Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier


Open Access Dissertation

Document Type


Degree Name

Doctor of Philosophy (PhD)

Degree Program

Computer Science

Year Degree Awarded


Month Degree Awarded


First Advisor

Hong Yu


Patient-centered care has been established as a fundamental approach to improve the quality of health care in a seminal report by the Institute of Medicine published at the start of the century. Improved access to health information and demand for greater transparency contributed to its move into the mainstream. Research has also demonstrated that actively involving patients in the management of their own health can lead to better outcomes, and potentially lower costs. However, despite the efforts in many areas of medicine to embrace patient-centered care, engaging patients is still considered a challenge. One of the barriers is the lack of effective tools to help patients understand their health conditions, options and their consequences. Patient portals are now widely adopted by hospitals and other healthcare practices to provide patients with the capabilities to view their own Electronic Health Records. They are a rich resource of information for patients. However, the language in the records are generally difficult for patients without training in medicine to understand. Furthermore, the amount of information can often be overwhelming as well. In this work, we propose computational approaches to foster patient engagement from three aspects by exploiting the rich information in the medical records. First, we design a framework to automatically generate health literacy instruments to measure a patient's literacy levels. This framework exploits readily available large scale corpora to generate instruments in a commonly used test format. Second, we investigate methods that can determine the readability of complex documents such as health records. We propose to rank document readability, instead of assigning a grade level or a pre-defined difficulty category. Lastly, we examine the problem of finding targeted educational materials to facilitate patient comprehension of medical notes. We study methods to formulate effective queries from specialized and long clinical narratives. In addition, we propose a neural network based method to identify medical concepts that are important to patients. The three aspects of this work address the issues of the overabundance and technical complexity of medical language in health records. We demonstrate that our approaches are effective with various experiments and evaluation metric.


Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.