to both the patient and the healthcare provider, a patient knowledge body object BOK1 is then generated, which includes queries associated with one or more devices of the patient and the results of those queries; Based on the association between any two or more of the repository, the patient, the provider, and the event, the BOK1 is updated to include data retrieved from the public and personal repositories of the patient and / or the provider; Applying the inference engine to the BOK1 enables the generation of a natural language summary NL1 of the patient's current understanding state; Applying the difference engine to NL1 and the natural language output NL2 corresponding to the event and based on the provider's knowledge body object BOK2, such that a natural language summary NL3 is generated of the difference between the patient's current state of understanding and the predicted state of understanding for the patient as targeted by the provider; and The NL3 is presented to the provider.

2. The system of claim 1, further comprising any modules, engines, data sources, other components and / or operating steps disclosed with reference to Figures 1 to 9.

Images