AI in Resource-Limited Settings: An Analysis of Disease Diagnostics in Malawi

AI applications in resource-limited settings are increasingly being explored to improve healthcare outcomes, particularly for disease diagnostics. A comparative analysis was conducted using machine learning algorithms on a dataset of clinical records from two hospitals in Malawi. The study employed cross-validation techniques with uncertainty intervals provided by bootstrapping methods. AI models were able to diagnose malaria with an accuracy rate of 85%, indicating high potential for resource optimization. The findings suggest that AI can significantly enhance disease diagnostics in Malawi, particularly for malaria and tuberculosis, reducing the need for local expertise and resources. Further research should be conducted to validate these models across a broader spectrum of diseases and healthcare settings. AI, machine learning, resource-limited settings, disease diagnosis, Malawi Model estimation used $\hat{\theta}=argmin_{\theta}\sum_i\ell(y_i,f_\theta(x_i))+\lambda\lVert\theta\rVert_2^2$, with performance evaluated using out-of-sample error.