General track on digital health

Health care for patients through Multiple Applications of WeChat (DHA2019-17)

Author/s: Wenjie Long, Yile Ning, Tingchun Wu, Qingqing Liu, Lingjun Wang, Huili Liao, Zhongqi Yang and Shaoxiang Xian

Abstract: WeChat (Wēixìn), the Chinese social media and multi-functional application,with more than one billion monthly active users. Meanwhile, it allows the users to connect privately, providing instant text, sharing photos, voice, and video calls ,etc. It is one of the most popular mobile chat app in China. Widely used and easy to set up technology makes it an ideal platform for health care, including registration, follow up, and monitoring,etc. Moreover, the patients can pay and handle prescriptions via e-pharmacy. The physicians are often looking for a secure and easy way to use mobile applications to share clinical images and data, and to obtain photographs, vital signs data and laboratory data from patients. Therefore, we will provide an overview of the application of WeChat in healthcare.


Designing a security-oriented medical cyber physical system for future smart healthcare (DHA2019-07)

Author/s: Shuai Luo, Hongwei Liu, Ershi Qi and Mingjing Liu

Abstract: Medical Cyber-Physical System (MCPS) is a new disruptive approach which enable smart health system to monitor, process and make autonomous decisions without involving doctors and caregivers. However, the security of MCPS is still a vital challenge for the healthcare industry. In this paper, we propose a security-oriented MCPS with security policy: physical defense policy, network defense policy and reaction policy. The physical defense policy is in charge of identifying the failed physical medical devices by creating the virtual medical devices in the edge of network. The network defense policy aims at detecting the suspicious network communications by creating the virtual network function. The reaction policy is used to manage and configure the physical and virtual medical devices and the network resources. In addition, an effective fuzzy alarms filter is designed based on the fuzzy if-then rules to reduce the false alarms caused by the network communication. A case study is used to document the security policy of the proposed MCPS and the effectiveness of the fuzzy alarm filter. The results show that the proposed MCPS is effective to guarantee the security of MCPS and the false alarms can be reduced through the proposed fuzzy alarms filter.


Recognition of Patient Reviews Based on Dynamic Mixed Sampling and Transfer Learning (DHA2019-08)

Author/s: Fei Xiang and Yaotan Xie

Abstract: In this study, a convolutional neural network model based on dynamic mixed -sampling and transfer learning is proposed to solve the imbalanced data problem of online patient reviews. We applied the method combined with dynamic mixed sampling and transfer learning to solve the imbalanced data problem, and then used deep learning architecture based on Word2vec and convolutional neural network for the distributed representation, feature extraction and topic classification of online patient reviews. Compared with traditional machine learning algorithm represented by SVM and single convolutional neural network, the topic recognition model based on dynamic mixed sampling and transfer learning has significantly improved the accuracy, recall rate and F1 value of prediction. In this study, the recognition model of patient reviews based on dynamic mixed sampling and transfer learning can apparently improve the recognition effect when dealing with imbalanced data.


The Cost-effectiveness Evaluation of Acupuncture and Moxibustion for Peripheral Neuropathic Pain with Network-meta Analysis of Frequency Method (DHA2019-19)

Author/s: Weixuan Zhao, Haoming Huang, Jingchun Zeng, Kun Liu, Shuxin Wang, Shiyu Lin, Wenjie Long, Lixia Li, Guohua Lin and Liming Lu

Abstract: (1) Background: Acupuncture and Moxibustion are traditional therapies of more cost-effective than medications in some cases, but there is not sufficient economic evidence. (2) Methods: Based on the frequency method, we conduct a network meta-analysis of RCTs on acupuncture and moxibustion for peripheral neuropathic pain and build models by graph theory of direct or indirect comparisons between interventions with the "netmeta" package under R software. We measure the input expense of outpatient treatments including direct and indirect cost. According to the results of meta-analysis, the RD/NNT is used as the criterion, and the efficacy of Western medicine is set as the control. The net cost of benefit and net cost of non-benefit of various acupuncture and moxibustion interventions are calculated to perform cost-effectiveness analysis.(3) Results: Among the net costs of various interventions, the top five acupuncture therapies with the lowest average benefit and net cost are "tapping + cupping", "acupuncture + cupping", "needle knife", "fire needle" and "warm needle", and they are 1219.39, 1916.87 ,2154.36, 2229.41 and 3420.44 RMB/week. (4) Conclusions: As the assembled result of network meta-analysis and cost-effectiveness analysis, "fire needle" and "warming needle" are better interventions for peripheral neuropathic pain.


