Systems, apparatuses, methods, and non-transitory computer-readable storage devices for person-specific cognitive impairment care support

Abstract

A computerized method has the steps of: obtaining a plurality of personal and cultural factors of a patient, obtaining a plurality of health factors of the patient, and generating one or more recommendations based on the plurality of personal and cultural factors and the plurality of health factors.

Description

[0001]SYSTEMS, APPARATUSES, METHODS, AND NON-TRANSITORY COMPUTER- READABLE STORAGE DEVICES FOR PERSON-SPECIFIC COGNITIVE IMPAIRMENT CARE SUPPORT CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of US Provisional Patent Application Serial No.63 / 666,935, filed July 02, 2024, the content of which is incorporated herein by reference in its entirety. FIELD OF THE DISCLOSURE The present disclosure relates generally to computer systems, apparatuses, methods, and non-transitory computer-readable storage devices, and in particular to computer systems, apparatuses, methods, and non-transitory computer-readable storage devices for person-specific cognitive impairment care support. BACKGROUND Dementia, or generally cognitive impairment, is a severe disease impairing human cognition. There are significant publicly available dementia support tools and techniques available in prior art. However, these tools are generally blanket training programs or isolated videos that don’t focus on providing individuals with context on when and where to use them appropriately. Additionally, with these tools, the expectation is on the individual looking at them to be able to decipher if a specific recommendation is appropriate for their use case. While specialized individuals in the dementia care space may have the appropriate education and experience to ensure the recommendation that they are providing is correct, ordinary people need to use their best guess and the prior-art approaches may still not work in their circumstances. This creates a problem as dementia is not a single type of disease, and rather a blanket term for a plurality of diseases such as Alzheimer’s disease, frontotemporal dementia, vascular dementia, Lewy body dementia, and the like. By providing the incorrect type of support, it can lead to significantly negative repercussions to the patient. US Patent No. 11,810,678 B2 to Soenksen, et al. discloses systems and methods for reducing acute incident risks for dementia patients. Benefits of the systems and methods include reducing healthcare costs, improving patient and caregiver outcomes and reducing caregiver burden. A server system analyzes data related to a dementia patient and the corresponding caregivers (e.g., family member, paid caregiver, physician) and calculates an acute incident risk. Based on the acute incident risk, the server system determines patient therapies to reduce the acute incident risk and also identifies caregiver education to improve the caregiver belief state, which reduces the acute incident risk. The system periodically re-calculates the acute incident risk and identifies patient therapies and caregiver education and motivation to further reduce or maintain the acute incident risk. However, while providing recommendations to caregivers, the method disclosed in US 11,810,678 B2 is associated to clinical risk and does not consider person or ethnic factors. US Patent Publication No.2019 / 0197632 A1 to Schneider discloses an automated memory reminder system, memory assistance system, gift and / or social engagement recommendation system and personal social network system which may be subscribed to by a user or a user may invite friends and family to participate. The personal social network comprises a smart device application and an administration panel functional algorithm called Not So Forgetful (NSF). NSF distinguishes as an event, gift and / or social engagement and memory reminder system that is personal to the user and may run on a computer processor personal to the user such as a mobile communications device, smart home, smart television, smart watch, smart glasses and the like. The NSF personal network may be limited in extent to, for example, friends, family and significant others including some business associates and the like. The NSF system including an NSF server and related databases and search engines may assist an individual diagnosed with a memory deficiency to remember friends and family via the smart glasses. For example, for light reminders, there is a list for things to do which may vary from week to week (such as “Honey-Do” checklist for shopping, dry cleaning, which may be generated from the user or a contact of the user, etc.) provided by an automated administration panel via the NSF server and editable also by the NSF user / member. The checklists also may provide recommendations for items to be purchased online or in-person through purchasing affiliate account links of smart device provided coupons from the purchasing affiliates providing database search engines as well as links to public source search engines. Algorithms include a trending algorithm utilizing NSF network and external data to develop, for example, a top ten gift and / or social engagement list for a teenager female, a gift and / or social engagement registry for recording gift and / or social engagements desired and a gift and / or social engagement depository for recording gift and / or social engagements purchases for supporting the trending algorithm based on demographics. The system disclosed in 2019 / 0197632 A1 is for supporting dementia caregivers, and it is a memory reminder system. However, this system is not a recommendation system for cognitive impairment care support. There are also programs available for dementia care (such as Help4Dementia by the Alzheimer’s Society of Canada). However, these programs merely provide a list of things a caregiver can do. None of them are for supporting patients at individual level. In prior art, other attempts give the patients all possible information and then the patients must determine what (if any) is applicable to them and their circumstance. This gets tricky when the behavioral manifestation of dementia differs significantly between frontotemporal dementia, Alzheimer’s disease, and Lewy body dementia, and the techniques to support differ depending on their personal and cultural circumstances. SUMMARY According to one aspect of this disclosure, there is provided a computerized method comprising: obtaining a plurality of personal and cultural factors of a patient; obtaining a plurality of health factors of the patient; and generating one or more recommendations based on the plurality of personal and cultural factors and the plurality of health factors. In some embodiments, the plurality of personal and cultural factors comprise: personal factors; professional factors; circumstantial factors; apprehension factors; or a combination thereof. In some embodiments, the plurality of health factors comprise: clinical diagnoses; allergies; or a combination thereof. In some embodiments, the personal factors comprise: age; religion; birth place; ethnicity; languages spoken; preferred language; gender; or a combination thereof. In some embodiments, the professional factors comprise: education; profession; or a combination thereof. In some embodiments, the circumstantial factors comprise: hobbies; growing-up history; community involvement; physical activities; sports; music; movies; television; pets; or a combination thereof. In some embodiments, the apprehension factors comprise: dislikes; triggers; fears; agitations; or a combination thereof. In some embodiments, the method further comprises: using a first artificial intelligence (AI) engine and a corresponding first AI model for: determining one or more other factors; and weighing the plurality of personal and cultural factors, the plurality of health factors, and the determined other factors; and wherein said generating the one or more recommendations based on the plurality of personal and cultural factors and the plurality of health factors comprises: generating the one or more recommendations based on the plurality of personal and cultural factors, the plurality of health factors, and the determined one or more other factors. In some embodiments, the first AI engine is a machine learning engine. In some embodiments, the first AI engine is further configured for: refining the first AI model based on the generated one or more recommendations. In some embodiments, the method further comprises: training the first AI model. In some embodiments, the method further comprises: retrieving one or more events as the one or more other factors; and filtering the retrieved one or more events based on nature thereof and context of the patient, to obtain one or more filtered events; and wherein said generating the one or more recommendations based on the plurality of personal and cultural factors, the plurality of health factors, and the determined one or more other factors comprises: generating the one or more recommendations based on the plurality of personal and cultural factors, the plurality of health factors, and the determined one or more filtered events. In some embodiments, said filtering the retrieved one or more events based on the nature thereof and the context of the patient comprises: analyzing the retrieved one or more events using a second AI model for obtaining a sensitivity score for each of the retrieved one or more events; determining a relevance score for each of the retrieved one or more events with respect to the patient based on the context of the patient; determining a final score for each of the retrieved one or more events based on the sensitivity score and the relevance score thereof; and suppressing at least one of the retrieved one or more events based on the final scores thereof. In some embodiments, said determining the relevance score for each of the retrieved one or more events with respect to the patient comprises: determining the relevance score for each of the retrieved one or more events with respect to the patient based on a degree of match between metadata of the event and a weighted profile of the patient. In some embodiments, said suppressing the at least one of the retrieved one or more events based on the final scores thereof comprises: for each event of the retrieved one or more events, if the final