Method for predicting communication strategies with patients and server computer thereof

Abstract

To provide a method for predicting communication strategies with patients. [Solution] The method includes receiving demographic data of a patient, finding a model persona in a persona database based on the patient's demographic data, generating one or more first tasks based on a first communication strategy that matches the attribute data of the model persona in the persona database, continuously sending the one or more first tasks based on the first communication strategy to the patient's client computer via a network, continuously receiving one or more first responses to the one or more first tasks from the client computer, and recording a first interaction history in the persona database, the first interaction history including the one or more first tasks and the corresponding one or more first responses.

Description

Cross-reference to related patent applications

[0001] This patent application is based on Provisional Patent Application No. 63687011, filed August 26, 2024. [Technical Field]

[0002] The present invention relates to a network system, and more particularly to a network system for predicting a communication strategy with a patient based on a persona database. [Background technology]

[0003] There are many different types of patients, each with different preferred communication times and capacities for engagement during an encounter. Clinics and new patients take time to determine the appropriate time and capacities for engagement. Also, preferred communication times and capacities for engagement with patients may change over time.

[0004] Therefore, there is a need to provide a mechanism for suggesting communication times and engagement capabilities for new patients by finding a comparable persona within the patient database. Summary of the Invention [Means for solving the problem]

[0005] According to one embodiment of the present application, there is provided a method for predicting a communication strategy with a patient, the method including: receiving demographic data of the patient; finding a model persona in a persona database based on the patient's demographic data; generating one or more first tasks based on a first communication strategy that matches attribute data of the model persona in the persona database; continuously transmitting the one or more first tasks based on the first communication strategy to a client computer of the patient via a network; continuously receiving one or more first responses to the one or more first tasks from the client computer; and recording a first interaction history including the one or more first tasks and the corresponding one or more first responses in the persona database.

[0006] According to one embodiment of the present application, a server computer comprises a networking device configured to connect to a network and a processor configured to execute instructions stored in a non-volatile memory and implement the above-described method for predicting a communication strategy with a patient.

[0007] The presented method and server computer can quickly suggest communication strategies for new patients or previously engaged patients whose demographic data has changed by finding comparable personas in the persona database. Additionally, one or a combination of machine learning models can be used to suggest modified communication strategies based on multiple rounds of interactions between the patient and the network system. Patient demographic data can update the persona database to broaden the comparison criteria. Furthermore, interaction history can be used as a training set to train one or a combination of machine learning models to improve communication strategy suggestions.

[0008] The advantages and principles of the present invention may be further understood from the following detailed description and drawings. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a block diagram of a network system 100 for predicting communication strategies with patients. [Figure 2] FIG. 2 illustrates a software architecture 200 implemented by a server for predicting communication strategies based on demographic data in a persona database, according to one embodiment of the present application. [Figure 3] FIG. 3 illustrates a database schema 300 of a proposed persona database according to one embodiment of the present application. [Figure 4] 4 is a block diagram of a computer 400 according to an embodiment of the present application. [Figure 5] 5 is a flowchart of a method 500 for predicting a communication strategy based on demographic data in a persona database according to an embodiment of the present application. [Figure 6A] FIG. 2 illustrates a portion of an entry in a persona database according to one embodiment of the present application. [Figure 6B] FIG. 2 illustrates a portion of an entry in a persona database according to one embodiment of the present application. [Figure 6C] FIG. 2 illustrates a portion of an entry in a persona database according to one embodiment of the present application. [Figure 6D] FIG. 2 illustrates a portion of an entry in a persona database according to one embodiment of the present application. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, several embodiments of the present application will be described in detail. However, in addition to the following description, the present invention can also be applied to other embodiments, and the scope of the present invention is not limited by such, but is limited by the scope of the claims. Also, for better understanding and clarity of the description, some components in the drawings are not necessarily to scale, and some are exaggerated relative to others and are irrelevant. If the relationship between two steps is not described, the order of their execution is not fixed to the order shown in the flowchart.

[0011] One aspect of the present application is to provide a networked system for predicting communication strategies with patients, including but not limited to time, duration, topic, and ability, based on records in a persona database. The persona database in the networked system collects a plurality of different persona records, each of which includes two parts: a first part of the persona records includes persona demographic data, and a second part of the persona records includes persona attribute data.

