Abstract
This systematic review investigates the use of AI-enabled mobile diagnostics for early detection and disease management, such as chronic obstructive pulmonary disease (COPD) and pneumonia. The objective was to identify the efficacy of machine learning (ML) models, including convolutional neural networks (CNNs), and decision trees. The study therefore points to the scientific and practical importance of using AI technologies in the treatment of the most life-threatening conditions in healthcare and in resource-scarce countries. Methodologically, this review combines chest X-rays, respiratory sound recordings, and clinical indicators through comprehensive searches across major academic databases. Key findings support a promising role of AI in innovative changes to traditional diagnostics of respiration but also point out several drawbacks, such as small sample sizes, heterogeneity of data, and lack of external validation. This work will contribute to the field by emphasizing the need for diverse, validated datasets that build trust and improve practical integration in healthcare settings. Practical implications based on the study's results reveal how AI-powered tools will support healthcare providers with diagnostics that are real-time and more accurate, further enhancing access to care and improvement in the management of patients across the globe.