score thereof is within a first score range, allowing the event to be used for generating the one or more recommendations; if the final score thereof is within a second score range, excluding the event from being used for generating the one or more recommendations; if the final score thereof is within a third score range, excluding a first event category associated with the event from being used for generating the one or more recommendations; or if the final score thereof is within a fourth score range, excluding one or more second event categories associated with the event from being used for generating the one or more recommendations. In some embodiments, said filtering the retrieved one or more events based on the nature thereof and the context of the patient comprises: filtering the retrieved one or more events based on one or more predefined severity criteria and relevance of the one or more events to the patient. In some embodiments, the method further comprises: removing duplicated events from the retrieved one or more events. In some embodiments, the AI engine is configured for receiving structured prompts. In some embodiments, the AI engine is configured for receiving prompts comprising: clinical information of the patient; personal attributes of the patient; circumstantial traits of the patient; apprehensions of the patient; and professional background of the patient. In some embodiments, the AI engine is configured for receiving prompts comprising natural language content; and wherein the method further comprises: identifying semantic intent of the natural language content, detecting culture-specific phrasing and linguistic markers in the natural language content, and tagging extracted attributes. In some embodiments, the method further comprises: using a weighted scoring model for dynamically assigning a weight to each of the plurality of personal and cultural factors for said generating the one or more recommendations. In some embodiments, said using the weighted scoring model for dynamically assigning the weight to each of the plurality of personal and cultural factors comprises: dynamically assigning the weight to each of the plurality of personal and cultural factors based on: clinical condition, language and cultural context, interactivity domain, and stage and severity of the patient. In some embodiments, said using the weighted scoring model for dynamically assigning the weight to each of the plurality of personal and cultural factors comprises: assigning statistical weights to the plurality of personal and cultural factors; and using or more cultural validation filters for adjusting the weights of the plurality of personal and cultural factors. In some embodiments, said using the weighted scoring model for dynamically assigning the weight to each of the plurality of personal and cultural factors comprises: uses linguistic parsing and circumstantial history to infer subculture; and dynamically assigning the weight to each of the plurality of personal and cultural factors based on the inferred subculture. In some embodiments, the method further comprises: filtering the plurality of personal and cultural factors based on the assigned weights thereof. In some embodiments, the method further comprises: using a language pattern and a vocabulary model to decode intent of an input of the first AI engine. In some embodiments, the method further comprises: retraining the first AI model based on caregiver’s interactions with the patient and caregiver’s post-recommendation feedback. In some embodiments, the method further comprises: using one or more scenario-based questions in a user’s primary language for assessing the user’s language comprehension; obtaining an assessment score for the user based on said assessing the user’s language comprehension; mapping the assessment score to a communication index; and storing the communication index and a language code of the user’s primary language. In some embodiments, the method further comprises: rendering the one or more recommendations based on the communication index and the language code of the user. In some embodiments, each language is associated with a subset of available communication indices. In some embodiments, the method further comprises: caching the one or more recommendations. According to one aspect of this disclosure, there is provided one or more processors for performing the above-described method. According to one aspect of this disclosure, there is provided one or more non-transitory computer-readable storage devices comprising computer-executable instructions, wherein the instructions, when executed, cause one or more circuits to perform the above-described method. BRIEF DESCRIPTION OF THE DRAWINGS For a more complete understanding of the disclosure, reference is made to the following description and accompanying drawings, in which: FIG. 1 is a schematic diagram of a computer network system, according to some embodiments of the present disclosure; FIG. 2 is a schematic diagram showing a simplified hardware structure of a computing device of the computer network system shown in FIG.1; FIG. 3 a schematic diagram showing a simplified software architecture of a computing device of the computer network system shown in FIG.1; FIG.4 is a block diagram showing a functional structure of the computer network system shown in FIG.1, according to some embodiments of this disclosure; FIG. 5 is a schematic diagram showing the details of the recommendation engine shown in FIG.4, according to some embodiments of this disclosure; FIG. 6 is a schematic diagram showing shows an example of the health analysis module shown in FIG.5, according to some embodiments of this disclosure; FIG.7 is a schematic diagram showing shows an example of the personal analysis module shown in FIG.5, according to some embodiments of this disclosure; FIG. 8 is a schematic diagram showing shows an example of the circumstantial analysis module shown in FIG.5, according to some embodiments of this disclosure; FIG. 9 is a schematic diagram showing shows an example of the professional analysis module shown in FIG.5, according to some embodiments of this disclosure; FIG.10 is a schematic diagram showing shows an example of the apprehension analysis module shown in FIG.5, according to some embodiments of this disclosure; FIG. 11 is a schematic diagram showing the details of the recommendation engine 302, according to some embodiments of this disclosure; and FIG.12 is a screenshot showing the user interface of the computer network system shown in FIG.1 for receiving patient’s inputs and outputting generated recommendations and other action outputs, according to some embodiments of this disclosure. DETAILED DESCRIPTION Turning now to FIG. 1, a computer network system is shown and is generally identified using reference numeral 100. As shown, the computer network system 100 comprises one or more server computers 102 and a plurality of client computing devices 104 functionally interconnected by a network 108, such as the Internet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), and / or the like, via suitable wired and / or wireless networking connections. The server computers 102 may be computing devices designed specifically for use as a server, and / or general-purpose computing devices acting as server computers while also being used by various users. Each server computer 102 may execute one or more server programs. The client computing devices 104 may be portable and / or non-portable computing devices such as laptop computers, tablets, smartphones, Personal Digital Assistants (PDAs), desktop computers, and / or the like. Each client computing device 104 may execute one or more client application programs which sometimes may be called “apps”. Generally, the computing devices 102 and 104 have a similar hardware structure such as a hardware structure shown in FIG. 2. As shown, the computing device 102 / 104 comprises a processing structure 122, a controlling structure 124, one or more non-transitory computer- readable memory or storage devices 126, a network interface 128, an input interface 130, and an output interface 132, functionally interconnected by a system bus 138. The computing device 102 / 104 may also comprise other components 134 coupled to the system bus 138. The processing structure 122 may be one or more single-core or multiple-core computing processors such as INTEL®microprocessors (INTEL is a registered trademark of Intel Corp., Santa Clara, CA, USA), AMD®microprocessors (AMD is a registered trademark of Advanced Micro Devices Inc., Sunnyvale, CA, USA), ARM®microprocessors (ARM is a registered trademark of Arm Ltd., Cambridge, UK) manufactured by a variety of manufactures such as Qualcomm of San Diego, California, USA, under the ARM®architecture, or the like. When the processing structure 122 comprises a plurality of processors, the processors thereof may collaborate via a specialized circuit such as a specialized bus or via the system bus 138. The processing structure 122 may also comprise one or more real-time processors,programmable logic controllers (PLCs), microcontroller units (MCUs), μ-controllers (UCs), specialized / customized processors and / or controllers using, for example, field-programmable gate array (FPGA) or application-specific integrated circuit (ASIC) technologies, and / or the like. Generally, each processor of the processing structure 122 comprises necessary circuitries implemented using technologies such as electrical and / or optical hardware components for executing one or more processes as the implementation purpose and / or the use case maybe, to perform various tasks. In many embodiments, the one or more processes may be implemented as firmware and / or software stored in the memory 126. Those skilled in the art will appreciate that, in these embodiments, the one or more processors of the processing structure 122, are usually of no use without meaningful firmware and / or software. Of course, those skilled the art will appreciate that a processor may be implemented using other technologies such as analog technologies. The controlling structure 124 comprises one or more controlling circuits, such as graphic controllers, input / output chipsets, and the like, for coordinating operations of various hardware components and modules of the computing device 102 / 104. The memory 126 comprises one or more one or more