[0012] The persona demographic data may include any combination of patient nickname, age, gender, social determinants of health (SDOH), hours worked per week or month, comorbidities, medications, symptoms, motivation, conversational intensity, and / or special time events. Special time events include, but are not limited to, reminder reset times (e.g., 10:15 PM) and home time (e.g., 7:15 PM). The persona attribute data may include any combination of preferred communication times, preferred communication durations, preferred communication topics, and daily task capabilities.

[0013] Communication strategies may be refined and / or modified to meet the patient's needs when received for new patients or patients with demographic changes. Thus, in one embodiment, the proposed network system may view patients with demographic changes as new records in the persona database. Note that a single patient may therefore be associated with one or more persona records in the database.

[0014] After admission of a new patient or a previously involved patient whose demographic data has changed, a model persona can be found in the persona database by comparing the record in the persona database with the new patient or patient whose demographic data has changed. In one embodiment, the comparison can be based on the patient's demographic data. The comparison can be performed in two ways: by a deterministic algorithm or by a machine learning model.

[0015] Because there are many columns of demographic data in the persona database, in a first example of a comparison algorithm, a model persona may be selected because its record has the most columns that are identical to the new patient. In a second example, the difference corresponding to each column in the demographic data between the existing record and the new patient is calculated. The record with the smallest sum of differences may be selected as the model persona. In a variation of the second example, the differences may each be weighted.

[0016] In a machine learning approach, one or a combination of machine learning models may be employed to find the model persona. The machine learning model may be trained based on demographic data from the persona database. Alternatively, the machine learning model may be trained based on demographic data provided by other social databases.

[0017] Some constraints or requirements may be imposed when finding a model persona. For example, a constraint may be required in the comparison that the gender of the initial patient and the model persona must be the same. Another constraint may require that the age difference be less than three years. Those skilled in the art can understand that the present application does not limit the method of finding a model persona that is closest to the initial patient according to demographic data.

[0018] Once a model persona is found, persona attribute data corresponding to the model persona can be retrieved from the persona database as an initial suggestion to the new patient. The suggested persona attribute data can include a communication strategy, including one or any combination of preferred communication time, preferred communication duration, preferred communication topic, and daily task ability.

[0019] In one embodiment, interaction history with the patient may also be recorded to provide more input to the prediction module. For example, the interaction history may include how long it takes the patient to complete a task or survey (i.e., relating to task performance), the patient's perspective on their health problem (i.e., relating to preferred communication topics), and the duration of the interaction (i.e., relating to preferred communication duration).

[0020] The prediction module may use the proposed attribute data retrieved from the persona database and optional interaction history input to predict attribute data for the patient. Based on the predicted attribute data of the communication strategy, the network system may appropriately communicate with the patient. If the patient is satisfied with the predicted attribute data of the communication strategy, the update module may update the persona database with the predicted attribute data along with the patient's corresponding demographic data.

[0021] In one embodiment, the interaction history can be analyzed to obtain the patient's tonicity preference. The analysis of tonicity preference can be performed by one or a combination of large-scale language models (LLMs) based on the patient's responses. The analyzed tonicity preference can be recorded in the patient's demographic and / or attribute data in a persona database so that the system can communicate with the patient at the preferred tonicity next time. The table below shows examples of two different conversational attitudes / tonicities corresponding to the same context. [Table 1]

[0022] In some examples, communication between the proposed system and the patient can be supported and provided by the LLM using contextual information and / or tension preferences as prompts to the LLM. Patient responses in the interaction history can also be extracted by the LLM. Those skilled in the art can understand that current LLMs can analyze tension and generate output according to tension requirements defined in input prompts.

[0023] If it takes multiple rounds of communication for the patient to be satisfied with the latest communication strategy, the update module may update the persona database with the latest communication strategy along with the patient's corresponding attribute data. In addition, the multiple rounds of communication may be further recorded as part of the training data. Another one or a combination of machine learning models may also be trained according to the training data of the multiple communications. In an alternative embodiment, the prediction module may utilize one or a combination of machine learning models to adjust the predicted attribute data.

[0024] 1, which illustrates a block diagram of a network system 100 for predicting communication strategies with patients according to one embodiment of the present application. The network system 100 may include a server 110, a network 120, a clinic computer 130 for clinic clerks, a client computer 140 for patients, and a DBMS server 112. In addition, the network system 100 may further include an optional artificial intelligence (AI) server 114 and an optional message server 150.

