Center for Digital Health

Overview

Contributed to projects that aimed to improve aspects of health with digital tools.

Role

Intern

Major Contributions

Social Media Dashboard

I helped design a dashboard that analyzed and visualized social media and sensor data for clients and their clinicians, with the goal of improving clinical outcomes in mental health settings.

Research

I conducted an analysis of peer reviewed publications and commercial products where social media and dashboards are being used for mental health.

The following are some of my findings.

Client preferences:

1. See whether their health has changed or deviates from normal range/health guidelines.
2. See correlations between behaviors and mood.
3. Avoid showing negative feedback.

Client concerns:

1. Privacy, they wanted to know who was viewing their data.

Clinician preferences:

1. Organize behavioral targets around students’ personalized goals.
2. Deliver feedback (positive reinforcement, reflection reminders, challenging negative thought).

Data analysis

I worked with the team identify analyses we could do based off of social media and wearable sensor data. 

Social media variables considered:

    • Amount of positive/negative language
    • Number of posts
    • Amount and content of posts/google searches between 12-5 am (dangerous hours)
    • Leisure language used
    • Amount of self-referential language
    • Stress and anxiety scores (based on historical data)
    • Map of places recorded on their social media (via events, posts, photos, check-ins)
    • Number of posts referring to past, present, or future

Self-report variables considered:

    • Self-reported mood or well-being (i.e. on a scale of 1-5, how did you feel this week?)

Wearable sensor variables considered:

    • Amount of sleep (restless and restful)
    • Number of steps
    • Distance travelled

Design

I then worked with the team’s Design Strategist to identify the content and layout of potential dashboard modules. 

Patient Readability Pilot

I developed a Patient Readability Pilot to examine whether module prototypes were readable and actionable for patients, to be implemented by the center’s research coordinator.

Google Review Analysis

I wrote Python scripts to analyze Yelp and Google reviews of substance abuse treatment facilities, for a project aimed at using big data to improve Pennsylvania’s health care system, through the hospital, the home, and the community. 

Prior to my involvement, the project focused only on Yelp reviews, which captured roughly 7% of online reviews. I expanded the scope of the project by writing code to analyze Google reviews, which allowed the team to capture an additional 57.5% of online reviews.