Examining the Health Impacts of Digital Technologies
Dr Vaibhav Rajan, Assistant Professor, School of Computing
Join Zoom Meeting
https://nus-sg.zoom.us/j/96318967271?pwd=L0hwNVFPRWtkUUI3Q0NqSUFSZ2hjZz09
Abstract:
Digital technologies have the tremendous potential to affect human's health status by supporting the care delivery process in healthcare organizations and reforming individual's lifestyles in daily life.
As the underlying foundation for various digital technologies used in health sector, Electronic Medical Records / Electronic Health Records (EMR/EHR), are the major digital system in the health sector. EMR/EHR stores and manages all the health-related data of patients. This enables the development of predictive analytics-based solutions for clinical decision support. Among various important health outcomes, identifying high-risk individuals with one or multiple diseases early is significantly imperative. However, the value of EHR in the disease predictive analytics is less explored. On the other hand, the digital technologies used outside the domain of health care delivery, may also have the ability to instigate profound changes in public health. The digital technologies used in daily life include fitness and non-fitness applications. Beyond the straightforward health effect of fitness applications which monitor daily diet and exercise, some non-fitness applications daily-used also have the potential to affect individual's health inadvertently. However, fewer papers observe the potential health effect of these non-fitness applications which is less detectable. In general, this thesis aims to fill up these two gaps.
The first paper in my thesis observes the problem of multiple disease predictive analytics (MDPA). Data-driven approaches for MDPA face the challenge of data imbalance because rare diseases tend to have much lesser data than common diseases. Insufficient data for rare diseases makes it difficult to leverage correlations with other diseases. These correlations may be studied and recorded in biomedical literature, but are rarely utilized in predictive analytics. The first essay presents an approach to augment the MDPA models, called Knowledge-Aware Approach (KAA) that learns clinical correlations from the rapidly growing body of clinical knowledge. KAA can be combined with any data-driven MLL model for MDPA to refine the predictions of the model.
In the second paper of my thesis, I pay attention to the health influence that results from non-fitness digital technology. Online food delivery (OFD) services provide users the access to a large amount of food in a handly available way. These services free out individuals from the kitchen workload to enjoy a more convenient life. However, the increasing availability of healthy and unhealthy food options in OFD services may reshape users' eating habits and influence the diet management for diet-related diseases, such as diabetes. Diabetes has become a worldwide leading cause of death. Given its severity, the second essay in this thesis aims to investigate the health impact of OFD services on diabetes. I construct a Difference-in-Difference model given the staggered launches of OFD services across counties in U.S. The estimations show that these OFD platforms can substantially increase the occurrence of diabetes. A variety of heterogenous factors are also considered to understand heterogeneity in the effects. The insights obtained can shield light on the diabetes prevention and management for all stakeholders, including consumers and platform runners.