JMIR Publications

ABSTRACT

Background:

With increased Internet penetration among older adults and the rapid growth of online health communities（OHCs）, the Internet has become one of the main ways for older adults to obtain health information. OHCs have stored a large amount of information on geriatric consultations, providing valuable resources for tapping the health information needs of elderly Chinese individual.

Objective:

This study aimed to explore how to reveal the health information needs of elderly Chinese individual based on OHCs and how to tap into the differential and evolutionary trends of the health information needs of elderly Chinese Individuals from age and sex perspective.

Methods:

We collected 3,605 geriatric consultations from the Good Doctor OHC as a data source. Age groups were established and incorporated into the latent Dirichlet allocation model (LDA), and a novel LDA based evolutionary approach (EA-LDA) for health information needs mining was proposed. First, the Q&A texts generated by users in the same age group were organized as one document; thus, the document-word corpus was transformed into an age-word corpus. Second, the extended TF-IDF was used to recalculate the weight of all words in each age-word document to give greater weight to key words most prominent in a particular age group. Finally, by sampling each document in the age-word document, the EA-LDA model generated topic words and age topics. The topic-word part of the model contained the need topics, and the age-topic part of the model was composed of the need topics distributed across each age group.

Results:

The results found 12 main health information needs and their evolutionary processes among elderly Chinese individuals: cancer and tumours (39.16%), tubercles (8.90%), eye diseases (2.90%), orthopaedic diseases (6.72%), skin diseases (5.50%), urinary system diseases (7.48%), cognitive dysfunction (1.37%), digestive system diseases (4.24%), respiratory diseases (4.91%), cardiovascular diseases (15.37%), anxiety and depression (2.62%), and oral problems (0.84%). Among these topics, the proportion of individuals who need information about cancer and tumours began to decline after the age of 80. Interest in orthopaedic diseases, skin diseases, urinary diseases, cognitive dysfunction, digestive diseases, respiratory diseases and cardiovascular diseases gradually increased with age, and tubercle gradually decreased. There were differences between males and females regarding health information needs for cancer and tumours, eye diseases, skin diseases, cognitive dysfunction, respiratory diseases and oral problems.

Conclusions:

In this study, a novel EA-LDA-based topic mining and evolutionary analysis approach was proposed for discovering online health information needs. The health information needs of elderly Chinese individuals are different in age and sex. These study results provide guidance for future targeted health education for the elderly will be helpful for OHCs seeking to optimize health information services and improve information quality.