[0025] Network 120 is configured to exchange data between server 110 and detector 130 and between server 110 and client computer 140. For example, network 120 may include an access network such as a WiFi / IEEE 802.11 network, a 3G / 4G / 5G / 6G-enabled network, an IEEE 802.3 network, a PSTN network, an optical fiber network, and / or an xDSL network. Server 110, detector 130, client computer 140, optional AI server 112, and / or message server 150 may be configured to connect to the access network. Network 120 may also include a backbone network, e.g., a telecommunications network, for connecting to the access network.

[0026] The exchange of information or messages between the server 110 and the clinic computer or client computer 140 may go through a message server 150. The message server 150 may implement one or more public messaging services for the server 110 and the detection device 130. For example, the message server 150 may be an email server, an instant messaging server (such as Line, WhatsApp®, Skype®, or Facetime®), or a message queue server. One of the functions provided by the message server 150 is asynchronous transmission, which means that messages are temporarily stored in the message server 150 until they are securely received by the server 110 and may be resent by the message server 150. Another function provided by the message server 150 is encrypted transmission, which ensures secure transmission. However, the message server 150 is optional in the present embodiment.

[0027] As an alternative to message server 150, SMS / MMS server 152 can send and receive short or multimedia messages from server 110 to clinic computer 130 and client computer 140, and vice versa. SMS / MMS server 152 can provide scheduling functionality to server 110, meaning that server 110 can pre-determine when to update and transmit information. Thus, at the pre-determined time, SMS / MMS server 152 can deliver information packaged in a short or multimedia message to clinic computer 130 or client computer 140 via a telecommunications network included in network 120. In addition, SMS / MMS server 152 can receive and deliver patient responses from client computer 140 to server 110.

[0028] After receiving a response from the clinic computer 130 or the client computer 140, the server 110 may analyze the information and the corresponding timestamps of the responses. The accumulated timestamps and corresponding responses may be used to infer the user's preferred communication times, durations, topics, and / or daily task abilities. Daily or weekly tasks may include meals (breakfast, lunch, afternoon tea, dinner), exercise, work, and sleep. The collection of daily tasks constitutes the user's behavior or daily schedule. Similarly, the collection of weekly or monthly tasks constitutes the user's behavior or cyclical schedule. In one embodiment, the analysis of the physiological information and the corresponding timestamps may be performed by applying one or more machine learning algorithms / models to obtain the user's cyclical schedule. The cyclical schedule consists of at least one task.

[0029] See FIG. 2, which illustrates a software architecture 200 implemented by a server for predicting communication strategies based on demographic data in a persona database, according to one embodiment of the present application. In one example, software architecture 200 may be implemented solely by server 110 as shown in FIG. 1. In another example, software architecture 200 may be collectively implemented by a combination of server 110, DBMS server 112, and / or AI server 114 as shown in FIG. 1.

[0030] Software architecture 200 may include a database management layer 210 for operating the persona database, a business logic layer 220 including a prediction module 224 and an update module 226 for implementing the methods provided herein, and an interface layer 230 for interacting with clinic computer 130 and client computer 140 via network 120. Optionally, an AI service layer 232 may be configured to provide inference services for software architecture 200.

[0031] In one embodiment, database management layer 210 may be incorporated into DBMS server 112, as shown in FIG. 1. In one embodiment, AI service layer 222 may be incorporated into AI server 114, as shown in FIG. 1. DBMS server 112 may include a storage farm to provide large-scale, inexpensive storage space. AI server 114 may include specialized hardware, such as a graphics processing unit and / or a neural network processing unit, for training and / or inferring machine learning models provided herein. Alternatively, AI service layer 222 may be provided by a commercially available service, such as Co-pilot, Gemini®, ChatGPT®, Perplexity®, DeepSeek®, and / or any other type of AI service.

[0032] Server 110 may itself be capable of implementing machine learning or AI models. In an alternative embodiment, AI server 114 may be further utilized by server 110 to generate task-specific health-related information for the user. In one example, AI server 114 may be configured to train or infer various machine learning algorithms / models used in embodiments of the present application. The machine learning models may be a combination of various neural networks, such as convolutional neural networks, deep neural networks, and transformer networks. Those skilled in the art will appreciate that AI server 114 may be provided by a vendor as a cloud service. For example, Google® cloud services, Amazon® AWS services, and Microsoft® Azure® services provide various types of neural network services that may be considered as the AI ​​server 114 herein. However, even without AI server 114, server 110 may have sufficient computing power to implement the machine learning algorithms / models presented herein.

