Key considerations for integrating health equity into your technology and data practices
Techquity: Designing and deploying technology to advance health equity
SameSky Health hosted a webinar, Key considerations for integrating health equity into your technology and data practices, to display how techquity can be interwoven into all aspects of business. Abner Mason, founder and CEO of SameSky Health, was joined by Josh Siegel, chief technology officer, and Hannah Arnson, director of data science at SameSky Health, to discuss how SameSky Health is using techquity to drive personalization at scale, to create unique, culturally tailored member experiences in order to build trust and increase member engagement. View the recording below.
Technology Development Process
Techquity focuses on technology adoption and utilization. Developing new technology is an intricate process that involves detailed thought and design work. We must start considering how the end user or member will use the technology on a daily basis. During the development process at SameSky Health, we engage focus groups and perform user research from populations that are going to use our solutions so that when the technology and ultimately, the cultural experience, is deployed, the ideas from the focus groups are incorporated into the experiences they receive. Technology needs to be built so it is accessible, relatable, and works on the platforms the members are using.
Once technology is in production, you must be open to feedback and constantly refine the problems. Take into consideration the lived experiences and measurable results as they are happening. Measuring the effectiveness across all ethnic or socioeconomic groups is the only way to make sure techquity is being included. The positive impact must be felt equitably across each individual population and not just as an average bar across all populations.
“We are in a unique moment in time here, a moment of opportunity, where the impact of technology on healthcare is increasing dramatically, and key healthcare stakeholders want to implement action to drive health equity. These two movements are driving techquity.””
Techquity and Artificial Intelligence (AI)
AI will have a profound impact on society and will boost health equity, but it will come with some challenges too. AI solutions are designed to not have bias, but if algorithms are built a specific way it might happen. AI isn’t a magical solution, and it requires huge human insight to avoid pitfalls. Equitable AI must include:
Fairness: Are there desperate impacts within specific populations? Look at data sorted by specific groups to identify potential bias that needs to be adjusted. Don’t just look at the overall data across all populations.
Transparency and interpretability: Do we as humans understand how the decisions are being made from the algorithms? As AI gets more complex this gets harder and more complicated. Go back into the data and understand where the results are coming from to help you understand the potential consequences of the results.
Privacy: Is PHI being protected? Are you following all HIPPA and other regulations? Is the data being used in accordance with informed consent?
Impact: What is the ultimate effect of the algorithm on society? What will the business impact be? Is the solution helping people? What will the impact be for the end users and also the employees interacting with the AI at any point?
Minimizing Bias
There are many different data points that must be taken into consideration when building technology and adjusting algorithms in order to avoid bias in the results. How can you work to minimize bias at your organization?
Awareness: Make sure it isn’t just data scientists and engineers asking questions about the technology but also business leaders. From the earliest stages of a project, understand the goals and think about who is impacted. What are some unintended consequences? How might culture, accent, or language impact how people interact and impact the performance of using AI?
Quantification: Understand the biases in underlying data and solution performance. Collect demographic information ahead of time to know possible disparities so that when building a model you understand the potential impacts.
Experimentation and Mitigation: There are many tools available that are designed to mitigate bias. Focus on the transparency stage and understand how the model is making its decisions.
Interpretation: How is the model being used? Who is using it? How is the model interpreting the results? Is this in accordance with the initial goals and assumptions set up in the beginning? Constantly analyze the data because what works today may not hold in 6 months.
Process Improvement: Math and statistics can only do so much to address biases. We need to close the loop on a societal level. Using technology and AI will help increase access to healthcare and remove underlying health disparities.
We are early in the effort to build a more equitable healthcare system in our country. As we build solutions to improve health equity, we need to receive input from those that will be using the solutions and take steps now to ensure we aren’t introducing unintentional bias. Be willing to partner with others and share ideas. We need to learn from each other in order to use technology to advance health equity.
Share this post:Related
