Primary Care Provider Preferences Regarding Artificial Intelligence in Point-of-Care Cancer Screening

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Primary Care Medicine

Background:

• Optimizing the user interface and user experience of artificial intelligence (AI) tools for cancer screening in primary care clinical decision-making remains unclear.
• Primary care clinicians report inadequate training in AI through their medical education.

Purpose of the Study:

• To assess US primary care clinicians' attitudes towards AI in cancer screening.
• To determine preferences for AI tool deployment in colorectal, breast, and lung cancer screening.

Main Methods:

• An electronic survey was distributed to US primary care clinicians.
• Survey assessed general attitudes toward AI and preferences for AI deployment aspects.
• Descriptive analysis was performed on 99 valid responses.

Main Results:

• A plurality of clinicians preferred AI tools recommending screening intervals over other outputs.
• Clinicians favored AI predictions flagged within the electronic health record for workflow integration.
• No majority preference for AI explainability methods was found for breast cancer screening.

Conclusions:

• AI tools are perceived as most helpful in cancer screening when providing prescriptive outputs like recommended intervals.
• Integration with electronic health records is a preferred workflow for AI deployment in cancer screening.
• Findings inform a framework for AI development and integration into primary care.