Human avatars and DE&I: a promising use case?
February 27, 2023
What does a typical patient look like?
The answer, of course, is that there is no such thing. Looking at traditional healthcare curricula, you’d be forgiven for thinking otherwise. From limited patient demographics and presentations to implicit biases, healthcare education suffers the chronic affliction of lack of diversity and representation.
This isn’t just confined to students. In clinical practice it manifests today in a wide range of issues, including:
- Disproportionately low female representation in senior healthcare positions
- Significantly greater mortality in Black and Minority Ethnic (BAME) mothers
- Disparities in LGBTQ+ healthcare
The life science industry has seen a similar shift, with key issues such as study participant demographics, senior stakeholders and addressing increased vaccine hesitancy in BAME patients.
Affecting patients and professionals alike, these issues have sparked a rethink of healthcare education delivery and how diversity can be promoted. The reality is, diversity is far from a “tick box” exercise: poorer outcomes for BAME patients contribute to the staggering $300 billion dollar annual cost of US healthcare inequalities and also exacerbates professional issues and attrition, costing each hospital millions per year. If we are to reduce costs, improve the working environment and improve patient care then these difficult, but essential, diversity conversations need to be at the forefront.
A need for inclusive clinical experience
Traditionally, patient-facing clinical experience is delivered in two main ways: real patient experience (such as bedside teaching) and simulated encounters (using trained actors or clinicians).
Whilst providing obvious benefits, both approaches are limited in patient demographics and diversity. The clinical environment provides a finite pool of patients, restricted by location and number. In addition, barriers to healthcare in certain populations mean, by definition, they will be less well-represented in the average patient population.
“If we are to reduce costs, improve the working environment and improve patient care then these difficult, but essential, diversity conversations need to be at the forefront”
Simulated patients played by actors provide slightly more flexibility and can be tailored to individual student learning needs. However, certain demographic characteristics cannot be “acted” and so the scenarios are still limited by the availability of simulated patients.
The changing face(s) of healthcare?
So how do we overcome the limitations of real people? We might need a virtual approach.
Technology has boomed during COVID-19 and it looks to be staying, particularly in the form of telehealth. Alongside this increased connectivity comes familiarity with video-conferencing software, distanced learning and general computer literacy. Herein lies the link to a potentially powerful tool for healthcare education: AI-powered digital humans.
AI humans provide several key benefits. From an educational standpoint, the opportunities are vast. Virtual patients can provide immersive student experiences across the full range of specialties and professions (and, indeed, other sectors entirely). From a practical perspective, they enable distanced learning without the limitations of patient/actor availability and very much align with technological aspirations in future healthcare models.
Of course, the main issue discussed here has been diversity. AI-powered virtual patients can be programmed to represent different patient demographics; ethnicities, genders, disabilities, ages and any other characteristic that meets the learning needs of users. This can be used to broaden learner experiences, challenging and reflecting on biases.
Even better, interactions with AI-powered virtual patients could be assessed and benchmarked to deliver granular detail on performance that could identify bias. For the individual, this would provide an objective insight into their implicit biases (for example, do they tend to ask different questions based on avatar demographics?) In life sciences, where bias in clinical trials and race-based prescribing have drawn concerns, adopting digital humans could address biases in a similar manner. On an organizational level, it could provide immensely useful data to inform future learning content that directly addresses areas for improvement.
Since even before the infamous Tuskegee syphilis trial, issues of race, gender and other protected characteristics were seen in healthcare. Decades on and we’re still watching the effects ripple across the population. If we are to embrace representation and challenge bias in healthcare education, AI-powered virtual patients may represent a key approach to improving the situation.