[0033] See FIG. 3 , which illustrates a database schema 300 of a proposed persona database according to one embodiment of the present application. Those skilled in the art will appreciate that database schema 300 is suitable for a relational database. However, database schema 300 may be modified to be implemented with a database other than a relational database. For clarity, database schema 300 as illustrated in FIG. 3 does not illustrate patient personal information and / or any other information that may be included in the persona database.

[0034] As described above, for one persona, there are two corresponding data, including demographic data 310 and attribute data 320. The demographic data 310 may include one or any combination of the patient's age, gender, social determinants of health (SDOH), working hours per week or month, comorbidities, medications, symptoms, motivation, conversational intensity, and / or special time events. The persona attribute data 320 may include any combination of preferred communication time, preferred communication duration, preferred communication topic, and task ability within a certain period of time (e.g., a day, a week, or a month). Additionally, the demographic data 310 and attribute data 320 may include more columns than those described herein.

[0035] See FIG. 4, which illustrates a block diagram of a computer 400 according to one embodiment of the present application. The various computers illustrated in FIG. 1 may be implemented as variations of the computer 400. Some components of the computer 400 may be missing or modified to suit the role they play. For example, the AI ​​server 114 may lack a display 440 and input devices 480, such as a keyboard and mouse. The graphics processing unit (GPU) 230 of the AI ​​server 114 may be modified to include dozens of neural network processing units (NPUs). Those skilled in the art, with knowledge of computer configurations, computer architectures, system software, and operating systems, can implement a computer such as that illustrated in FIG. 1 based on the block diagram illustrated in FIG. 4. The clinic computer 130 and the client computer 140 may also have input devices 280, such as a touchscreen, touch panel, keyboard, or buttons, to receive user or patient responses.

[0036] The computer 400 includes at least one central processing unit (CPU) 410, a memory module 420 for system operation, an optional GPU 430 for generating items shown on a display 440, a peripheral connection device 450 for implementing industry-standard interfaces such as PCI, PCI-Express®, SCSI, SATA, USB®, etc., a networking device 460 for connecting to the network 120, a storage device 470 for storing an operating system, application programs, and data for implementing steps provided herein, one or more input devices 480, and one or more output devices 490. The operating system executed by the CPU 410 is configured to control the computer 400. Various driver programs and application programs under the operating system may also be stored in non-volatile memory such as the storage device 470 and used to implement steps of embodied methods provided herein.

[0037] 5, which illustrates a flowchart of a method 500 for predicting a communication strategy based on demographic data in a persona database according to an embodiment of the present application. The method 500 may be performed by the network system 100, particularly by the server 110. If there is no direct or indirect causal relationship between any two steps as illustrated in FIG. 5, the present application does not limit the order of their execution. The prediction method 500 may start from step 510.

[0038] Step 510: Receive patient information and demographic data. The server 110 may receive the patient information and demographic data via the clinic computer 130 and / or the client computer 140. If the patient is a new patient, the flow may proceed to step 520.

[0039] Step 520: Find the best-matching (matching, or closest if not matching) model persona in the persona database based on the patient's demographic data. To do so, persona attribute data corresponding to the model persona can be retrieved from the persona database. As described above, the best-matching model persona can be found in the persona database by a deterministic method or by a machine learning model. One or more requirements or constraints can also be applied to the finding step.

[0040] Step 530: Generate one or more tasks based on communication strategies tailored to persona attribute data and / or patient demographic data corresponding to the model persona.

[0041] In one embodiment, one or more tasks are generated according to contextual information and tension preferences recorded in the model persona's demographic data. Alternatively, a default tension may be employed to generate the one or more tasks. As described above, generation may be assisted by one or more LLMs. Each of the tasks is scheduled based on a communication strategy.

[0042] In another embodiment, one or more tasks are generated according to the patient's contextual information and demographic data. For example, when the patient specifically indicates that they will be home at 7:00 PM, the communication strategy generated based on the model persona may be modified to start communication 30 minutes after they arrive home, i.e., at 7:30 PM. In other words, one or more variations defined in the patient's demographic data may be taken into account when generating the communication strategy in this step.

