AI personalized health education system for elderly patients with head and neck cancer

Abstract

The invention belongs to the technical field of medical health management, and particularly relates to an AI personalized health education system for elderly patients with head and neck cancer, which comprises a content generation and management module, a personalized push engine, an interaction and artificial triage module, an elderly friendly interaction module and a data monitoring and safety module, by adopting an AI + manual hierarchical response mechanism and combining elderly friendly interactive design and operation training support, the participation obstacles that elderly patients are inconvenient to move and poor in informatization ability are broken through, and the intervention compliance and sustainability are improved; according to the intervention method, strict double-group random control test design is adopted, multi-dimensional and multi-time-point effect evaluation is combined with a scientific statistical analysis method, the improvement effect of AI-assisted health education on cancer recurrence, fear, psychological state and life quality of a patient can be accurately verified, and a solid evidence-based basis is provided for clinical popularization.

Description

Technical Field

[0001] This invention relates to the field of medical and health management technology, specifically to an AI-based personalized health education system for elderly patients with head and neck cancer. Background Technology

[0002] Elderly patients with head and neck cancer refer to those aged ≥60 years with malignant tumors concentrated in the head and neck region, including the oral cavity, pharynx, nasal cavity, sinuses, thyroid, etc. They are an important part of the head and neck cancer population, which is gradually increasing in proportion. These patients are often induced by factors such as long-term smoking, drinking, chronic inflammation, or HPV infection, and often have underlying conditions such as hypertension, diabetes, and cardiovascular disease, which increases the complexity of their condition.

[0003] In clinical practice, early symptoms of head and neck cancer in elderly patients are easily overlooked. These symptoms often manifest as a foreign body sensation in the throat, hoarseness, difficulty swallowing, painless lumps in the neck, or persistent oral ulcers. By the time of diagnosis, the cancer is often in the middle or late stages, making treatment more difficult. Due to declining physiological functions, they have lower tolerance for conventional treatments such as surgery, radiotherapy, and chemotherapy, and are prone to adverse reactions such as bone marrow suppression, decreased immunity, and organ damage. Individualized treatment plans are necessary. At the same time, the psychological state and quality of life of elderly patients need to be given special attention. The physical pain caused by the disease, the stress of treatment, and the impact on their ability to live independently can easily lead to emotional problems such as anxiety and depression. They require dual support from family care and medical care to ensure both treatment effectiveness and quality of life in their later years.

[0004] Existing health education models suffer from severe homogenization of educational content and a lack of personalized strategies for different diseases and treatment stages; insufficient follow-up continuity and lack of support during critical rehabilitation windows; and low intervention compliance and limited actual benefits due to mobility limitations and operational difficulties for elderly patients. Although some digital health interventions have shown potential in cancer rehabilitation, most suffer from heterogeneous intervention components, opaque algorithms, or a lack of age-appropriate design, making it difficult to promote and apply them among elderly head and neck cancer patients. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an AI-powered personalized health education system for elderly patients with head and neck cancer. This system solves several problems with existing health education models, including severe homogenization of educational content, a lack of personalized strategies for different diseases and treatment stages, insufficient follow-up continuity, and a lack of support during critical recovery windows. Furthermore, elderly patients often struggle with mobility and operational limitations, making it difficult for them to participate in offline or complex digital interventions, resulting in low intervention compliance and limited actual benefits. While some digital health interventions have shown potential in cancer rehabilitation, most suffer from heterogeneous intervention components, opaque algorithms, or a lack of age-appropriate design, hindering their widespread application among elderly head and neck cancer patients.