non-transitory computer-readable storage devices or media accessible by the processing structure 122 and the controlling structure 124 for reading and / or storing instructions for the processing structure 122 to execute, and for reading and / or storing data, including input data and data generated by the processing structure 122 and the controlling structure 124. The memory 126 may be volatile and / or non-volatile, non- removable or removable memory such as RAM, ROM, EEPROM, solid-state memory, hard disks, CD, DVD, flash memory, or the like. In use, the memory 126 is generally divided into a plurality of portions for different use purposes. For example, a portion of the memory 126 (denoted as storage memory herein) may be used for long-term data storing, for example, for storing files or databases. Another portion of the memory 126 may be used as the system memory for storing data during processing (denoted as working memory herein). The network interface 128 comprises one or more network modules for connecting to other computing devices or networks through the network 108 by using suitable wired and / or wireless communication technologies such as Ethernet, WI-FI®(WI-FI is a registered trademark of Wi-Fi Alliance, Austin, TX, USA), BLUETOOTH®(BLUETOOTH is a registered trademark of Bluetooth Sig Inc., Kirkland, WA, USA), Bluetooth Low Energy (BLE), Z-Wave, Long Range (LoRa), ZIGBEE®(ZIGBEE is a registered trademark of ZigBee Alliance Corp., San Ramon, CA, USA), wireless broadband communication technologies such as Global System for Mobile Communications (GSM), Code Division Multiple Access (CDMA), Universal Mobile Telecommunications System (UMTS), Worldwide Interoperability for Microwave Access (WiMAX), CDMA2000, Long Term Evolution (LTE), 3GPP, 5G New Radio (5G NR) and / or other 5G networks, and / or the like. In some embodiments, parallel ports, serial ports, USB connections, optical connections, or the like may also be used for connecting other computing devices or networks although they are usually considered as input / output interfaces for connecting input / output devices. The input interface 130 comprises one or more input modules for one or more users to input data via, for example, touch-sensitive screens, touch-sensitive whiteboards, touch-pads, keyboards, computer nice, trackballs, microphones, scanners, cameras, and / or the like. The input interface 130 may be a physically integrated part of the computing device 102 / 104 (for example, the touch-pad of a laptop computer or the touch-sensitive screen of a tablet), or may be a device physically separated from but functionally coupled to, other components of the computing device 102 / 104 (for example, a computer mouse). The input interface 130, in some implementation, may be integrated with a display output to form a touch-sensitive screen or a touch-sensitive whiteboard. The output interface 132 comprises one or more output modules for output data to a user. Examples of the output modules include displays (such as monitors, LCD displays, LED displays, projectors, and the like), speakers, printers, virtual reality (VR) headsets, augmented reality (AR) goggles, and / or the like. The output interface 132 may be a physically integrated part of the computing device 102 / 104 (for example, the display of a laptop computer or a tablet), or may be a device physically separate from but functionally coupled to other components of the computing device 102 / 104 (for example, the monitor of a desktop computer). The computing device 102 / 104 may also comprise other components 134 such as one or more positioning modules, temperature sensors, barometers, inertial measurement units (IMUs), and / or the like. Examples of the positioning modules may be one or more global navigation satellite system (GNSS) components (for example, one or more components for operation with the Global Positioning System (GPS) of USA, Global'naya Navigatsionnaya Sputnikovaya Sistema (GLONASS) of Russia, the Galileo positioning system of the European Union, and / or the Beidou system of China). The system bus 138 interconnects various components 122 to 134 enabling them to transmit and receive data and control signals to and from each other. FIG.3 shows a simplified software architecture of the computing device 102 or 104. The software architecture comprises an application layer 162, an operating system 166, a logical input / output (I / O) interface 168, and a logical memory 172. The application layer 162, operating system 166, and logical I / O interface 168 are generally implemented as computer-executable instructions or code in the form of software programs or firmware programs stored in the logical memory 172 which may be executed by the processing structure 122. Herein, a software or firmware program is a set of computer-executable instructions or code stored in one or more non-transitory computer-readable storage devices or media such as the memory 126, and may be read and executed by the processing structure 122 and / or other suitable components of the computing device 102 / 104 for performing one or more processes. Those skilled in the art will appreciate that a program may be implemented as either software or firmware, depending on the design purposes and requirements. Therefore, for ease of description, the terms “software” and “firmware” may be interchangeably used hereinafter. Herein, a process has a general meaning equivalent to that of a method, and does not necessarily correspond to the concept of computing process (which is the instance of a computer program being executed). More specifically, a process herein is a defined method implemented as software or firmware programs executable by hardware components for processing data (such as data received from users, other computing devices, other components of the computing device 102 / 104, and / or the like). A process may comprise or use one or more functions for processing data as designed. Herein, a function is a defined sub-process or sub-method for computing, calculating, or otherwise processing input data in a defined manner and generating or otherwise producing output data. Alternatively, a process may be implemented as one or more hardware structures having necessary electrical and / or optical components, circuits, logic gates, integrated circuit (IC) chips, and / or the like. Referring back to FIG. 3, the application layer 162 comprises one or more application programs 164 executed by or performed by the processing structure 122 for performing various tasks. The operating system 166 manages various hardware components of the computing device 102 or 104 via the logical I / O interface 168, manages the logical memory 172, and manages and supports the application programs 164. The operating system 166 is also in communication with other computing devices (not shown) via the network 108 to allow the application programs 164 to communicate with programs running on other computing devices. As those skilled in the art will appreciate, the operating system 166 may be any suitable operating system such as MICROSOFT®WINDOWS®(MICROSOFT and WINDOWS are registered trademarks of the Microsoft Corp., Redmond, WA, USA), APPLE®OS X, APPLE®iOS (APPLE is a registered trademark of Apple Inc., Cupertino, CA, USA), Linux, ANDROID®(ANDROID is a registered trademark of Google Inc., Mountain View, CA, USA), or the like. The computing devices 102 and 104 of the computer network system 100 may all have the same operating system, or may have different operating systems. The logical I / O interface 168 comprises one or more device drivers 170 for communicating with respective input and output interfaces 130 and 132 for receiving data therefrom and sending data thereto. Received data may be sent to the application layer 162 for being processed by one or more application programs 164. Data generated by the application programs 164 may be sent to the logical I / O interface 168 for outputting to various output devices (via the output interface 132). The logical memory 172 is a logical mapping of the physical memory 126 for facilitating the application programs 164 to access. In this embodiment, the logical memory 172 comprises a storage memory area that may be mapped to a non-volatile physical memory such as hard disks, solid-state disks, flash drives, and / or the like, generally for long-term data storage therein. The logical memory 172 also comprises a working memory area that is generally mapped to high- speed, and in some implementations, volatile physical memory such as RAM, generally for application programs 164 to temporarily store data during program execution. For example, an application program 164 may load data from the storage memory area into the working memory area, and may store data generated during its execution into the working memory area. The application program 164 may also store some data into the storage memory area as required or in response to a user’s command. In a server computer 102, the application layer 162 generally comprises one or more server-side application programs 164 which provide(s) server functions for managing network communication with client computing devices 104 and facilitating collaboration between the server computer 102 and the client computing devices 104. Herein, the term “server” may refer to a server computer 102 from a hardware point of view, or to a logical server from a software point of view, depending on the context. As described above, the processing structure 122 is usually of no use without meaningful firmware and / or software. Similarly, while a computer system 100 may have the potential to perform various tasks, it cannot perform any tasks and is of no use without meaningful firmware and / or software. As will be described in more detail later, the computer system 100 described herein, as a combination of hardware and software, generally produces tangible results tied to the physical world, wherein the tangible results such as those described herein may lead to improvements to the computer and system themselves. In some embodiments, the computer network system 100 is a person-specific cognitive impairment care support system. FIG. 4 shows a functional structure of the person-specific cognitive impairment care support system. As shown, the system 100 comprises a recommendation engine 302 collaborating with a back-end application programming interface (API) server 