Table 3: Challenges, risks, and mitigation strategies for AI and digital health in medical education.
|
Challenge Domain |
Description |
Proposed Mitigation Strategies |
Urgency |
Key References |
|
Content accuracy and hallucination |
LLMs may generate plausible but factually incorrect content, risking propagation of misinformation |
Mandatory expert review; retrieval-augmented generation; iterative validation; transparent disclosure of limitations |
High |
[12,22] |
|
Algorithmic bias |
Models may reproduce racial, gender, or socioeconomic biases present in training data, affecting vignettes and recommendations |
Diverse training datasets; regular bias audits; inclusive design teams; equitable representation |
High |
[13,20,59] |
|
Data privacy and security |
Student performance data, biometric data from wearables, and simulated patient data raise privacy concerns |
GDPR/HIPAA compliance; anonymization; institutional review; transparent data governance |
High |
[16,17] |
|
Digital divide |
Unequal access to technology, connectivity, and infrastructure limits participation of learners in LMICs |
Low-bandwidth solutions; mobile-first design; institutional investment; international collaboration and funding |
High |
[9,23] |
|
Faculty readiness |
Many educators lack training in AI and digital health, hindering effective curricular integration |
Faculty development programs; communities of practice; mentorship; institutional incentives |
Medium |
[1,24] |
|
Academic integrity |
LLMs enable dishonesty in assignments, assessments, and scholarship, challenging traditional evaluation |
Detection tools; redesigned assessments emphasizing reasoning; proctoring; honor codes |
Medium |
[2,19,60] |
|
Over-reliance on technology |
Excessive dependence on AI may diminish independent reasoning, empathy, and humanistic skills |
Balanced curriculum design; emphasis on clinical exposure; reflective practice; human–AI collaboration |
Medium |
[5,21] |
|
Regulatory and governance gaps |
Lack of standardized guidelines for AI use; variable institutional policies across regions |
National/international frameworks; accreditation standards; ethical AI governance committees |
Medium |
[13,25] |
Note: Urgency levels were categorized as High (requiring immediate attention), Medium (important but less time-sensitive), or Low (long-term consideration). GDPR = General Data Protection Regulation; HIPAA = Health Insurance Portability and Accountability Act; LMICs = Low- and Middle-Income Countries.