[0006] To achieve the above objectives, this invention provides the following technical solution: It includes a content generation and management module, a personalized push engine, an interactive and manual triage module, an elderly-friendly interaction module, and a data monitoring and security module. The content generation and management module involves physicians and specialist nurses collaboratively creating and reviewing health education content, covering video, text, and animation formats. It establishes a knowledge base containing specific content for different pre-operative and post-operative stages and different disease types, and implements regular content updates and version control. The personalized push engine employs a rule-based intelligent algorithm to accurately push content based on the patient's treatment stage (pre-operative, 1 month, 3 months, 6 months, 12 months post-operative), disease type (laryngeal cancer / oral cancer / nasopharyngeal cancer / thyroid cancer, etc.), and symptom assessment results. Simultaneously, it triggers enhanced content pushes based on patient interaction feedback. The interactive and manual triage module features AI-powered rapid response (≤30 seconds), and provides background triage for unanswerable questions or red flag signals. It fulfills the service commitment of an average nurse response time of <5 minutes during working hours (8:00-20:00) and processing by 9:00 the next day during off-hours, establishing a nurse-physician tiered response mechanism. The elderly-friendly interaction module uses an interface design with a font size ≥18px and a contrast ratio ≥7:1, supports voice questions and one-click access to Q&A, provides 1:1 operation training, and a family proxy login channel. The data monitoring and security module records patient behavior indicators (weekly active days, video completion rate, number of interactions), triggers intervention reminders for patients with low compliance, and uses anonymized IDs, encrypted data storage, and tiered permission management to ensure privacy and security.

[0007] As a preferred technical solution of the present invention, the content creation process of the content generation and management module (1) includes: determining the topic and outline through a weekly co-creation meeting of "physician + specialist nurse", the responsible nurse writing the popular version of the content (≤800 words), the attending physician reviewing the medical accuracy, the head nurse reviewing the readability and elderly friendliness, marking the version number, release date and ≤1 month of the proposed review date for the content, and completing the rapid review within 48 hours when the guidelines are updated.

[0008] As a preferred technical solution of the present invention, the push rules of the personalized push engine (2) include: configuring corresponding knowledge packages according to the treatment stage, setting specific rehabilitation content according to the disease type (such as focusing on swallowing and articulation training for laryngeal cancer / oral cancer), pushing enhanced content after assessing the patient's symptoms by telephone or WeChat video, and intelligently recommending medical treatment channels for emergencies (nearby emergency room at night, daytime hospital visit).

[0009] As a preferred technical solution of the present invention, the hierarchical response mechanism of the interactive and manual triage module is as follows: nursing-manageable problems (nursing guidance, medication routine, functional exercises) are handled in a closed loop by specialist nurses; complex problems are answered and suggested by associate chief physicians; and patients with complex symptoms are directly recommended to register for a follow-up visit.

[0010] As a preferred technical solution of the present invention, the following steps are included:

[0011] S1: Screening and grouping of study subjects: Patients aged ≥60 years who were pathologically diagnosed with head and neck malignant tumors and had completed radical surgery, had independent communication ability, and had a PHQ-9 score ≥5 or a GAD-7 score ≥5 were selected and randomly assigned to the intervention group and the control group in a 1:1 ratio, with 50 patients in each group and a total sample size of 100 patients.

[0012] S2: Intervention Implementation:

[0013] Intervention Group: Intervention was implemented through the AI-assisted health education system, covering the period from preoperative to 12 months postoperatively. Thematic content was pushed twice a week at fixed times (Tuesday or Saturday), and real-time interactive Q&A was provided daily from 8:00 to 20:00. Preoperative education and anxiety self-help techniques were pushed. During the postoperative hospitalization period, short videos related to nursing care, pain management, and rehabilitation training were pushed daily. After discharge, swallowing / articulation training, nutrition management, sleep and mood regulation, and preparation for follow-up visits were pushed in stages. Check-in screening was carried out at 1, 4, 8, and 12 weeks postoperatively, and nurses followed up within 48 hours for high-risk patients.

[0014] Control group: Standardized text messages were sent at the same frequency as the intervention group, including perioperative education, dietary guidelines, rehabilitation exercise guidelines, and follow-up visit reminders. Consultations and suggestions beyond the scope of the text messages were sent to offline clinics or routine nursing channels.

