Qingyuan Song

Qingyuan Song

Auto-generated profile, based on publicly available information

Qingyuan Song

Vanderbilt University, Nashville, TN, USA

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 November 2024 in Journal of Medical Internet Research
Verified
Journal Article
Open access
LWLeyao Wang
Leyao Wang
ZWZhiyu Wan
Zhiyu Wan
CNCongning Ni
Congning Ni
QSQingyuan Song
Qingyuan Song
YLYang Li
Yang Li
ZYZhijun Yin
Zhijun Yin
Background: The launch of ChatGPT (OpenAI) in November 2022 attracted public attention and academic interest to large language models (LLMs), facilitating the emergence of many other innovative LLMs. These LLMs have been applied in various fields, including health care. Numerous studies have since been conducted regarding how to use state-of-the-art LLMs in health-related scenarios. Objective: This review aims to summarize applications of and concerns regarding conversational LLMs in health care and provide an agenda for future research in this field. Methods: We used PubMed, ACM, and the IEEE digital libraries as primary sources for this review. We followed the guidance of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) to screen and select peer-reviewed research articles that (1) were related to health care applications and conversational LLMs and (2) were published before September 1, 2023, the date when we started paper collection. We investigated these papers and classified them according to their applications and concerns. Results: Our search initially identified 820 papers according to targeted keywords, out of which 65 (7.9%) papers met our criteria and were included in the review. The most popular conversational LLM was ChatGPT (60/65, 92% of papers), followed by Bard (Google LLC; 1/65, 2% of papers), LLaMA (Meta; 1/65, 2% of papers), and other LLMs (6/65, 9% papers). These papers were classified into four categories of applications: (1) summarization, (2) medical knowledge inquiry, (3) prediction (eg, diagnosis, treatment recommendation, and drug synergy), and (4) administration (eg, documentation and information collection), and four categories of concerns: (1) reliability (eg, training data quality, accuracy, interpretability, and consistency in responses), (2) bias, (3) privacy, and (4) public acceptability. There were 49 (75%) papers using LLMs for either summarization or medical knowledge inquiry, or both, and there are 58 (89%) papers expressing concerns about either reliability or bias, or both. We found that conversational LLMs exhibited promising results in summarization and providing general medical knowledge to patients with a relatively high accuracy. However, conversational LLMs such as ChatGPT are not always able to provide reliable answers to complex health-related tasks (eg, diagnosis) that require specialized domain expertise. While bias or privacy issues are often noted as concerns, no experiments in our reviewed papers thoughtfully examined how conversational LLMs lead to these issues in health care research. Conclusions: Future studies should focus on improving the reliability of LLM applications in complex health-related tasks, as well as investigating the mechanisms of how LLM applications bring bias and privacy issues. Considering the vast accessibility of LLMs, legal, social, and technical efforts are all needed to address concerns about LLMs to promote, improve, and regularize the application of LLMs in health care.

27 April 2024 in medRxiv
Verified
Preprint
Open access
PDF file
LWLeyao Wang
Leyao Wang
ZWZhiyu Wan
Zhiyu Wan
CNCongning Ni
Congning Ni
QSQingyuan Song
Qingyuan Song
YLYang Li
Yang Li
ZYZhijun Yin
Zhijun Yin
Background: The launch of the Chat Generative Pre-trained Transformer (ChatGPT) in November 2022 has attracted public attention and academic interest to large language models (LLMs), facilitating the emergence of many other innovative LLMs. These LLMs have been applied in various fields, including healthcare. Numerous studies have since been conducted regarding how to employ state-of-the-art LLMs in health-related scenarios to assist patients, doctors, and public health administrators. Objective: This review aims to summarize the applications and concerns of applying conversational LLMs in healthcare and provide an agenda for future research on LLMs in healthcare. Methods: We utilized PubMed, ACM, and IEEE digital libraries as primary sources for this review. We followed the guidance of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRIMSA) to screen and select peer-reviewed research articles that (1) were related to both healthcare applications and conversational LLMs and (2) were published before September 1 st , 2023, the date when we started paper collection and screening. We investigated these papers and classified them according to their applications and concerns. Results: Our search initially identified 820 papers according to targeted keywords, out of which 65 papers met our criteria and were included in the review. The most popular conversational LLM was ChatGPT from OpenAI (60), followed by Bard from Google (1), Large Language Model Meta AI (LLaMA) from Meta (1), and other LLMs (5). These papers were classified into four categories in terms of their applications: 1) summarization, 2) medical knowledge inquiry, 3) prediction, and 4) administration, and four categories of concerns: 1) reliability, 2) bias, 3) privacy, and 4) public acceptability. There are 49 (75%) research papers using LLMs for summarization and/or medical knowledge inquiry, and 58 (89%) research papers expressing concerns about reliability and/or bias. We found that conversational LLMs exhibit promising results in summarization and providing medical knowledge to patients with a relatively high accuracy. However, conversational LLMs like ChatGPT are not able to provide reliable answers to complex health-related tasks that require specialized domain expertise. Additionally, no experiments in our reviewed papers have been conducted to thoughtfully examine how conversational LLMs lead to bias or privacy issues in healthcare research. Conclusions: Future studies should focus on improving the reliability of LLM applications in complex health-related tasks, as well as investigating the mechanisms of how LLM applications brought bias and privacy issues. Considering the vast accessibility of LLMs, legal, social, and technical efforts are all needed to address concerns about LLMs to promote, improve, and regularize the application of LLMs in healthcare.

