Chris Hemphill Defining the Through Line for Data Analysis & AI to Reclaim the Algorithm

Chris Hemphill

Chris Hemphill is a data scientist who has worked in healthcare and digital mental health. We explored his career evolution, current research, and philosophy on using AI and apps to understand human mental health needs and other technology adoption for the greater good.

Chris graduated in 2008 with a marketing degree, notably during the financial crisis, which he recognizes made him fortunate to secure a job. While in college, he worked selling cars, and one of his customers offered his first post-graduation job opportunity over lunch. Chris entered healthcare immediately after graduation in 2008 and has remained in the healthcare vertical throughout his career.

The pivot to data science came in 2018 while working in an Excel-heavy operations role. Chris had a mentor who taught him operational fundamentals, but he became captivated watching colleagues on the data science team perform what seemed like "tremendous things"—analyzing multiple data elements and nuances to make predictions about people's lives.

When he asked the head of the data science team what they'd need to learn, the response was daunting: Python, programming languages, statistics, and more. Initially, Chris considered all of this "way outside of my limited, small brain domain." However, he enrolled in a data science course through General Assembly (a learning platform offering both online and in-person classes).

Chris's approach was highly practical and obsessive. Each week, he would:

1. Attend the General Assembly class

2. Immediately apply that week's learnings to his daily job

3. Ensure the skills became "sticky" through real-world application

Chris became particularly obsessed with data cleanup—a foundational but often unglamorous aspect of data science. This obsession and immediate application strategy successfully transitioned him into a data science role.

The transition required significant personal commitment. At the time, he was playing keyboards and synthesizers in "a pretty decent band" performing synth rock in Birmingham, Alabama. He gave up the band to dedicate time to learning data science, describing it as "giving up one big love for another."

Before joining Woebot Health, Chris worked with the NYU McSilver Institute for Poverty Policy Research. This organization conducts research that informs policy at local and varying governmental levels. One notable project he mentioned was "Ring the Alarm" - research focused on Black youth suicide that included policy recommendations for the NIH (National Institutes of Health) and other federal agencies. While Chris notes these policy suggestions have been made, he’s uncertain about federal implementation at the time of the conversation. Chris describes his position at Woebot Health as "another highlight of my career, for different reasons." He’s excited about working in the digital mental health space and enabling access for populations who otherwise wouldn't receive mental health support. Chris has interviewed people in the White House responsible for health care work that focuses on knowledge as a social determinant of health.

Chris highlights a critical gap in mental health care: When someone is referred to a behavioral health professional, it takes an average of 4-6 weeks to be seen. During this waiting period, patients might be prescribed an SSRI (Selective Serotonin Re-uptake Inhibitor, like Zoloft), but their immediate mental health needs go unaddressed—a dangerous gap in care.

Woebot Health had served 1.4 million people seeking mental health help through their app. Chris’s role focused on understanding:

• Conversation flows

• Behavioral patterns

• Thought processes

Since hiring 500 people to manually read and analyze all conversations would be impractical, they employed AI and natural language processing. Critically, their analysis is based on free-text conversations—actual back-and-forth exchanges between users and the platform—not survey responses or structured questionnaires. This captures authentic, unprompted concerns.

The analysis of large volumes of mental health conversation data has been personally transformative for Chris which he describes as having changed his life.

Their research uncovered that among the top 10 concerns people bring to the platform are:

• Financial worries

• Body image issues

• Fears about medical procedures and what they might hear from surgeons about upcoming surgeries

This research had immediate, profound personal relevance. When Chris learned his mother would need surgery the previous year, he could tell it was causing her stress. He flew back to Mobile, Alabama to be with his mother.

He reflects that without having seen how "potent and powerful" pre-surgery anxiety is in the dataset, he wouldn't have had that visceral response or understood the urgency of being present for his mom during that time.

Chris has developed a thoughtful philosophy about the intersection of data, technology, and human experience. He observes that data, particularly Electronic Medical Records (EMR), tends to dehumanize human experiences. EMR data captures:

• Biomarkers

• Names

• Dates of birth

• Clinical measurements

But it fundamentally doesn't capture the feeling that drove someone to urgent care or the emergency department. There's a human being behind every row in a spreadsheet—someone concerned enough about their health or quality of life that they sought medical intervention, possibly to prevent death or significant loss of function. The communication gap between medical systems and patients is large.

While acknowledging it "sounds a little marketing schtick," Chris genuinely believes free-text exchanges on platforms like Woebot come closer to capturing authentic human feelings and concerns. Unlike structured medical data, conversational data reveals:

• How people actually talk about their care

• Language differences across different populations

• Unfiltered, unprompted concerns

  
  

Chris is particularly interested in bridging the gap between:

• Quantitative data: Specific, measurable information about the human body and standardized health issues

• Qualitative data: Individual differences based on who someone is as a person, how they communicate, their cultural context

This integration of technical expertise with human perspective is central to Chris’s approach. In 2022, Chris made a deliberate commitment to himself: by the end of the year, he would be heavily involved in behavioral health. This wasn't accidental—he had specific criteria:

1. Behavioral health focus: Working directly on mental health challenges

2. Value-based care model: He’s a strong believer in value-based care over fee-for-service models (though he added "no shade on fee-for-service"). Value-based care (where providers are paid based on patient health outcomes) over fee-for-service (where providers are paid for each service) reflects a values-driven approach to healthcare—prioritizing patient outcomes and quality of care over volume of services.

Chris continues to successfully navigate career transitions by bringing his exceptional data analysis skills to big health care problems. That analysis is steeped in deep empathy for the human experiences behind the data.

Key Themes:

1. Applied Learning: Obsessive, immediate application of new skills to real-world problems

2. Human-Centered Data Science: Maintaining awareness that data represents real human experiences, not abstractions

3. Access to Care: Passion for enabling mental health support for underserved populations

4. Authentic Data: Value of free-text, conversational data over structured surveys for understanding genuine human concerns

5. Personal Connection: Research findings that directly inform personal life decisions and empathy

6. Interdisciplinary Thinking: Combining marketing background, technical skills, and healthcare domain expertise