[0043] Step 540: Continuously transmit one or more tasks to the patient's client computer based on the schedule of the one or more tasks. The transmission may be directly from the server 110 to the client computer 140. Alternatively, the transmission may be indirect from the server 110 to the client computer 140 via the message server 150 or the SMS / MMS server 152.

[0044] Step 545: Continuously receive responses from client computers to one or more tasks and timestamps of the received responses. In one embodiment, the timestamps are received with the tasks. In an alternative embodiment, the timestamps are generated locally within server 110.

[0045] Step 550: Record an interaction history with the patient based on steps 540 and 545. In one embodiment, after one round of interaction, the patient may indicate a satisfaction level with the communication strategies implemented in that one round of interaction. The satisfaction level may be a binary value of like or dislike. Alternatively, the satisfaction level may be a score of 1 to 5 or 1 to 10. The satisfaction level threshold may be 4 out of 1 to 5 or 7 out of 1 to 10. If a value greater than the threshold is evaluated, it means that the patient is satisfied with the communication strategies implemented in the one round of communication.

[0046] Step 560: Determine whether the patient is satisfied with the communication strategy. The determination may be based on the satisfaction score provided by the patient. If the patient is satisfied with the communication strategy embodied in the interaction history recorded in step 550, the flow may end here. Alternatively, the flow may proceed further to step 570. Otherwise, if the patient is not satisfied with the communication strategy embodied in the interaction history recorded in step 550, the flow may proceed to step 565.

[0047] Step 565: Receive modified communication strategy. In one embodiment, the modified communication strategy may be provided by the patient. In another embodiment, the modified communication strategy may be generated by the server 110. In this embodiment, the modified communication strategy is a variation of the communication strategy used in step 530. After the modified communication strategy is generated by one or a combination of machine learning models, the flow may return to step 530 with the modified communication strategy.

[0048] Step 570: Compare the patient's demographic data with the demographic data of the selected persona model. If there are no differences, the flow proceeds to step 590. Otherwise, the flow proceeds to step 580.

[0049] Step 580: If the patient's demographic data differs from that from the selected persona model, create a new persona model with demographic and attribute data that reflects the patient.

[0050] Step 590: Update the persona attribute data corresponding to the patient in the persona database based on the satisfactory communication strategy.

[0051] See Figures 6A-6D, which illustrate entries in a persona database at various stages of a method 500 according to one embodiment of the present application. A first table of persona demographic data 310 and a second table of persona attribute data 320 are presented in Figures 6A-6D, respectively.

[0052] In this embodiment, a 64-year-old woman who is still employed interacts with network system 100. In step 510 as shown in Figure 5, the woman enters her information and her demographic data into server 110. Or alternatively, a clinic clerk enters the woman's information and her demographic data into server 110. The flow of method 500 proceeds to step 520.

[0053] Assume then that a model persona is found in the database based on the woman's demographic data. As shown in Figure 6A, the model persona that most closely resembles the woman is a 64-year-old man who is still working. The model persona has demographic data that indicates that the preferred communication timing for engagement is Monday at 8:00 AM, and the engagement window / capability is approximately 15 minutes. The man can answer three questions within his capability window.

[0054] Based on the communication strategy of the model persona's attribute data, a similar communication strategy is adopted to generate three tasks for the woman. In step 540, the three prepared questions are sent consecutively to the woman's client computer at 8:00 AM on Monday. Responses to the three questions received from the woman's client computer are received by the server in step 545. Time stamps corresponding to the responses are received together with the responses or generated locally at the server. This round of interaction history with the woman is recorded in a database schema 340 as shown in FIG. 3.

[0055] However, the woman is not satisfied with the communication strategy. The result of the determination in step 560 is not OK. The flow may proceed to step 565 to receive a revised communication strategy. The following week, the flow may proceed to have another round of interactions according to the revised communication strategy. Thus, steps 530-560 are repeated again for multiple interactions.

[0056] In one embodiment, in step 565, an updated version of the modified communication strategy is estimated by one or a combination of machine learning models based on the recorded interaction history. By adopting the updated version of the modified communication strategy, step 560 determines that the woman is satisfied. Because the woman does not have a record in the persona database, flow proceeds from step 570 to step 580. In step 580, a new persona corresponding to the woman is created using her demographic data and her preferred attribute data in the database. The result is shown in Figure 6B.