31 May 2024 in JMIR Aging
Verified
Journal Article
Open access
ZYZhijun Yin
Zhijun Yin
LSLauren Stratton
Lauren Stratton
QSQingyuan Song
Qingyuan Song
CNCongning Ni
Congning Ni
LSLijun Song
Lijun Song
BMBradley Malin
Bradley Malin
Background: Informal dementia caregivers are those who care for a person living with dementia and do not receive payment (eg, family members, friends, or other unpaid caregivers). These informal caregivers are subject to substantial mental, physical, and financial burdens. Online communities enable these caregivers to exchange caregiving strategies and communicate experiences with other caregivers whom they generally do not know in real life. Research has demonstrated the benefits of peer support in online communities, but this research is limited, focusing merely on caregivers who are already online community users. Objective: We aimed to investigate the perceptions and utilization of online peer support through a survey. Methods: Following the Andersen and Newman Framework of Health Services Utilization and using REDCap (Research Electronic Data Capture), we designed and administered a survey to investigate the perceptions and utilization of online peer support among informal dementia caregivers. Specifically, we collected types of information that influence whether an informal dementia caregiver accesses online peer support: predisposing factors, which refer to the sociocultural characteristics of caregivers, relationships between caregivers and people living with dementia, and belief in the value of online peer support; enabling factors, which refer to the logistic aspects of accessing online peer support (eg, eHealth literacy and access to high-speed internet); and need factors, which are the most immediate causes of seeking online peer support. We also collected data on caregivers’ experiences with accessing online communities. We distributed the survey link on November 14, 2022, within two online locations: the Alzheimer’s Association website (as an advertisement) and ALZConnected (an online community organized by the Alzheimer’s Association). We collected all responses on February 23, 2023, and conducted a regression analysis to identifyn factors that were associated with accessing online peer support. Results: We collected responses from 172 dementia caregivers. Of these participants, 140 (81.4%) completed the entire survey. These caregivers were aged 19 to 87 (mean 54, SD 13.5) years, and a majority were female (123/140, 87.9%) and White (126/140, 90%). Our findings show that the behavior of accessing any online community was significantly associated with participants’ belief in the value of online peer support (P=.006). Moreover, of the 40 non–online community caregivers, 33 (83%) had a belief score above 24—the score that was assigned when a neutral option was selected for each belief question. The most common reasons for not accessing any online community were having no time to do so (14/140, 10%) and having insufficient online information–searching skills (9/140, 6.4%). Conclusions: Our findings suggest that online peer support is valuable, but practical strategies are needed to assist informal dementia caregivers who have limited time or online information–searching skills.

