Artificial Intelligence in Disease Diagnosis: A Scoping Review of Applications in Resource-Limited Healthcare Settings in Malawi

Abstract

Artificial Intelligence (AI) applications in disease diagnosis have shown promise in resource-limited healthcare settings globally, including Malawi where AI can potentially address challenges such as a shortage of medical professionals and limited diagnostic infrastructure. A systematic approach using PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines to identify, select, synthesize, and report studies published between and that utilised AI in disease diagnosis within Malawi's healthcare system. AI applications were predominantly used for early detection of infectious diseases such as malaria and tuberculosis (TB), with a notable proportion (85%) leveraging machine learning algorithms to predict patient outcomes based on clinical data and demographic information. These models often exhibited high accuracy rates, though variability was observed across different AI tools. AI holds significant potential for improving disease diagnosis in Malawi's healthcare settings by augmenting the diagnostic capabilities of medical professionals and reducing reliance on expensive laboratory tests. Future research should prioritise developing and validating AI models that are culturally and contextually appropriate to ensure they can be effectively integrated into existing healthcare systems, while also addressing ethical concerns regarding data privacy and bias in AI algorithms. 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.

Ähnliche Arbeiten