Machine Learning and Artificial Intelligence for Infectious Disease Surveillance, Diagnosis, and Prognosis

Area of Science:

• Computational biology
• Epidemiology
• Medical informatics

Background:

• Big data, digital phenotyping, and public datasets enable AI/ML in infectious disease management.
• Current reviews lack scoping, hindering optimal AI/ML model selection for clinical practice.
• AI and ML offer powerful tools for analyzing complex clinical and molecular data in infectious diseases.

Purpose of the Study:

• To conduct a scoping literature review identifying ML models and applications relevant to infectious disease management.
• To propose an actionable workflow for implementing ML models in clinical practice.
• To determine the most suitable AI/ML models for infectious disease surveillance, diagnosis, and prognosis.

Main Methods:

• Literature search on PubMed, Google Scholar, and ScienceDirect (Jan 2020 - Apr 2024).
• Keywords: AI, ML, public health, surveillance, diagnosis, prognosis, infectious disease.
• Included 77 studies focusing on surveillance, prognosis, and diagnosis; excluded studies lacking public datasets or ML model descriptions.

Main Results:

• Different data types in infectious disease management necessitate distinct AI/ML models for optimal performance.
• Explainable AI and ensemble learning models demonstrate broad applicability and high predictive accuracy.
• Most reviewed studies lack validation across diverse cohorts, limiting generalizability.

Conclusions:

• Explainable AI and ensemble learning models are promising for enhancing infectious disease surveillance, diagnosis, and prognosis.
• Integration of these ML models into clinical workflows can augment decision-making.
• Further validation in diverse populations is crucial for widespread clinical adoption of AI/ML in infectious diseases.