[0057] According to an embodiment, six months later, the same woman (age 64.5), who is still employed, again interacts with network system 100 (step 510). In step 520, the model persona found in the persona database is, in fact, herself.

[0058] The model persona's corresponding attribute data from the persona record found in the persona database, as shown in FIG. 6B, indicates that Monday at 10:00 AM is preferred and the communication window is approximately 30 minutes. She is also available to take on five tasks. Therefore, five tasks are generated in step 530 and scheduled to be sent to her client computer one by one at 10:00 AM on Monday in step 540. Responses and satisfaction levels for these five tasks are continuously received by server 110 in step 545. This round of interaction history is recorded in the persona database in step 550.

[0059] However, in step 560, the woman is not satisfied with this one round of interaction. The flow proceeds to step 565 to modify the communication strategy. After multiple rounds of interaction, the woman is finally satisfied with the communication strategy estimated by one or a combination of the machine learning models. The flow then proceeds from step 560 to step 570.

[0060] In step 560, the woman's age is incremented, meaning that her demographic data is changed. Therefore, the flow proceeds to step 580. A new record corresponding to the woman is added with new attribute data, see FIG. 6C. The preferred communication timing for the attribute data shown in the bottom row is Monday at 9:00 AM. The window of seven tasks can then last for one hour.

[0061] Further, after the above embodiment, the woman turns 65 and has just retired. In step 510, she interacts with the network system 100 again. According to the persona database, in step 520, the model persona record closest to the woman is still her own record. According to this model persona, the preferred communication timing of the communication strategy recorded in the corresponding attribute data is Monday at 9:00 AM. Also, the window of seven tasks may last for one hour. Therefore, in step 530, seven tasks are generated according to the context information and intensity preferences and scheduled to be sent to her client computer at 9:00 AM on Monday.

[0062] Responses and satisfaction levels for these seven tasks are continuously received by server 110 in step 545. This round of interaction history is recorded in the persona database in step 550.

[0063] However, in step 560, the woman is not satisfied with this one round of interaction. The flow proceeds to step 565 to modify the communication strategy. After multiple rounds of interaction, the woman is finally satisfied with the communication strategy estimated by one or a combination of the machine learning models. The flow then proceeds from step 560 to step 570.

[0064] In step 560, the woman's age is incremented, meaning that her demographic data is changed. Therefore, the flow proceeds to step 580. A new record corresponding to the woman is added with new attribute data, see FIG. 6D. The preferred communication timing for the attribute data shown in the bottom row is Monday at 8:00 AM. The window for the three tasks can then last 30 minutes.

[0065] One or a combination of machine learning models employed to estimate modified communication strategies may be trained according to the interaction history recorded in the persona database. Each round of interaction is labeled with a satisfaction level. A training set including the interaction history and corresponding labels may be used to train one or a combination of machine learning models.

[0066] In one embodiment, the computational resources required for training may be provided by AI server 114 of network system 100 as shown in Figure 1. Training may be performed periodically.

[0067] According to one embodiment of the present application, there is provided a method for predicting a communication strategy with a patient, the method including: receiving demographic data of the patient; finding a model persona in a persona database based on the patient's demographic data; generating one or more first tasks based on a first communication strategy that matches attribute data of the model persona in the persona database; continuously transmitting the one or more first tasks based on the first communication strategy to a client computer of the patient via a network; continuously receiving one or more first responses to the one or more first tasks from the client computer; and recording a first interaction history including the one or more first tasks and the corresponding one or more first responses in the persona database.

[0068] Preferably, the model persona's demographic data is closest to the patient's demographic data in order to match the closest record to the patient's demographic data.

[0069] Preferably, to add a new record for a first-time patient, the method further includes determining whether the patient is a first-time patient or whether the patient's demographic data differs from the patient's existing demographic data in the persona database, and creating a new persona record in the persona database with the patient's demographic data and attribute data corresponding to the first communication strategy.

[0070] Preferably, for updating a record of a previously engaged patient, the method further includes updating attribute data of the existing persona record based on the first communication strategy when the patient's demographic data is identical to the demographic data of the patient's existing persona record.