304 for generating recommendations for patients’ cognitive impairment care support. The recommendation engine 302 uses a recommendation engine database 306 for facilitating the recommendation generation. The back- end API server 304 uses a user account database 308 to manage various users. Users may be from various sources and may communicate with the recommendation engine 302 and back-end API server 304 via various methods. In the example shown in FIG. 4, some users 320 may use their client-computing devices 104 to communicate with the back-end API server 304 via a front-end API server 322 and / or via third-party front-end integrations 324. Some users (not shown) may use patient care focused retrieval-augmented generation (RAG) 326 to communicate with the recommendation engine 302 via a large language model (LLM) API 328. With the support of the recommendation engine 302 and the LLM API, the patient care focused RAG 326 obtains information from the users (such as patients or caregivers), interprets the obtained information using a LLM model (not shown) via the LLM API 328 to derive the patient’s cultural and / or circumstantial context, and provides the interpretation results to the recommendation engine 302 to receive one or more recommendations for feeding back to the users of the patient care focused RAG 326. In some embodiments, the recommendation engine 302 provides recommendations based on various personal and cultural identifiers to ensure that individuals have the correct support techniques given their individual needs. This is important because it ensures that individuals with cognitive impairment such as dementia are appropriately cared for regardless of the skill or knowledge of their caregivers. FIG. 5 is a schematic diagram showing the details of the recommendation engine 302, according to some embodiments of this disclosure. As shown, the recommendation engine 302 uses a plurality of modules 402 to 410 for analyzing patient’s information and generating recommendations. More specifically, the recommendation engine 302 obtains support selection (step 422), and optionally obtains domain selection (step 424). Based on the obtained support selection and domain selection, the recommendation engine 302 analyzes patient’s preferences (step 426) and analyzes patient’s aversions (step 428), and then generates personalized recommendations (step 430). The support selection 422 includes information related to patient support, such as activities, support tips and techniques, de-escalation techniques, and / or the like. The domain selection 424 includes information related to the patient’s current and historical activities such as intellectual activities, emotional activities, physical activities, spiritual activities, social activities, occupational activities, and / or the like. TABLE 1 below summarizes the exemplary factors of the support selection 422 and the domain selection 424. TABLE 1: EXEMPLARY FACTORS OF SUPPORT SELECTION AND THE DOMAIN SELECTION Factors (Inputs) Modules Acti Sup De- Inte Em Phy Spir Soc Occ In analyzing patient’s preferences (step 426), the recommendation engine 302 uses a health module 404 for determining patient’s health information, a personal analysis module 404 for determining patient’s personal information, a circumstantial analysis module 406 for determining patient’s circumstantial information, and a professional analysis module 408 for determining patient’s professional information. In analyzing patient’s aversions (step 428), the recommendation engine 302 uses an apprehension analysis module 410. FIG.6 shows an example of the health analysis module 402, which determines a plurality of health-related information of the patient such as the patient’s clinical diagnosis 502, allergies 504, and other special considerations 506. FIG. 7 shows an example of the personal analysis module 404, which determines a plurality of personal information of the patient such as the patient’s age 512, religion 514, birth country 516, birth city 520, ethnicity 522, the languages that the patient speaks 524, preferred language 526, gender 518, and / or the like. FIG.8 shows an example of the circumstantial analysis module 406, which determines a plurality of circumstantial statuses of the patient such as the patient’s hobbies 542, growing-up history 544, community involvement 546, preferred physical activities 548, preferred sports 550, preferred music 552, preferred movies 554, preferred television shows 556, pet preference 558, and / or the like. FIG. 9 shows an example of the professional analysis module 408, which determines a plurality of profession factors of the patient such as the patient’s profession 572, education 574, and / or the like. FIG.10 shows an example of the apprehension analysis module 410, which determines a plurality of apprehension factors of the patient such as the patient’s dislikes 592, triggers 594, special fears 596, other special agitation 598, and / or the like. TABLE 2 below summarizes the exemplary factors used in analyzing the patient’s preferences (step 426) and aversions (step 428). TABLE 2: EXEMPLARY FACTORS USED IN ANALYZING THE PATIENT’S PREFERENCES AND AVERSIONS Age Reli Birt Birt Birt Eth Lan Pref Gen Clin Alle Oth Edu Prof Hobbies Circumstantial Growing Up Com Phy Spo Mus Mo Tele Pets Disl Trig Fear Agi By using above-described factors, the recommendation engine 302 may obtain a comprehensive understanding of the patient’s current and historical statuses, the patient’s personality and social characteristics, and / or the patient’s individual and cultural characteristics, and then use the patient’s individual and cultural characteristics, and clinical diagnoses to generate personalized recommendations adapted to the patient’s cognitive impairment conditions in the patient’s context. FIG. 11 is a schematic diagram showing the details of the recommendation engine 302, according to some embodiments of this disclosure. As shown, the recommendation engine 302 receives active input of support selection 602, active input of domain selection 604, and external events and / or facts 606, and generates personalized recommendations 608 and (if needed) other action outputs 610. The recommendation engine 302 comprises an active ontology module 622 controlling a plurality of other modules 632 to 642 to analyze the inputs 602 to 606 and generates the outputs 608 and 610. More specifically, the active ontology module 622 controls the following modules for analysis: • a cultural validation module 632 analyzing the patient’s cultural status based on the personal factors (see examples described above) obtained by a person module 652 and the apprehension factors (see examples described above) obtained by an apprehension module 656; • a person-specific validation module 634 analyzing the patient’s personal status based on the personal factors obtained by the person module 652, the circumstantial factors obtained by a circumstantial module 654, the apprehension factors obtained by the apprehension module 656, and the professional factors obtained by a professional module 658; and • a clinical validation module 636 analyzing the patient’s clinical diagnostic status based on the patient’s health factors obtained by a health module 660. The active ontology module 622 also controls the following modules for analyzing the patient’s natural language input: • a language pattern recognizer 638 for recognizing language patterns in the patient’s natural language input; • a language interpretation module 640 using the recognized language patterns and a vocabulary 642 for interpreting the patient’s natural language input. The active ontology module 622 combines the outputs of the modules 632 to 642 to generate personalized recommendations 608 and other action outputs 610. In these embodiments, a dialog flow processor 672 and an output processor 674 are used for formatting and / or wording the generated recommendations 608 and other action outputs 610 into a form adapted to the patient’s culture and in the patient’s culturally specific language suitable and easy for the patient to understand. FIG.12 is a screenshot showing the user interface for receiving patient’s inputs 602 to 606 and outputting the generated recommendations 608 and other action outputs 610. In some embodiments, the active ontology module 622 may learn and provide recommendations without being limited to the content stored in the recommendation engine database 306. For example, the active ontology module 622 may derive the patient’s culture and personality, and provides recommendations of suitable media (such as TV show, music, and / or the like) for the patient to use. Those skilled in the art will appreciate that it may be very technically complicated to determine, among a large variety of patient-related factors, what are important in specific contexts to generate recommendations. The technical challenges are in that, for example, not all cultures are readily accessible to the designer of the system 100. Moreover value-specific identifiable categories may not weight in the same way. For example, an Asian individuals who grew up in the urban center would have more weighting put into academic based recommendations then someone from rural Africa. Additionally, defining the boundaries on which inputs could prevent specific outputs was a challenge. Generally, while it is desirable to make the recommendation generation process an objective process, many such processes in prior art are impacted by may subjective judgements. Therefore, in some embodiments, the system 100 also comprises an artificial intelligence (AI) engine such as a machine learning engine, which is trained using various patient-related factors such as personal factors, apprehension factors, and / or the like (for example, the exemplary factors listed in TABLE 2) to properly weigh the patient-related factors and adapt the recommendation generation process to each patient’s specific characteristics and contexts for generating personalized recommendations, thereby achieving improved patient care focused RAG. In these embodiments, the AI engine considers the patient-related events (past and present) in the respective country to understand the patient’s personal contexts. The AI model also takes into account the patient’s personal and professional experiences (such as the patient’s profession, the patient’s current historical family