Guideline for Developing
Responsible AI in
HealthcareEir Health
Read more
AI tools should
aim to alleviate
existing health
disparities Eir Health
Principle 1
1 *
AI should address
inequalities in
healthcare access
and outcomes
Eir Health
Recommendations
Ensure diverse representation
in training data
Evaluate AI performance
across different populations
Consider social determinants
of health in AI models
Outcomes of AI
tools should be
clinically
meaningful
Principle 2
Eir Health
2 * Focus on outcomes that
matter to patients and
clinicians
Eir Health
Define relevant clinical endpoints in
Recommendations
collaboration with healthcare
professionals
Prioritize patient-centered outcomes
over surrogate markers
Validate AI predictions against real-
world clinical outcomes
Eir Health
AI tools should aim
to reduce
overdiagnosis and
overtreatment
Principle 3
Avoid unnecessary
interventions and
healthcare costs
3 *
Eir Health
Recommendations
Incorporate risk stratification in AI
models
Design AI to support appropriate care
decisions
Evaluate AI impact on downstream
healthcare utilization
AI tools should
aspire to have high
healthcare value
and avoid diverting
resources from
higher-priority Eir Health
areas
Principle 4
4 *
Improve outcomes
while considering
costs
Eir Health
Recommendations
Conduct cost-effectiveness analyses of
AI implementations
Prioritize AI projects with potential for
high-value impact
Consider both immediate and long-
term financial implications
Eir Health
AI tools should
consider the
biographical
drivers of health
Principle 5
Account for social and
environmental factors
affecting health
5 *
Eir Health
Recommendations
Incorporate social determinants of
health in AI models
Develop AI that can integrate diverse
data sources
Ensure AI respects patient privacy while
considering biographical factors
Principle 6
AI tools should be
designed to be
easily tailored to
the local
population Eir Health
6 * Adapt to specific
healthcare contexts and
populations
Eir Health
Recommendations
Design AI with modularity for local
customization
Develop transfer learning techniques
for model adaptation
Collaborate with local healthcare
providers for implementation
Eir Health
AI tools should
promote a
learning
healthcare system
Principle 7
Contribute to continuous
improvement in healthcare
delivery
7 *
Eir Health
Recommendations
Implement feedback loops for AI model
updates
Design AI to support quality
improvement initiatives
Integrate AI with existing healthcare
data systems
Eir Health
AI tools should
facilitate shared
decision-making
Principle 8
8 * Support patient-
provider
communication and
informed choices
Eir Health
Recommendations
Develop interpretable AI models
Create user-friendly interfaces for AI-
assisted decision-making
Train healthcare providers in effectively
using AI tools with patients
Eir Health
Reporting clinically
Alleviating health meaningful outcomes
disparities
Reducing
Facilitate shared overdiagnosis and
decision-making overtreatment
Trust Having high
Promote a learning
healthcare value
health system
Incorporating
Easily tailored to biography
local population
Read more about the framework here: https://www.nature.com/articles/s43856-023-00279-9.pdf
Questions about Responsible AI or
AI in Healthcare?
REACH OUT TO
DISCUSS
WWW.EIRHEALTH.AI
Sigrid Berge van Rooijen
