Senior Design Projects
ECS193 A/B Winter & Spring 2021
Decision Support for Radiation Oncology
| ********** | |
| Name | Sonja Dieterich |
| Affiliation | UC Davis Health |
Project's details
| Project title | Decision Support for Radiation Oncology |
| Background | What is the goal of this project? To practice evidence based medicine, Radiation Oncologists have to memorize hundreds of treatment recommendations based on a decision tree including patients’ pathology, cancer staging, age, co-morbidities and other factors. As new clinical trial results are published, studies have shown multi-year delays to move new best evidence into routine clinical practice. For example, more than 30% of women with N2/N3 breast cancer in the US still do not receive the recommended post-mastectomy radiation, increasing the local recurrence rate by 20%-27% and reducing overall survival by 8-9%. Clinicians in small rural clinics in the US and in developing countries face additional barriers to implement new evidence, because they often do not have resources to access peer-reviewed literature or attend conferences to update their clinical expertise. Hypotheses: 1. Decision support software can shorten the time from clinical trials publication to widespread implementation of new treatment regimens in Radiation Oncology. 2. Decision support software can increase the percentage of patients who are eligible for radiation treatments to receive radiation treatments. 3. Decision support software will reduce the time clinicians need to spend on researching evidence based, individually tailored treatment regimens. |
| Description | What has been done so far? A previous ECS-193 and since then an undergraduate student have build a literature search tool for early stage breast cancer, brain and spine (CNS) cancers, and pediatric disease which can be accessed through a browser. The software is hosted on GitHub. The tool searches for literature based on a closest match form the gold standard of the papers cited in the Handbook. The software provides some rudimentary tools for users to provide feedback on listed articles as relevant/not relevant, and for marking user favorites. In addition, the undergraduate student has been working on NLP analysis of the literature search result papers. We are currently use the abstracts to conduct a principal components analysis to determine proximity to the gold standard papers. Our hope is that this will improve relevance of the top search results shown. |
| Deliverable | What are the goals for the 2021 ECS 193 project?,br>In the first usability tests of the decision support software, it became clear that the currently implemented ranking system of search results does not meet the software user needs. We therefore started work on exploring different ranking options. The next step for software development is to integrate these ranking methods, provide infrastructure to add new ranking options in the future, and capture user input to quantitatively study user feedback on the relevance of literature ranking for the different methods. Specifically, the ECS-193 team should: 1) Adapt the decision support software infrastructure such that different ranking methodologies can be more easily plugged in for further evaluation of ranking methodologies. The ranking strategy in use should be easily selectable on the GUI by either the software administrator (global default choice) or the user (local choice). 2) Create a tool to export user rankings of search results to allow for quantitative analysis of user feedback on ranking methods. |
| Skill set desirable | N/A |
| Phone number | ********** |
| Client time availability | 30-60 min weekly or more |
| IP requirement | Open source project |
| Attachment | N/A |
| Selected | No |
| Team members | N/A |
| TA | N/A |