status) in order curate the recommended outputs. In some embodiments, the AI engine also evaluates the effectiveness of the recommendations to continuously learn what recommendations are effective with different individual profiles, and refine the AI model used. In above embodiments, examples of various factors related to cognitive impairment supports are described. Those skilled in the art will appreciate that the factors described above are examples only, and in other embodiments, other factors may also be used for determining appropriate cognitive impairment supports. In some embodiments, the system 100 generates recommendations adapting to currently occurring events. For example, the system 100 uses passive sensitivity filtering based on real-time events and patient context. More specifically, the system 100, such as the AI engine thereof, operates passive sensitivity filtering as a passive, background process that does not require user input. At a fixed interval, such as once daily, multiple time each day, one or more times on predefined days, or the like, the system 100 automatically retrieves current events and / or trending topics (such as from external public sources; collective denoted “events”). The AI engine analyzes these events to determine if any content types should be omitted from that day’s personalized caregiving recommendations in order to avoid emotional, cultural, or psychological harm. If the AI engine determines that certain topics are sensitive or potentially distressing for specific individuals, the system 100 suppresses those suggestions (that is, excludes those suggestions) silently without showing or sending any alerts or disclaimers to caregivers or users, thereby avoiding generating problematic recommendations while maintaining a seamless caregiving experience. The following example illustrates how the system 100 determines what to evaluate. Each day, the system 100 gathers or otherwise retrieves social events from a set of real- time data sources, such as: • Google News API, • Google Trends, • Reddit (for example, / r / worldnews), • Twitter (X) trending topics, • Government and public advisory feeds (for example, disaster or conflict alerts), • and / or the like. Each retrieved social event is evaluated using a LLM API, which returns a structured analysis result containing: • a short summary of the event, • a list of descriptive tags (e.g., “war”, “religion”, “elder abuse”), • a Base Sensitivity Score (BSS) (such as from 0 to 25) based on a plurality of categories such as the five categories or dimensions listed in TABLE 3, Base Sensitivity Score (BSS) Matrix, • and / or the like. TABLE 3: BASE SENSITIVITY SCORE (BSS) MATRIX E ) s Ge rs Sociocultural Relevance (0-5) Whether the topic touches on religious, political, or cultural tensions As another example, the system 100 may use filtering based on connect user profiles. After scoring, the system 100 filters out any events that are not relevant to the current Connect user base, so as to reduce unnecessary processing and prevent the system 100 from evaluating topics that would never affect any users. Herein, the current Connect user base refers to a group of users registered in the system 100 and currently serviced by the system 100 as a group. As those skilled in the art will appreciate, the system 100 may provide service to multiple groups of users (each group may be a company, an organization, or the like). When providing service to the users of one of the groups, filtering out irrelevant events means filtering out events that are irrelevant to the users of this particular group (the users of this group do not need to login and use the system 100 at the same time, nor need to be “online” (that is, logged-in) when the system 100 filters out irrelevant events). Of course, when providing service to another group of users, the system 100 may determine that a similar or same event is also irrelevant to this group, and then filters out that event. For example, if an event is related to a specific country such as Norway, but no user in the group is from Norway or has the Norwegian ethnicity, the system 100 then filters out that event. This user-matching logic is governed by the personal analysis module 404, which draws from stored demographic and cultural data including: • age, • religion, • birth country, • birth province / state, • birth city, • ethnicity, • languages spoken, • preferred language, • one or more relevance modifier scores (RMS), wherein a RMS may be determined by the personal analysis module 404 using the information contained therein, or predefined such as the RMS listed in TABLE 4, RMS Matrix, • and / or the like. Herein, each RMS is for weighting the relevance between the patient and a factor or characteristic the RMS is associated therewith. Pati Birt Rel Eth Lan Pref erred anguage a c w meda s domnan anguage + Age (75+) Event relates to older populations +1 After above-described filtering (the passive sensitivity filtering based on real-time events and patient context, or the filtering based on connect user profiles), the system 100 obtains a final score for each retrieved social event. Then, the system 100 excludes or otherwise suppresses the social events based on the final scores thereof, such as excluding or otherwise suppressing the social events based on their final scores and one or more suppression thresholds. In other words, the system 100 uses the BSS (which generally determines how triggering the social events are) with the RMS (that is, the relevance of the social events) and then determines what the system 100 shall deal with the social events based on the suppression threshold matrix (STM). TABLE 5 shows an example of the STM. TABLE 5: SUPPRESSION THRESHOLD MATRIX (STM) Final Score Range (BSS + RMS) System Behavior 0–1 15– 20– 25+ In this example, if the final score of a retrieved social event is between zero (0) and 14, recommendations related to this retrieved social event are allowed. If the final score of a retrieved social event is between 15 and 19, mild suppression is used, that is, individual prompts are silently excluded. Herein, each retrieved social event is associated with a category (for example, as its primary category) and may be related to one or more other categories (for example, as its secondary categories). If the final score of a retrieved social event is between 20 and 24, moderate suppression is used, that is, the entire content of the primary category that the retrieved social event is associated therewith is omitted, for limiting potentially distressing or triggering prompts without broadly affecting the caregiving experience. The moderate suppression is a selective category suppression, wherein only the content directly related to the event (that is, the content of the event’s primary category) is excluded. For example, some tags may lead to moderate suppression. Examples of such tags are: • “Discuss today’s news”, • “Cultural conversations”, and • “Storytelling prompts about religion or politics”. In moderate suppression, indirect content (such as the content of the event’s secondary categories) remains intact. For example, as an event’s secondary categories, music recommendations, care routines, sensory activities, and / or the like are not affected in moderate suppression. By using moderate suppression, the system 100 is fine tuned to be cautious, but does not overhaul the full suggestion set. An example use case may be as follows: A caregiver of a first religion supports an elder of a second religion. There is a high-profile news story about religious violence abroad (that is, an event associated with a primary category of religion). The system 100 determines a moderate suppression of this event, and excludes prompts related to news or religion that day, while still suggesting music, bathing tips, or familiar routines. If the final score of a retrieved social event is greater than 25, severe suppression is used, that is, the content of the primary category and the one or more secondary categories that the retrieved social event is associated therewith is suppressed, thereby preventing any emotionally adjacent content from surfacing when a triggering topic is both globally intense and personally relevant to the user. The severe suppression is a broad category suppression, wherein the content of a plurality of categories is excluded. In other words, not only the content of the primary category is suppressed, adjacent or associated content types (that is, those of the secondary categories) are also removed (for example, if “war” is the event, content of “military-themed stories” or “veteran memories” may be suppressed). Moreover, the content of the following categories may also be suppressed: • “Reminiscence activities”, • “Conversations involving nationality, race, or migration”, and • “Any prompts that encourage reflection on the world”. The severe suppression thus creates a deeply filtered recommendation list that protects the user from emotional or cultural distress, particularly when their background makes them highly susceptible. Below is an example use case: A patient was born in a first country with a first religion and a first native language. A violent event occurs in a city of the first country with high global coverage. The system suppresses not only news and religion prompts, but also reminiscing about home country, conversations about identity, or any content linked to conflict, resistance, or geography. Below shows an example output: { "event": "Middle East conflict escalates", "base_score": 22, "relevance_modifiers": { "birth_country": 3, "religion": 2, "language": 1 }, "final_score": 28, "suppressed_tags": ["war", "religion", "current events"], "caregiver_alert_displayed": false } As can be seen, this example output indicates that multiple categories ("war", "religion", and "current events") are suppressed. In some embodiments, the system 100 uses a two-stage filtering method for solving the overfiltering or inaccurate suppression issue. In these embodiments, the system 100 first filters the retrieved events based on one or more global severity criteria predefined for the main categories that the retrieved events belong to. Then, the system 100 assesses the relevance of the remaining events with respect to the corresponding user to obtain personalized relevance, and filters the remaining events based on their personalized relevance. In