[0015] S3: Effectiveness Assessment: At baseline (T0, admission date), 1 month post-surgery (T1), 3 months post-surgery (T2), 6 months post-surgery (T3), and 12 months post-surgery (T4), a multidimensional assessment was conducted using the WHO-QoL-Old scale, the PSS stress perception scale, the FoP-Q-SF cancer recurrence fear scale, the PHQ-9 depression assessment scale, the GAD-7 anxiety assessment scale, and the MMSE cognitive function scale.

[0016] S4: Statistical Analysis of Data: Descriptive statistics, independent samples t-test, repeated measures ANOVA, and covariance analysis (controlling for baseline differences) are used to analyze the assessment data and verify the intervention effect.

[0017] According to claim 5, the method of an AI-based personalized health education system for elderly patients with head and neck cancer is characterized in that the exclusion criteria for research subjects in step S1 include: comorbid severe mental illness or cognitive impairment, comorbid severe heart, liver, or kidney failure, postoperative plans to participate in other large-scale intervention trials, inability to use a smartphone or lack of information receiving conditions, and other situations deemed unsuitable for enrollment by the researchers.

[0018] As a preferred technical solution of the present invention, the system activation process of the intervention group in S2 includes: within 24 hours of patient admission, a 15-minute 1:1 operation training (including family members) is conducted, and the patient completes the onboarding tasks of following the official account, playing the video, and sending questions, and an illustrated operation manual is distributed.

[0019] As a preferred technical solution of the present invention, the assessment implementation process of S3 is as follows: In T0 and T1, trained nurses guide patients to fill out questionnaires face-to-face in the ward (reading questions aloud if necessary); in T2-T4, questionnaires are pushed out via WeChat mini program first, and those who have not completed the questionnaires are followed up by telephone within 48 hours for supplementary testing. The assessment environment should be quiet, and the duration of each test should be ≤15 minutes. Tests should be avoided within 30 minutes after the use of the analgesia pump. The assessor is a nurse who did not participate in the intervention. Anonymous ID is used to administer the test and enter the data. Double entry and 5% random review are implemented.

[0020] As a preferred technical solution of the present invention, it also includes compliance management measures: patients in the intervention group automatically receive points after completing the viewing and check-in tasks, and can redeem rehabilitation-related gifts when the points reach the target; for patients who have no activity for 7 consecutive days or whose video completion rate is <30%, telephone follow-up is automatically triggered, and compliance is improved through system reminders, one-on-one guidance from nurses, assistance from family members, and bedside re-education in sequence.

[0021] As a preferred technical solution of the present invention, it also includes an adverse event handling procedure: if a serious adverse event occurs (such as a clear suicidal tendency or an acute mental disorder onset), the patient's participation in the study shall be immediately suspended, and the case shall be handled in accordance with the hospital's emergency procedures. After evaluation by the on-duty physician, the patient shall be referred to a psychiatric or specialized medical institution, and the event shall be recorded and reported to the ethics committee.

[0022] Compared with existing technologies, this invention provides an AI-based personalized health education system for elderly patients with head and neck cancer, which has the following beneficial effects:

[0023] 1. This AI-based personalized health education system for elderly patients with head and neck cancer overcomes the participation barriers caused by elderly patients' limited mobility and poor information literacy by adopting a graded response mechanism of AI + human intervention, combined with elderly-friendly interactive design and operation training support, thereby improving the adherence and sustainability of the intervention. The intervention method adopts a rigorous two-group randomized controlled trial design.