[0071] Preferably, the one or more first tasks are generated based on the tension preferences of the model persona, such that the one or more first tasks are generated based on the tension preferences in the demographic data of the model persona.

[0072] Preferably, to generate the one or more first tasks based on the patient's tension preferences, the one or more first tasks are generated based on the tension preferences in the patient's demographic data.

[0073] Preferably, to generate the one or more first tasks by a machine learning model, the one or more first tasks are generated by a large-scale language model (LLM).

[0074] Preferably, the one or more first tasks are scheduled according to a special time event in the patient's demographic data, such as the patient's time of returning home or leaving home, to schedule the one or more first tasks based on a special time event.

[0075] Preferably, for having another round of interactions when the most recent round of interactions is unsatisfactory, the method further includes receiving a first satisfaction level of the first interaction history from the patient's client computer, generating a second communication strategy when the first satisfaction level of the first interaction history is lower than a threshold, generating one or more second tasks based on the second communication strategy, continuously transmitting the one or more second tasks based on the second communication strategy to the client computer via a network, continuously receiving one or more second responses to the one or more second tasks from the client computer, and recording the second interaction history including the one or more second tasks and the corresponding one or more second responses in the persona database.

[0076] Preferably, to add a new record for a first-time patient, the method further includes determining whether the patient is a first-time patient or whether the patient's demographic data differs from the patient's existing demographic data in the persona database, and creating a new persona record in the persona database with the patient's demographic data and attribute data corresponding to the second communication strategy.

[0077] Preferably, for updating a record of a previously engaged patient, the method further includes updating attribute data of the existing persona record based on the second communication strategy when the patient's demographic data is identical to the demographic data of the patient's existing persona record.

[0078] Preferably, for having a third round of interactions when the most recent round of interactions is unsatisfactory, the method further includes receiving a second satisfaction level of the second interaction history from the patient's client computer, generating a third communication strategy when the second satisfaction level of the second interaction history is lower than a threshold, generating one or more third tasks based on the third communication strategy, continuously sending the one or more third tasks based on the third communication strategy to the client computer via a network, continuously receiving one or more third responses to the one or more third tasks from the client computer, and recording the third interaction history including the one or more third tasks and the corresponding one or more third responses in the persona database.

[0079] Preferably, the second communication strategy is estimated by one or a combination of machine learning models to generate a modified communication strategy based on the machine learning models.

[0080] Preferably, to train the machine learning models, one or a combination of the machine learning models is trained with a training set that includes interaction histories and labels that include the satisfaction levels of the corresponding interaction histories.

[0081] Preferably, the patient demographic data includes one or any combination of the patient's nickname, age, sex, social determinants of health (SDOH), periodic (e.g., daily, weekly, or monthly) work hours, comorbidities, medications, symptoms, motivation, conversational intensity, and special time events.

[0082] Preferably, the attribute data in the persona database includes one or any combination of preferred communication time, preferred communication duration, preferred communication topic, and ability to perform a task within a period of time.

[0083] Preferably, the patient demographic data is received from a clinic computer other than the patient's client computer, with the assistance of a clinic clerk to enter the patient's data.

[0084] Preferably, to transmit one or more first tasks in advance, the one or more first tasks are transmitted to a message server before their scheduled times, and the message server transmits each of the one or more first tasks to a client computer in accordance with the scheduled times.

[0085] Preferably, to find the model persona with a non-deterministic algorithm, the model persona is found as a result of extrapolating a machine learning model to patient demographic data and a persona database.

[0086] According to one embodiment of the present application, a server computer comprises a networking device configured to connect to a network and a processor configured to execute instructions stored in a non-volatile memory and implement the above-described method for predicting a communication strategy with a patient.

[0087] The presented method and server computer can quickly suggest communication strategies for new patients or previously engaged patients whose demographic data has changed by finding comparable personas in the persona database. One or a combination of machine learning models can also be used to suggest modified communication strategies based on multiple rounds of interaction between the patient and the network system. Patient demographic data can update the persona database to broaden the comparison criteria. Interaction history can also be used as a training set to train one or a combination of machine learning models to improve communication strategy suggestions.

[0088] While the present invention has been described in terms of what are presently considered to be the most practical and preferred embodiments, it is to be understood that the invention is not necessarily limited to the described embodiments, but rather is intended to cover various modifications and similar arrangements included within the spirit and scope of the appended claims, which are to be accorded the broadest interpretation and all such modifications and similar structures.