Measuring healthcare services productivity with environmental constraints in China (DHA2019-23)

Author/s: Jinna Yu, Tingting Zhang and Zhen Liu

Abstract: This paper analyzes the growth and decomposition of Total Factor Productivity (TFP) in the provision of healthcare services across 31 provinces in China. We used balanced panel data for 2005-2016 covering 31 Provinces in China and the data from 2006-2017 China Health and Family Planning Statistical Yearbook. We employed the Metafrontier undesirable super-efficiency slack-based measure model and the global Malmquist-Luenberger index to measure the growth and decomposition of TFP. The results show that: (1) Considering the entire country, healthcare services’ TFP shows a slow decline, which is due to insufficient technical innovation. However, an increase in technical catch-up can alleviate this trend. (2) The effect of technical catch-up in the eastern, middle, and western regions of China is significant across all three, while the effect of technical innovation was negative. (3) There are differences between province-level TFPs. These findings suggest that, under existing environmental constraints, relevant government departments should improve technical innovation in the supply of healthcare services and medical waste treatment, increase technical efficiency in the factor of healthcare production, strengthen regional health planning, and balance the development of regional healthcare.


A Machine Learning Approach to the Analysis of Suicide-Related Online Videos: A Case Study of Singapore (DHA2019-05)

Author/s: Caifeng Zhang, Rui Ma and Shuyang Li

Abstract: A recent World Health Organization report indicates that individuals aged 15 to 29 are at higher risk of committing suicide across many societies and cultures in recent years. In Singapore, suicide is known to be the leading cause of death for this age group. The impact of suicide on the deceased’s family members and communities is devastating and long-lasting. While not all instances of suicidal ideation may lead to self-harm and actual suicidal acts, it places individuals at an increased risk of suicide. In the past decade, the world has seen a significant increase in the use of social media platforms such as Twitter, Instagram, Facebook and YouTube, coinciding with a rapid adoption of smart phone devices around the world. Such platforms are widely accessed often for entertainment and social networking. However, in recent years, media reported about individuals broadcasting their suicidality on social media sites. Recent studies have shown that social media sites such as Twitter can be used as suicide prevention tools. If early detection of suicide warning signs can be done, suicide deaths can be prevented through early intervention. In this study, we analyzed suicide-related YouTube videos in Singapore. In particular, we analyzed user comments using Natural Language Processing with English and Singlish (colloquial Singaporean English). We developed a customized corpus which includes suicide and other mental health-related terms as well as widely used online jargon, short-hands and acronyms. Lastly, to discover topics, we performed topic modeling using Latent Dirichlet Allocation (LDA).


Exploring user’s willingness to disclose personal information on online medical platform: the role of privacy concern (DHA2019-27)

Author/s: Bernadine Lye, Jane Seah and Kyong Jin Shim

Abstract: Online medical platform has become people's essential choice to seek medical service. On the one hand, online medical platform can help users reduce time and treatment cost. On the other hand, individuals face the threat of leaking private medical information. The paper considers satisfaction as one factor that may affect user’s willingness to disclose personal information. Using survey approach, this study discusses the factors, the security control of platform, perceived usefulness, satisfaction, information sensitivity, privacy concern in platforms and users’ trust in platforms that may influence users’ willingness to disclose personal information. The results indicate that user’s concern on online medical platform does not produce a negative impact on the willingness to disclose personal information. This may be caused by people’s higher willingness to save cost through online medial platform, although they have larger risk of personal medical information leakage. The findings provide implications for theory and practice.

Digitalisation Adding Value to Healthcare

©2020 by DHA2020.