2017-summer-internship

iDASH is pleased to announce the availability of summer research internships for undergraduates, graduate students, and postgraduates.

 

Program Structure

Each intern is paired with one primary mentor and, in some cases, secondary mentors. Interns may work independently or on small teams for their projects. Interns are generally expected to reside in San Diego during the internship, and are required to attend a kickoff meeting at the beginning of the program* and the Internship Symposium at the end of the program to present the results of their work. During the internship period, iDASH will host weekly lunch seminars to be attended by all interns and selected program mentors and leaders. Interns are also expected to schedule regular 1:1 meetings with their mentor(s) throughout the internship period. Monthly social activities will also be hosted for interns – in summers past, this included potluck lunches, day hikes, and a surfing session.

 

Timeline

February 10

Application Period opens

March 24

Applications will no longer be accepted

April 14

Notification of acceptance or waitlist status

April 21

Response required from intern to guarantee participation

May 30/June 12*

Program begins with kickoff meeting

July 14

Midterm evaluations

August 4

Final presentations

August 4/18*

Program concludes

* Students who are unavailable due to their school calendar for the May 30th kickoff meeting will start on June 12th and finish on August 18th.

 

Participating Faculty and Project Information

Jina Huh

Project 1
 
In this project, we will work with UCSD CAPS to develop a mobile-based intervention to help college students improve mental health. The mobile-based intervention involves expressive arts therapy sessions in an online support group form. The student will be asked to conduct focus groups and user testing of the application to develop the intervention. 
 
Project 2
 
In this project, we will analyze the effect of various interface features supported by the bedside tablets in Jacobs Medical Center. We will assess the efficacy of watching visual healing videos against other clinical and administrative outcomes (e.g., length of hospital stay, reported pain level). We will also investigate the association between various usage patterns of the tablet vs. clinical and administrative outcomes. 
 

Xiaoqian Jiang

Project 1

Distributed conditional random field (CRF). The CRF is a popular model in natural language processing and content analysis, which can be useful to biomedical challenges. However, a major bottleneck is that data are often hosted in distributed sources, which cannot be shared in the raw format due to privacy concerns. This project will develop privacy-preserving distributed CRF model with/without approximation to ensure utility and privacy. 
 

Project 2

Learning on encrypted data. This project will focus on homomorphic encrypted learning algorithm development for outsourced data in a public cloud. The major task is to solve real problem with scalability and efficiency in the learning processing when inputs are encrypted. 
 

Hyeoneui Kim

Project 1

Assessing EHR data quality: develop comprehensive quality metrics. Review of existing methods is expected as well 

Project 2

Developing an algorithmic process of filtering clinical data based on patient's data sharing preference, before making the data available for research 
 

Shuang Wang

Development novel methods to protect genome data privacy and security 
 
Description: Genome data sharing is essential to promote scientific discovery, improve healthcare quality, and support meaningful use. However, data privacy is a major concern that may hinder genome data sharing and health information exchange for research. In this project, students will utilize cryptographic technologies, such as secure multiparty computation and homomorphic encryption to develop new tools to safeguard genome data analysis in public cloud or across different institutions. 
 


 Dr. Lucila Ohno-Machado & Dr. Tsung-Ting Kuo (Tim)

Project 1

In this era of big data science, it has become increasingly difficult to efficiently search for, view, and use datasets for biomedical and healthcare research. National initiatives are in place that support data sharing; however, there are still regulations and challenges to overcome. In this research project, we will develop and implement a Blockchain-based data sharing framework to address the challenges with data sharing. Blockchain is a distributed ledger/database technology composed of a chain of blocks for transactions, for example, used with the Bitcoin cryptocurrency. Based on state-of- the-art Blockchain platforms, we will develop and implement data structures and algorithms to construct a self-workable ecosystem. The result will be a framework that improves data accessibility for researchers. 
 
Project 2 
 
Predictive modeling can advance research and facilitate quality improvement initiatives and substantiate research results, especially when data from multiple healthcare systems can be included. However, current, state-of-the-art privacy-preserving predictive modeling frameworks are still centralized, in other words, the models from distributed sites are integrated in a central server to build a global model. This centralization carries several risks, e.g., single-point-of-failure at the central server. To improve the security and robustness of predictive modeling frameworks, we will develop and implement novel and advanced algorithms on decentralized Blockchain networks (a distributed ledger/database technology adopted by the Bitcoin cryptocurrency) to build better models. The outcome will be algorithms that improve the predictive power of data from multiple healthcare systems through a distributed system. 
 
Project 3
 
With the wide adoption of electronic health record systems, building more accurate predictive models using data from distributed healthcare systems becomes a possibility. Recently, researchers have started to focus on decentralized privacy- preserving machine learning methods based on new technologies such as Blockchain (a chain of blocks of transactions that form a distributed ledger or database). However, current state-of-the-art, distributed privacy-preserving predictive modeling methods are designed for fully connected networks, and the algorithms may not be applied directly to real-world, complicated (i.e., partially connected) networks. In this research project, we will develop and implement new algorithms to deal with the privacy-preserving prediction problems on such complicated networks. The output will be algorithms that improve the predictive power of data from multiple healthcare systems within complicated networks. 

 

Luca Bonomi

Project 1

Modern information systems enable the collection of a large amount of temporal medical data (e.g. EHR). Despite the importance of this data for various medical applications (e.g. prediction of risk factors, evolution of diagnoses) the extraction of useful knowledge is very challenging due to the high dimensionality and the noisy nature of the data. In this project, we aim to develop robust solutions for discovering recurrent structures in the data to address the aforementioned challenges. 

Project 2
 
Data sharing and integration solutions provide the foundation for the current inter- institutional medical research. In this setting, the privacy concerns regarding the sensitive medical information and large amount of data pose significant challenges. The goal of this project is twofold: (1) study how current techniques could benefit from domain specific knowledge and (2) develop practical privacy preserving solutions for data sharing and integration. 
 

Previous Internships

iDASH has hosted dozens of undergraduate and graduate students over the past five summers. To read internship alumni bios and see videos from their Symposium presentations, please use the links in the dropdown menu above.

View presentations from the 2016 Internship Symposium here.

 

Application
Application period ended March 24, 2017.