Emerging trends in multi-modal artificial intelligence for clinical decision support: A narrative review

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Decision Support

Background:

• Traditional clinical decision support systems (CDSSs) are limited by unimodal data, impacting predictive accuracy.
• Multimodal artificial intelligence (MMAI) integrates diverse data types (images, text, audio, video) for enhanced analysis.
• MMAI offers potential to overcome limitations of unimodal data in healthcare.

Purpose of the Study:

• To review the evolving role of MMAI in healthcare.
• To explore MMAI's application in improving diagnostic sensitivity and treatment personalization.
• To examine challenges and provide recommendations for MMAI adoption in clinical settings.

Main Methods:

• Narrative review of MMAI technologies and their clinical applications.
• Examination of MMAI evolution, including large language, vision, vision-language, and large multimodal models.
• Analysis of ethical, technical, and infrastructure challenges associated with MMAI.

Main Results:

• MMAI integration improves diagnostic accuracy and treatment optimization through heterogeneous data synthesis.
• Various MMAI technologies show practical applications in clinical settings.
• Key challenges include data quality, model interpretability, bias, and interoperability.

Conclusions:

• MMAI holds significant promise for transforming modern medicine and patient care.
• Addressing current limitations is crucial for realizing MMAI's full potential.
• Strategic recommendations are provided for responsible MMAI adoption by stakeholders.