some embodiments, the system 100 uses structured prompts, format validation, feedback loop auditing, and / or the like for solving the issue of LLM hallucination or inconsistent tagging. More specifically, instead of using natural-language or loosely framed prompts, the system 100 sends strictly templated prompts that request, for example: • a fixed number of tags (for example, 3 to 5 maximum), • scores for five specific dimensions (for example, emotional intensity, media saturation, etc.), • a one-paragraph summary. The following is an example of a structured prompt: “You are a safety classifier for dementia caregiving. Given this event, return a JSON with: • summary • tags (list of 3–5 thematic labels) • scores (five named fields, 0–5 each) • total_score (sum of all five scores).” The structured prompt improves predictability and consistency across calls. Accordingly, when receiving a prompt, the system 100 validates its format before using the prompt for further processing. The feedback loop auditing is related to AI model training. A subset of outputs is logged and reviewed (such as manually reviewed) to evaluate LLM behavior over time. If hallucinated or missed tags are discovered, examples are collected and used to refine the prompt. An internal quality score is tracked to determine when prompt tuning or model version updates are needed. Feedback loop auditing creates a learning system that gets smarter and more stable over time. In some embodiments, the system 100 uses event deduplication (that is, determining and removing duplicated events), prioritization-reduced LLM load (lowering priorities of some LLM loads based on, e.g., one or more predefined criteria), and / or the like for reducing the latency during daily event evaluation. In some embodiments, the system 100 uses dynamically filled prompts for the AI engine to preserve user experience so as to prevent users from misunderstanding or missing content. More specifically, when the system 100 suppresses or otherwise blocks a recommendation, the blocked or suppressed recommendation is replaced with an alternative recommendation that is appropriate. In some embodiments, the system 100 automatically generates recommendations for users. For example, to implement the automated recommendation generation, the system 100 accepts structured and unstructured data inputs such as: • clinical information: diagnoses (for example, Alzheimer’s disease, frontotemporal dementia), allergies, medication restrictions, and / or the like; • personal attributes: age, birthplace, ethnicity, languages spoken and / or preferred, gender identity, and / or the like; • circumstantial traits: hobbies, cultural upbringing, lifestyle, favorite music or films, and / or the like; • apprehensions: fears, dislikes, sensory triggers, sources of agitation, and / or the like; • professional background: occupation, education, career-based knowledge biases, and / or the like; • and / or the like. In some embodiments, the system 100 standardizes and interprets inputs (such as above- described data inputs). More specifically, a preprocessing module maps user input into a normalized schema. Natural language responses are interpreted using a language interpretation module (for example, via LLM API integration), which: • identifies semantic intent, • detects culture-specific phrasing and linguistic markers (for example, formal Hindi-influenced English vs. Caribbean Creole), and • tags extracted attributes for downstream use. In some embodiments, the system 100 uses a weighted scoring model for automatically generating recommendations. By using the weighted scoring model, the system 100 dynamically assigns a weight to each factor based on: • the clinical condition (for example, diagnosis of frontotemporal dementia increases the weight of behavioral and aversion-related inputs); • the language and cultural context (for example, non-verbal cues are weighted differently for Indigenous vs. East Asian profiles); • the interactivity domain (for example, emotional or spiritual support contexts elevate preferences over clinical rigidity); • the stage and severity of dementia, derived from historical entries and clinical notes; • and / or the like. This scoring model acts as an intermediate decision layer between input parsing and recommendation selection. Each possible recommendation is assigned a relevance score based on the degree of match between its metadata (such as tags, conditions, applicability, and / or the like) and the patient’s currently weighted profile vector. A threshold is enforced to filter out recommendations that fall below a clinical or cultural acceptability score threshold. In some embodiments, the system 100, or more specifically the AI engine thereof, infers personalized contexts via suitable ontology modules. For example, the system 100 may use an active ontology module to arrange: • cultural validation: to ensure output is aligned with the patient’s norms, such as avoiding direct eye contact for certain Indigenous individuals; • person-specific validation: to incorporate circumstantial and professional context (for example, engineers may respond better to logic-driven activities); • clinical validation: to confirm therapeutic appropriateness; • and / or the like. In some embodiments, the system 100 automatically generates tailored recommendations. By using validated inputs and scores, the system 100 may: • select and rank appropriate recommendations; • apply context-specific filters (such as avoiding showing content involving dogs to users flagged as fearful); • format results into appropriate language and cultural tone via a dialog flow processor and output formatter; • and / or the like. In some embodiments, the system 100 uses the weighted scoring model and dynamically reprioritizes (for example, dynamically re-weights) the factors for solving the issues related to conflicting traits (such as strong cultural aversions but clinical indications suggesting exposure is helpful) in patients’ profiles (that is, balancing competing factors). For example, if a frontotemporal diagnosis is present, agitation triggers are given higher weight to prevent harm. In some embodiments, the system 100 uses a dual-layer scoring process for solving the bias in output (for example, default LLM and machine learning (ML) models may overrepresent dominant (Western, English-speaking) caregiving norms. In these embodiments, the system 100 first uses statistical weights tuned to diverse population profiles, and then applies one or more post-generation cultural validation filters (containing, for example, one or more culture-based validation criteria) to adjust the weights of the factors, to reduce or even eliminate the bias in output. In these embodiments, the AI models (such as the LLM and / or ML models) are fine-tuned in training using representative datasets. In some embodiments, the system 100 uses linguistic parsing and circumstantial history to reduce the intra-national cultural variation and solve the issue that the country-level cultural tagging may be otherwise too coarse. More specifically, the system 100 uses linguistic parsing and circumstantial history to infer subculture, and assign internal scoring differentials. For example, a Tamil speaker from Chennai with a British educational background would activate different weights than a Punjabi Sikh from Amritsar. In some embodiments, the system 100 uses a configurable threshold to solve the issue of oversaturation of recommendations (that is, without constraints, the system 100 may otherwise generate too many or poorly targeted recommendations). In these embodiments, the weighted score must exceed a configurable threshold to be used or otherwise taken into account, which ensures that only high-confidence, validated recommendations are presented. In some embodiments, the system 100 incorporates a language pattern recognizer and vocabulary model to decode intent, and re-weight inputs accordingly. This prevents over-reliance on specific-language-centric constructs such as English-centric constructs, thereby solving the difficulty in interpreting natural language inputs for non-native speakers, wherein users often provide incomplete or culturally coded language responses. In some embodiments, the scoring model may be implemented as follows. 1. Scoring Vector Representation Each patient profile is encoded as a multi-dimensional vector:" = [()$), (*$*, … , (.$.] (1)where: • $, (i = 1, 2, …) is a normalized feature value (for example, frontotemporal diagnosis = 1, Alzheimer’s = 0, etc.); • (,(i = 1, 2, …) is the learned or predefined weight for that feature; • " [0, 1] or having a value determined based on a categorical mapping (for example, language = “Punjabi” giving rise to f = 0.75 under culture sensitivity dimension). 2. Recommendation Match Score Each potential recommendation is associated with a weight vector #&(j = 1, 2, …): #- = ['-), '-*, … , '-.] (2)where '&%reflects how suitable recommendation #&is for feature i. The match score is calculated as: Score-(3) Only scores above a threshold (for on a zero to one are to users. An example is described as follows: In this example, the patient: • has frontotemporal dementia ($)= 1, () = 0.4);• speaks Cree ($*= 1, (* = 0.2); and• has an aversion to dogs ($+= 1, (+ = 0.4).In this example, the recommendation weight vector for this patient is R = [1, 0.5, 0]. Then, the match score for Score = = = 0.5In this example, the threshold is 0.65. As the match score 0.5 is smaller than the threshold of 0.65, the recommendation of “therapy dog visits” is discarded. On the other hand, a recommendation for “nature walk with music” (having a weight vector R = [1, 1, 1] ) gives Score = / (, $, ', = (0.4 " 1 " 1) + (0.2 " 1 " 1) + (0.4 " 1 " 1) = 1.0As the match score of 1.0 is greater than the threshold of 0.65, the recommendation for “nature walk with music” is accepted. In some embodiments, the system 100 uses weight calibration, wherein weights may be: • static, defined by, for example, clinical guidance (such as from gerontologists, neurologists, and / or the like); and / or • dynamic, adjusted via machine learning (for example, reinforcement learning based on caregiver feedback, success logs, aversion incidents, and / or the like). In some embodiments, a retraining step is triggered periodically using caregiver interactions and post-recommendation feedback (explicit thumbs up / down or