[0024] 2. This AI-powered personalized health education system for elderly patients with head and neck cancer can accurately verify the effects of AI-assisted health education on patients' fear of cancer recurrence, psychological state, and quality of life through multi-dimensional and multi-time-point effect evaluation, combined with scientific statistical analysis methods, providing solid evidence-based basis for clinical promotion. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0026] Example

[0027] Example 1: Construction of an AI-Assisted Health Education System

[0028] Content Generation and Management Module Construction: A content creation team composed of head and neck surgeons and specialist nurses was formed. Based on the *American Cancer Society Guidelines for the Care of Head and Neck Cancer Survivors*, the *European Association for Head and Neck Cancer Recommendations for the Care of Head and Neck Cancer Survivors*, and domestic guidelines for the diagnosis and treatment of head and neck tumors, and combined with frequently asked clinical questions, core content directions were determined, including preoperative education, postoperative care, pain management, swallowing and articulation training, nutritional guidance, emotional regulation, and preparation for follow-up visits. A co-creation meeting was held every Tuesday to determine topics and outlines. The responsible nurse wrote a simplified version of the content (≤800 words), the attending physician reviewed the medical accuracy, and the head nurse reviewed readability and age-friendliness. Content was produced in the form of videos (3-5 minutes), text and graphics (presented in bullet points, highlighting key points), and animations (visualizing complex processes), and marked with version number, release date, and review date. A routine review was conducted monthly, and a rapid review was completed within 48 hours when guidelines were updated.

[0029] Personalized Push Engine Development: A rule-based push algorithm was developed based on the WeChat Official Account platform, setting three core push dimensions: treatment stage (preoperative, 1 month postoperative, 3 months, 6 months, 12 months), disease type (laryngeal cancer / oral cancer / nasopharyngeal cancer / thyroid cancer), and symptom assessment results. Patient basic information (age, disease, surgery time) and symptom assessment data are entered into the backend management system. The engine automatically matches the corresponding knowledge package and pushes it at a preset frequency. For example, for a patient 1 month postoperatively with laryngeal cancer, swallowing-articulation training videos and articles are pushed twice a week. If the symptom assessment shows worsening swallowing difficulties, additional reinforcement training content is pushed.

[0030] Interactive and manual triage module deployment: An AI Q&A robot is integrated into the WeChat official account, with a pre-set database of frequently asked questions (covering categories such as nursing, medication, rehabilitation, and medical visits), enabling rapid responses within 30 seconds. A backend triage system is developed, setting question recognition rules to automatically triage questions that AI cannot answer or red flag signals (such as a PHQ-9 score ≥15, severe pain, suicidal tendencies, etc.) to the specialist nurse's workstation. A nurse response reminder mechanism is configured, providing real-time reminders to nurses to handle new questions during working hours (8:00-20:00) via a WeChat mini-program, ensuring an average response time of <5 minutes; outside of working hours, a confirmation SMS is automatically sent to inform the patient that the issue will be addressed before 9:00 the following day. A tiered response path is established, allowing nurses to directly handle nursing-related issues, submitting complex questions for review and answers by an associate chief physician, and recommending follow-up appointments for those requiring further diagnosis and treatment.

[0031] Optimized elderly-friendly interaction modules: The WeChat official account interface adopts a design standard of 18px font size and 7:1 contrast, adds a voice question function (supporting Mandarin recognition), and sets up three one-click access points: "Rehabilitation Guidance," "Interactive Q&A," and "My Tasks." A user manual with illustrations has been created, covering basic operations such as following the official account, playing videos, asking voice questions, and completing daily tasks. Within 24 hours of patient admission, a specialist nurse will provide 15 minutes of 1:1 operation training, which family members can participate in, ensuring that both patients and their families master the basic operation methods. For patients without smartphones, family members can use their own mobile phones to log in to the patient's account to receive push notifications and assist with interactive questions.

[0032] Data monitoring and security module implementation: A behavior monitoring module was developed to record in real time indicators such as the number of active days per week, video completion rate, and number of interactions. The backend generates adherence reports and automatically triggers telephone follow-up reminders for patients who have been inactive for 7 consecutive days or whose video completion rate is <30%. Data storage uses Alibaba Cloud servers, implementing encrypted data transmission and storage. A three-tiered access control system (administrator, researcher, and evaluator) is implemented, allowing only authorized personnel to access the data. An operation log system is established to record all data access, modification, and export operations, ensuring data traceability. After the study, the data will be exported and encrypted and stored on a dedicated hospital server, retained for 3 years in accordance with relevant regulations.