Claims

1. 1. A method for predicting a communication strategy with a patient, comprising: receiving patient demographic data; and Finding a model persona in a persona database based on the patient's demographic data; generating one or more first tasks based on a first communication strategy that matches attribute data of the model persona in the persona database; continuously transmitting the one or more first tasks based on the first communication strategy to the patient's client computer over a network; sequentially receiving one or more first responses to the one or more first tasks from the client computer; recording a first interaction history including the one or more first tasks and corresponding one or more first responses in the persona database; A method comprising:

2. The method of claim 1 , wherein the demographic data of the model persona most closely resembles the demographic data of the patient.

3. determining whether the patient is a new patient or whether the patient's demographic data differs from the patient's existing demographic data in the persona database; creating a new persona record in the persona database with the patient's demographic data and attribute data corresponding to the first communication strategy; The method of claim 1 further comprising:

4. 4. The method of claim 3, further comprising: when the patient's demographic data is the same as the demographic data of an existing persona record for the patient, updating attribute data of the existing persona record based on the first communication strategy.

5. The method of claim 1 , wherein the one or more first tasks are generated based on tension preferences within demographic data of the model persona.

6. The method of claim 1 , wherein the one or more first tasks are generated based on tension preferences in the patient's demographic data.

7. The method of claim 1 , wherein the one or more first tasks are generated by a large language model (LLM).

8. The method of claim 1 , wherein the one or more first tasks are scheduled according to a particular time event in the patient's demographic data.

9. receiving a first satisfaction level of the first interaction history from the client computer of the patient; generating a second communication strategy when the first satisfaction level of the first interaction history is below a threshold; and generating one or more second tasks based on the second communication strategy; and continuously transmitting the one or more second tasks based on the second communication strategy to the client computer via the network; successively receiving one or more second responses to the one or more second tasks from the client computer; recording a second interaction history including the one or more second tasks and corresponding one or more second responses in the persona database; The method of claim 1 further comprising:

10. determining whether the patient is a new patient or whether the patient's demographic data differs from the patient's existing demographic data in the persona database; creating a new persona record in the persona database with the patient demographic data and attribute data corresponding to the second communication strategy; The method of claim 9 further comprising:

11. updating the demographic data of the patient's existing persona record based on the second communication strategy when the demographic data of the patient is the same as the demographic data of the patient's existing persona record. The method of claim 10 further comprising:

12. receiving a second satisfaction level of the second interaction history from the client computer of the patient; generating a third communication strategy when the second satisfaction level of the second interaction history is lower than the threshold; and generating one or more third tasks based on the third communication strategy; and continuously transmitting the one or more third tasks based on the third communication strategy to the client computer via the network; sequentially receiving one or more third responses to the one or more third tasks from the client computer; recording a third interaction history in the persona database, the third interaction history including the one or more third tasks and corresponding one or more third responses; The method of claim 9 further comprising:

13. The method of claim 9 , wherein the second communication strategy is estimated by one or a combination of machine learning models.

14. The method of claim 13 , wherein one or a combination of the machine learning models is trained with a training set including interaction histories and labels including satisfaction levels of the corresponding interaction histories.

15. 10. The method of claim 1, wherein the patient demographic data includes one or any combination of the patient's nickname, age, gender, social determinants of health (SDOH), cyclical work hours, comorbidities, medications, symptoms, motivation, conversational intensity, and special time events.

16. The method of claim 1 , wherein the attribute data in the persona database includes one or any combination of preferred communication time, preferred communication duration, preferred communication topic, and ability for tasks within a duration.

17. The method of claim 1 , wherein the patient demographic data is received from a clinic computer other than the patient's client computer.

18. 2. The method of claim 1, wherein the one or more first tasks are transmitted to a message server prior to a scheduled time, and the message server transmits each of the one or more first tasks to the client computer in accordance with the scheduled time.

19. The method of claim 1 , wherein the model persona is found as a result of extrapolating a machine learning model to the patient demographic data and the persona database.

20. A server computer comprising a networking device configured to connect to a network and a processor configured to execute instructions stored in non-volatile memory and implement the method of claim 1 for predicting communication strategies with a patient.

Images