inferred from outcomes) for training the AI engines / models used in the system 100. In some embodiments, to quantify relevance, the system 100 converts both patient inputs and candidate recommendations into weighted feature vectors. Each recommendation is scored using a dot-product similarity function between the patient profile vector and the recommendation vector. Only those exceeding a configurable clinical confidence threshold are shown. This vector- based method ensures that outputs are mathematically consistent, interpretable, and tunable over time. In some embodiments, the system 100 uses adaptive language and literacy personalization for generating caregiver recommendations. The adaptive language and literacy personalization enables multi-format, multi-language, and multi-literacy recommendations rendering to accommodate caregivers with varying language proficiencies and comprehension levels. Thus, in these embodiments, the system 100 may adapt language to the need of the user, wherein the term “language” refers to the natural language such as English, Spanish, and / or the like, and refers to the way the communication is conducted (for example, elementary English proficiency for certain users and more advanced language proficiency for others). Although the clinical recommendation is generated once based on the person living with dementia (for example, via BSS, RMS, and STM scoring (see Table 4 for an example)), each caregiver receives a personalized rendering of that same recommendation, adapted to their individual comprehension ability and language preference. In these embodiments, the system 100 uses caregiver profile vectorization to assign a persistent communication profile to each caregiver, wherein the persistent communication profile comprises a plurality of fields such as: • preferred_language: ISO 639-1 language code, • communication_index (CI): Integer 1–5 representing literacy comprehension tier, • assessment_score: Raw score from initial language comprehension test (0–100). Below shows an example of the persistent communication profile: { "caregiver_id": "12345", "preferred_language": "English", "secondary_language": "Tagalog", "communication_index": 3, "modality_preferences": ["text", "image"], "visual_support": true, "assessment_score": 67, "last_modified": "2025-06-13T08:00:00Z" } In these embodiments, the system 100 uses onboarding language comprehension assessment. As those skilled in the art understand, self-reported proficiency is unreliable, especially for non-native speakers. Instead, the system 100 includes an adaptive comprehension quiz, such as a five-question adaptive comprehension quiz, delivered during onboarding or when adding a new caregiver, for onboarding language comprehension assessment. The adaptive comprehension quiz comprises one or more scenario-based questions translated into the user’s primary language. For ease of use, the one or more questions are in multiple-choice format with visual support where possible. The one or more questions cover both vocabulary interpretation (for example, “mobility”, “hydration”), and sentence comprehension and intent parsing An example question is: “If a care suggestion says: ‘Encourage hydration,’ what should you do?” A. Remind them to drink more water B. Take them for a walk C. Turn off the lights D. Call their doctor The assessment score is mapped to a communication index (communication_index), for example: • CI 1: Extremely limited literacy / visual only, • CI 2: Basic literacy (grade 3–5 equivalent), • CI 3: Intermediate (grade 6–9), • CI 4: Fluent conversational, • CI 5: Clinical or professional-level comprehension. In some embodiments, the system 100 uses a multi-variant recommendation library, wherein each core recommendation is stored therein as a set of rendered variants, indexed by: • language_code (e.g., en, fr, crk, es), • communication_index (1–5), • modality: text / audio / image / video. Below is an example: { "recommendation_id": "R-00321", "base_vector": [...], "variants": [ { "language_code": "en", "communication_index": 5, "text": "Increase mobilization through passive flexibility routines to delay sarcopenia." }, { "language_code": "en", "communication_index": 2, "text": "Help them stretch so they stay strong and don’t get stiff." }, { "language_code": "crk", "communication_index": 2, "audio_url": " / audio / crk / R-00321_ci2.mp3" } ] } In some embodiments, the system 100 uses a dynamic rendering pipeline, wherein, each time when a recommendation is displayed or pushed to a caregiver, the system matches the caregiver profile (caregiver_profile) to a recommendation variant for a specific language code and a specific communication index (recommendation.variant[language_code + CI]) for rendering the output. If a perfect match is not found, the fallback hierarchy (that is, the order for selecting the language of the output) is: 1. Same language, lower CI, 2. Same language, higher CI, 3. Caregiver’s secondary language, 4. System default (CI 3, English). In some embodiments, the system 100 uses a lightweight language diagnostic quiz to assign communication index levels objectively, for solving the issue that self-reporting may be inconsistent and culturally biased. In some embodiments, the system 100 uses a recommendation variant architecture (recommendation.variant[]) that has the same base logic and multiple output styles to ensure message consistency while enabling personalization, thereby solving the issue that care teams needed to view the same recommendation, but each caregiver had different literacy and language needs. In some embodiments, the system 100 may limit the communication indices for each language, for example, only CI 2, 3, and 5 variants are required per language, wherein CI 1 uses icons / audio, and CI 4 can fallback to CI 5 with formatting simplification, for solving the issue of variant explosion and storage bloat for multiple CI / language combinations. In some embodiments, the system 100 may pre-cache one or more recommendations, such as next 10 recommendations, per caregiver upon login using language and CI filter, for reducing the latency when rendering large multilingual teams. some embodiments, the system 100 uses standardized “caregiver-safe” glossary to convert complex terms to plain language at CI # 3, for solving the issue that caregivers may otherwise misinterpret clinical terms (for example, “hydration,” “agitation”). With above methods, the system 100 delivers the same underlying care recommendation to all caregivers supporting a person with dementia, while dynamically tailoring the rendered message based on each caregiver’s communication profile. This profile is derived from an embedded comprehension assessment and includes a communication index (CI), preferred languages, and content modality preferences. Each recommendation is stored in a structured format with variants indexed by language and CI level. The output layer matches the caregiver profile to the most appropriate variant and delivers the corresponding rendering (e.g., text, visual, or audio). This ensures that all caregivers—regardless of language fluency or educational background—receive clinically aligned guidance in a format they can understand and act upon. While dementia is used as an example for describing above embodiments, the methods disclosed herein may also or alternatively used for patients of other diseases. In above embodiments, a computer network system is used as the person-specific cognitive impairment care support system. In some other embodiments, the person-specific cognitive impairment care support system such as the functionalities and methods described above, may be implemented in a single computing device such as a single client computing device 102 or a single server computer 104. Herein, various embodiments of person-specific cognitive impairment care support system and methods are described. In some embodiments, the methods disclosed herein may be implemented as one or more circuits of a module, a device, an apparatus, a system, and / or the like. In some embodiments, the methods disclosed herein may be implemented as computer-executable instructions stored in one or more non-transitory computer-readable storage devices such that, the instructions, when executed, may cause one or more circuits to perform the methods disclosed herein. Those skilled in the art will appreciate that, cognitive impairment (such as dementia) supports vary depending on cultural and personal characteristics. Therefore, clinical discretion is usually utilized for determining the best support techniques. The person-specific cognitive impairment care support system and related methods disclosed herein take into account patients’ cultural and personal characteristics, and provide various advantages, such as: • allows culturally, personally and clinically specific care for an individual, regardless of the experience or qualifications of the caregiver (user); • providing personalized recommendations using an objective analytical process (which is usually a subject process in prior art); • alleviating the patient’s burden to have detailed understanding of the types of cognitive impairment (such as dementias) in order to understand what approach would be appropriate; • taking into account cultural implications that are important for a person with dementia, while prior-art systems fail to consider such implications. Generally, individuals living with dementia lost their short-term memory first, while retaining their long term, which means they retain the long term cultural and person information. Therefore, the system and methods disclosed herein provide improved care to patient by taking into account cultural implications. Those skilled in the art will appreciate that the various embodiments and / or features described herein may be customized and / or combined as needed or desired. For example, one or more features in an embodiment may be excluded or otherwise omitted as needed or desired. As another example, one or more features in one embodiment may be combined with one or more features of one or more other embodiments as needed or desired. Moreover, although embodiments have been described above with reference to the accompanying drawings, those of skill in the art will appreciate that variations and modifications may be made without departing from the scope thereof as defined by the appended claims.