[0033] Example 2: Implementation of AI-Assisted Health Education Intervention Method

[0034] Study Participant Recruitment and Grouping: Patient recruitment will take place from January 2026 to January 2027 at the Head and Neck Surgery Ward and Follow-up Clinic of the Cancer Hospital of the Chinese Academy of Medical Sciences (Shenzhen). Research nurses will screen patients according to inclusion and exclusion criteria, and will provide eligible patients with a detailed explanation of the study's purpose, procedures, risks, and benefits. After patients voluntarily participate and sign an informed consent form, they will be randomly assigned to either the intervention or control group using a third-party "lottery assistant" mini-program. An average of 10-12 patients are expected to be enrolled per month, with 100 patients recruited within one year.

[0035] Intervention implementation process:

[0036] Intervention Group: Patients completed training on the AI-assisted health education system within 24 hours of admission, followed the corresponding WeChat official account, and completed the initial setup task. During the preoperative phase, preoperative education and anxiety self-help techniques were pushed out according to plan, and nurses checked the completion of tasks. During postoperative hospitalization, short rehabilitation-related videos were pushed out daily, and patients completed a check-in task on the 3rd day postoperatively. Nurses reviewed the risk list in the background and provided targeted health education. After discharge, patients received two themed push notifications per week, and completed check-in screenings at 1, 4, 8, and 12 weeks postoperatively. High-risk patients were followed up by nurses within 48 hours. Patients could ask questions via voice or text at any time through the official account; the AI ​​robot responded quickly, and complex questions were answered by nurses or physicians.

[0037] Control group: Patients were informed of the intervention method upon admission and received standardized text messages every Tuesday and Saturday at 18:00, including perioperative education, dietary guidelines, rehabilitation exercise guidelines, and follow-up appointment reminders. Patients with additional consultation needs were advised to seek assistance through in-person outpatient visits or routine nursing channels.

[0038] Effectiveness assessment was conducted at five time points: baseline (T0), 1 month post-surgery (T1), 3 months post-surgery (T2), 6 months post-surgery (T3), and 12 months post-surgery (T4). At T0 and T1, research nurses guided patients in completing the questionnaire face-to-face in the ward, providing assistance by reading the questions aloud when necessary. From T2 to T4, a questionnaire link was sent to patients via WeChat mini-program for self-completion. Patients who did not complete the questionnaire within 48 hours were followed up by a telephone call with the research nurses for a make-up test. Throughout the assessment process, the research nurses only knew the patients' anonymous IDs and were unaware of the group assignments, ensuring the objectivity of the assessment.

[0039] Statistical Analysis: After all data were entered into the computer, statistical analysis was performed using SPSS 28.0 and R 4.3.1 software. First, descriptive statistics were conducted to analyze the demographic characteristics, disease-related information, and baseline levels of each assessment indicator. Independent samples t-tests were used to compare the differences in each assessment indicator between the two groups of patients. Repeated measures ANOVA was used to analyze the trends of changes in each indicator at different time points in both groups of patients. Analysis of covariance was used to control for the influence of baseline differences and accurately assess the effectiveness of the intervention. Two-tailed tests were used for statistical analysis, with P < 0.05 considered statistically significant.

[0040] Quality Control and Adverse Event Management: A comprehensive quality control system was established. During the content creation phase, dual review was implemented to ensure scientific accuracy. During data collection, dual data entry and random verification were conducted to ensure data authenticity and reliability. During intervention implementation, nurses' responses and patient compliance were regularly checked to promptly identify and resolve problems. If a patient experiences a serious adverse event (such as suicidal tendencies or acute mental disorder episodes), their participation in the study was immediately terminated, handled according to the hospital's emergency procedures, referred to a psychiatric or specialized medical institution, and reported to the ethics committee. For patients who voluntarily withdrew, experienced worsening conditions, or were lost to follow-up, the relevant reasons were recorded, and a combination of intention-to-treat (ITT) and protocol set (PP) methods was used in statistical analysis.