17 November 2023 in Journal of Medical Internet Research
Verified
Journal Article
Open access
CNCongning Ni
Congning Ni
QSQingyuan Song
Qingyuan Song
BMBradley Malin
Bradley Malin
LSLijun Song
Lijun Song
PCPatricia Commiskey
Patricia Commiskey
ZYZhijun Yin
Zhijun Yin
Background: Alzheimer disease or related dementias (ADRD) are severe neurological disorders that impair the thinking and memory skills of older adults. Most persons living with dementia receive care at home from their family members or other unpaid informal caregivers; this results in significant mental, physical, and financial challenges for these caregivers. To combat these challenges, many informal ADRD caregivers seek social support in online environments. Although research examining online caregiving discussions is growing, few investigations have distinguished caregivers according to their kin relationships with persons living with dementias. Various studies have suggested that caregivers in different relationships experience distinct caregiving challenges and support needs. Objective: This study aims to examine and compare the online behaviors of adult-child and spousal caregivers, the 2 largest groups of informal ADRD caregivers, in an open online community. Methods: We collected posts from ALZConnected, an online community managed by the Alzheimer’s Association. To gain insights into online behaviors, we first applied structural topic modeling to identify topics and topic prevalence between adult-child and spousal caregivers. Next, we applied VADER (Valence Aware Dictionary for Sentiment Reasoning) and LIWC (Linguistic Inquiry and Word Count) to evaluate sentiment changes in the online posts over time for both types of caregivers. We further built machine learning models to distinguish the posts of each caregiver type and evaluated them in terms of precision, recall, F1-score, and area under the precision-recall curve. Finally, we applied the best prediction model to compare the temporal trend of relationship-predicting capacities in posts between the 2 types of caregivers. Results: Our analysis showed that the number of posts from both types of caregivers followed a long-tailed distribution, indicating that most caregivers in this online community were infrequent users. In comparison with adult-child caregivers, spousal caregivers tended to be more active in the community, publishing more posts and engaging in discussions on a wider range of caregiving topics. Spousal caregivers also exhibited slower growth in positive emotional communication over time. The best machine learning model for predicting adult-child, spousal, or other caregivers achieved an area under the precision-recall curve of 81.3%. The subsequent trend analysis showed that it became more difficult to predict adult-child caregiver posts than spousal caregiver posts over time. This suggests that adult-child and spousal caregivers might gradually shift their discussions from questions that are more directly related to their own experiences and needs to questions that are more general and applicable to other types of caregivers. Conclusions: Our findings suggest that it is important for researchers and community organizers to consider the heterogeneity of caregiving experiences and subsequent online behaviors among different types of caregivers when tailoring online peer support to meet the specific needs of each caregiver group.

4 December 2024 in JMIR Aging
Verified
Journal Article
Open access
CNCongning Ni
Congning Ni
QSQingyuan Song
Qingyuan Song
QCQingxia Chen
Qingxia Chen
LSLijun Song
Lijun Song
PCPatricia Commiskey
Patricia Commiskey
ZYZhijun Yin
Zhijun Yin
Background: Alzheimer disease and related dementias (ADRD) are a growing global health challenge. ADRD place significant physical, emotional, and financial burdens on informal caregivers and negatively affects their well-being. Web-based social media platforms have emerged as valuable sources of peer support for these caregivers. However, there has been limited investigation into how web-based peer support might influence their mental well-being. Objective: This study aims to examine the dynamics of sentiment scores, a major indicator of mental well-being, among informal ADRD caregivers, specifically how their sentiment changes as they participate in caregiving experience discussions within 2 ADRD web-based communities. Methods: We collected data from 2 large web-based ADRD caregiving communities, ALZConnected (from November 2011 to August 2022) and TalkingPoint (from March 2003 to November 2022). Using the Valence Aware Dictionary for Sentiment Reasoning and Linguistic Inquiry and Word Count, we calculated sentiment scores for each post and evaluated how the initial sentiment score of a topic initiator evolves within a discussion thread. Structured topic modeling and regression analysis were used to identify the primary topics consistently associated with sentiment changes within these threads. We investigated longitudinal sentiment trends to identify patterns of sentimental stability or enhancement due to prolonged engagement in web-based communities by plotting linear interpolation lines of the sentiment values of each individual user. Results: The ALZConnected dataset comprised 532,992 posts, consisting of 57,641 topic threads and 475,351 comments. The TalkingPoint dataset was composed of 846,344 posts, consisting of 81,068 topic threads and 765,276 comments. Our research revealed that topic initiators experienced a notable increase in sentiment as they engaged in subsequent discussions within their threads, with a significant uptick in positivity in the short term. This phenomenon is part of a broader trend of steadily rising positive sentiment among ADRD caregivers. Using structured topic modeling, we cataloged a diverse range of topics that included both emotional aspects, such as family emotions, and practical concerns, such as diagnosis and treatment and everyday care practices. We observed that sentiment scores were positively aligned with discussions about family and daily routines life (coefficient=3.53; P<.001), while topics related to illness (coefficient=–1.37; P<.001) and caregiving facilities (coefficient=–1.98; P<.001) tended to correlate with lower sentiment scores. This evidence highlights the significant impact that both the time of participation and the posting content have on the sentiment changes of caregivers. Conclusions: This study identifies sentiment changes among informal ADRD caregivers through their interactions in 2 extensive web-based communities. These findings emphasize the importance of early emotional support within a topic thread and demonstrate a predominantly positive sentiment in these communities over time. These further highlight the value of web-based peer support and its potential to enhance the emotional well-being of informal ADRD caregivers.