Claims

WHAT IS CLAIMED IS:

1. A computerized method comprising:obtaining a plurality of personal and cultural factors of a patient; obtaining a plurality of health factors of the patient; and generating one or more recommendations based on the plurality of personal and cultural factors and the plurality of health factors.

2. The method of claim 1, wherein the plurality of personal and cultural factors comprise:personal factors; professional factors; circumstantial factors; apprehension factors; or a combination thereof.

3. The method of claim 1 or 2, wherein the plurality of health factors comprise:clinical diagnoses; allergies; or a combination thereof.

4. The method of any one of claims 1 to 3, wherein the personal factors comprise:age; religion; birth place; ethnicity; languages spoken; preferred language; gender; or a combination thereof.

5. The method of any one of claims 1 to 4, wherein the professional factors comprise:education; profession; ora combination thereof.

6. The method of any one of claims 1 to 5, wherein the circumstantial factors comprise:hobbies; growing-up history; community involvement; physical activities; sports; music; movies; television; pets; or a combination thereof.

7. The method of any one of claims 1 to 6, wherein the apprehension factors comprise:dislikes; triggers; fears; agitations; or a combination thereof.

8. The method of any one of claims 1 to 7 further comprising:using a first artificial intelligence (AI) engine and a corresponding first AI model for: determining one or more other factors; and weighing the plurality of personal and cultural factors, the plurality of health factors, and the determined other factors; and wherein said generating the one or more recommendations based on the plurality of personal and cultural factors and the plurality of health factors comprises: generating the one or more recommendations based on the plurality of personal and cultural factors, the plurality of health factors, and the determined one or more other factors.

9. The method of claim 8, wherein the first AI engine is a machine learning engine.

10. The method of claim 8 or 9, wherein the first AI engine is further configured for:refining the first AI model based on the generated one or more recommendations.

11. The method of any one of claims 8 to 10 further comprising:training the first AI model.

12. The method of any one of claims 8 to 11 further comprising:retrieving one or more events as the one or more other factors; and filtering the retrieved one or more events based on nature thereof and context of the patient, to obtain one or more filtered events; and wherein said generating the one or more recommendations based on the plurality of personal and cultural factors, the plurality of health factors, and the determined one or more other factors comprises: generating the one or more recommendations based on the plurality of personal and cultural factors, the plurality of health factors, and the determined one or more filtered events.

13. The method of claim 12, wherein said filtering the retrieved one or more events based on thenature thereof and the context of the patient comprises: analyzing the retrieved one or more events using a second AI model for obtaining a sensitivity score for each of the retrieved one or more events; determining a relevance score for each of the retrieved one or more events with respect to the patient based on the context of the patient; determining a final score for each of the retrieved one or more events based on the sensitivity score and the relevance score thereof; and suppressing at least one of the retrieved one or more events based on the final scores thereof.

14. The method of claim 13, wherein said determining the relevance score for each of the retrievedone or more events with respect to the patient comprises:determining the relevance score for each of the retrieved one or more events with respect to the patient based on a degree of match between metadata of the event and a weighted profile of the patient.

15. The method of claim 13 or 14, wherein said suppressing the at least one of the retrieved oneor more events based on the final scores thereof comprises: for each event of the retrieved one or more events, if the final score thereof is within a first score range, allowing the event to be used for generating the one or more recommendations; if the final score thereof is within a second score range, excluding the event from being used for generating the one or more recommendations; if the final score thereof is within a third score range, excluding a first event category associated with the event from being used for generating the one or more recommendations; or if the final score thereof is within a fourth score range, excluding one or more second event categories associated with the event from being used for generating the one or more recommendations.

16. The method of any one of claims 12 to 15, wherein said filtering the retrieved one or moreevents based on the nature thereof and the context of the patient comprises: filtering the retrieved one or more events based on one or more predefined severity criteria and relevance of the one or more events to the patient.

17. The method of any one of claims 12 to 16 further comprising:removing duplicated events from the retrieved one or more events.

18. The method of any one of claims 8 to 17, wherein the AI engine is configured for receivingstructured prompts.

19. The method of any one of claims 8 to 18, wherein the AI engine is configured for receivingprompts comprising: clinical information of the patient; personal attributes of the patient; circumstantial traits of the patient;apprehensions of the patient; and professional background of the patient.

20. The method of any one of claims 8 to 19, wherein the AI engine is configured for receivingprompts comprising natural language content; and wherein the method further comprises: identifying semantic intent of the natural language content, detecting culture-specific phrasing and linguistic markers in the natural language content, and tagging extracted attributes.

21. The method of any one of claims 8 to 20 further comprising:using a weighted scoring model for dynamically assigning a weight to each of the plurality of personal and cultural factors for said generating the one or more recommendations.

22. The method of claim 21, wherein said using the weighted scoring model for dynamicallyassigning the weight to each of the plurality of personal and cultural factors comprises: dynamically assigning the weight to each of the plurality of personal and cultural factors based on: clinical condition, language and cultural context, interactivity domain, and stage and severity of the patient.

23. The method of claim 21 or 22, wherein said using the weighted scoring model for dynamicallyassigning the weight to each of the plurality of personal and cultural factors comprises: assigning statistical weights to the plurality of personal and cultural factors; using or more cultural validation filters for adjusting the weights of the plurality of personal and cultural factors.

24. The method of any one of claims 21 to 23, wherein said using the weighted scoring model fordynamically assigning the weight to each of the plurality of personal and cultural factors comprises:uses linguistic parsing and circumstantial history to infer subculture; and dynamically assigning the weight to each of the plurality of personal and cultural factors based on the inferred subculture.

25. The method of any one of claims 21 to 24 further comprising:filtering the plurality of personal and cultural factors based on the assigned weights thereof.

26. The method of any one of claims 8 to 25 further comprising:using a language pattern and a vocabulary model to decode intent of an input of the first AI engine.

27. The method of any one of claims 8 to 26 further comprising:retraining the first AI model based on caregiver’s interactions with the patient and caregiver’s post-recommendation feedback.

28. The method of any one of claims 8 to 27 further comprising:using one or more scenario-based questions in a user’s primary language for assessing the user’s language comprehension; obtaining an assessment score for the user based on said assessing the user’s language comprehension; mapping the assessment score to a communication index; and storing the communication index and a language code of the user’s primary language.

29. The method of claim 28 further comprising:rendering the one or more recommendations based on the communication index and the language code of the user.

30. The method of claim 28 or 29, wherein each language is associated with a subset of availablecommunication indices.

31. The method of any one claims 1 to 30 further comprising:caching the one or more recommendations.

32. One or more processors for performing the method of any one of claims 1 to 31.

33. One or more non-transitory computer-readable storage devices comprising computer-executable instructions, wherein the instructions, when executed, cause one or more circuits to perform the method of any one of claims 1 to 31.

Images