[0041] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An AI-powered personalized health education system for elderly patients with head and neck cancer, characterized in that: The system includes a content generation and management module, a personalized push engine, an interactive and manual triage module, an elderly-friendly interaction module, and a data monitoring and security module. The content generation and management module involves physicians and specialist nurses collaboratively creating and reviewing health education content, encompassing video, text, and animation formats. It establishes a knowledge base with specific content for different pre-operative and post-operative stages and different diseases, and implements regular content updates and version control. The personalized push engine uses a rule-based intelligent algorithm to accurately push content based on the patient's treatment stage (pre-operative, 1 month, 3 months, 6 months, 12 months post-operative), disease type (laryngeal cancer / oral cancer / nasopharyngeal cancer / thyroid cancer, etc.), and symptom assessment results. It also triggers enhanced content pushes based on patient interaction feedback. The interactive and manual triage module... The diagnosis module features AI-powered rapid response (≤30 seconds), and triages unanswerable questions or red flag signals in the background. It fulfills the service commitment of an average nurse response time of <5 minutes during working hours (8:00-20:00) and processing by 9:00 the next day during off-hours, establishing a nurse-physician tiered response mechanism. The elderly-friendly interaction module uses an interface design with a font size ≥18px and a contrast ratio ≥7:1, supports voice questions and one-click access to Q&A, provides 1:1 operation training, and a family member proxy login channel. The data monitoring and security module records patient behavior indicators (weekly active days, video completion rate, number of interactions), triggers intervention reminders for patients with low compliance, and uses anonymized IDs, encrypted data storage, and tiered permission management to ensure privacy and security.

2. The AI-based personalized health education system for elderly head and neck cancer patients according to claim 1, characterized in that, The content creation process of the content generation and management module (1) includes: determining the topic and outline through a weekly co-creation meeting of "physician + specialist nurse", the responsible nurse writing the popular version of the content (≤800 words), the attending physician reviewing the medical accuracy, the head nurse reviewing the readability and age-friendliness, the content being marked with the version number, release date and a proposed review date of ≤1 month, and completing the rapid review within 48 hours when the guidelines are updated.

3. The system according to claim 1, characterized in that, The push rules of the personalized push engine (2) include: configuring corresponding knowledge packages according to the treatment stage, setting specific rehabilitation content according to the disease type (such as focusing on swallowing and articulation training for laryngeal cancer / oral cancer), pushing enhanced content after assessing the patient's symptoms by telephone or WeChat video, and intelligently recommending medical treatment channels for emergencies (nearby emergency room at night, daytime hospital visit).

4. The AI-based personalized health education system for elderly head and neck cancer patients according to claim 1, characterized in that, The hierarchical response mechanism of the interactive and manual triage module is as follows: nursing-manageable issues (nursing guidance, medication routines, functional exercises) are handled in a closed loop by specialist nurses; complex issues are addressed and suggested by associate chief physicians. For those with complex symptoms, we recommend making an appointment for a follow-up visit.

5. An AI-assisted health education intervention method based on the AI-personalized health education system for elderly head and neck cancer patients as described in any one of claims 1-4, characterized in that, Includes the following steps: S1: Screening and grouping of study subjects: Patients aged ≥60 years who were pathologically diagnosed with head and neck malignant tumors and had completed radical surgery, had independent communication ability, and had a PHQ-9 score ≥5 or a GAD-7 score ≥5 were selected and randomly assigned to the intervention group and the control group in a 1:1 ratio, with 50 patients in each group and a total sample size of 100 patients. S2: Intervention Implementation: Intervention Group: Intervention was implemented through the AI-assisted health education system, covering the period from preoperative to 12 months postoperatively. Thematic content was pushed twice a week at fixed times (Tuesday or Saturday), and real-time interactive Q&A was provided daily from 8:00 to 20:

00. Preoperative education and anxiety self-help techniques were pushed. During the postoperative hospitalization period, short videos related to nursing care, pain management, and rehabilitation training were pushed daily. After discharge, swallowing / articulation training, nutrition management, sleep and mood regulation, and preparation for follow-up visits were pushed in stages. Check-in screening was carried out at 1, 4, 8, and 12 weeks postoperatively, and nurses followed up within 48 hours for high-risk patients. Control group: Standardized text messages were sent at the same frequency as the intervention group, including perioperative education, dietary guidelines, rehabilitation exercise guidelines, and follow-up visit reminders. Consultations and suggestions beyond the scope of the text messages were sent to offline clinics or routine nursing channels. S3: Effectiveness Assessment: At baseline (T0, admission date), 1 month post-surgery (T1), 3 months post-surgery (T2), 6 months post-surgery (T3), and 12 months post-surgery (T4), a multidimensional assessment was conducted using the WHO-QoL-Old scale, the PSS stress perception scale, the FoP-Q-SF cancer recurrence fear scale, the PHQ-9 depression assessment scale, the GAD-7 anxiety assessment scale, and the MMSE cognitive function scale. S4: Statistical Analysis of Data: Descriptive statistics, independent samples t-test, repeated measures ANOVA, and covariance analysis (controlling for baseline differences) are used to analyze the assessment data and verify the intervention effect. According to claim 5, the method of an AI-based personalized health education system for elderly patients with head and neck cancer is characterized in that the exclusion criteria for research subjects in step S1 include: comorbid severe mental illness or cognitive impairment, comorbid severe heart, liver, or kidney failure, postoperative plans to participate in other large-scale intervention trials, inability to use a smartphone or lack of information receiving conditions, and other situations deemed unsuitable for enrollment by the researchers.

6. The method of an AI-based personalized health education system for elderly head and neck cancer patients according to claim 5, characterized in that, The system activation process for the intervention group in S2 includes: 15 minutes of 1:1 operation training (including family members) for patients within 24 hours of admission, completing the onboarding tasks of following the official account, playing videos, and sending questions, and receiving an illustrated operation manual.

7. The method of an AI-based personalized health education system for elderly head and neck cancer patients according to claim 5, characterized in that, The S3 assessment implementation process is as follows: For T0 and T1, trained nurses will guide patients to fill out the questionnaire face-to-face in the ward (reading the questions aloud if necessary); for T2-T4, the questionnaire will be pushed out via WeChat mini program first. Those who have not completed the questionnaire will be followed up by telephone within 48 hours for a make-up test. The assessment environment should be quiet, and the duration of each test should be ≤15 minutes. Tests should be avoided within 30 minutes after the use of the analgesia pump. The assessor is a nurse who did not participate in the intervention. Anonymous IDs will be used to administer the test and enter the data. Double entry and 5% random review will be implemented.

8. The method of an AI-based personalized health education system for elderly head and neck cancer patients according to claim 5, characterized in that, It also includes compliance management measures: patients in the intervention group automatically earn points after completing viewing and check-in tasks, and can redeem rehabilitation-related gifts when they reach the required points; for patients who have no activity for 7 consecutive days or whose video completion rate is less than 30%, a telephone follow-up is automatically triggered, and compliance is improved through system reminders, one-on-one guidance from nurses, assistance from family members, and bedside re-education.

9. The method of an AI-based personalized health education system for elderly head and neck cancer patients according to claim 5, characterized in that, It also includes adverse event handling procedures: if a serious adverse event occurs (such as a clear suicidal tendency or an acute mental disorder episode), the patient's participation in the study will be immediately suspended, and the case will be handled in accordance with the hospital's emergency procedures. The on-duty physician will assess the case and refer the patient to a psychiatric or specialized medical institution. At the same time, the event will be recorded and reported to the ethics committee.