From NVIDIA: 2025-04-17 12:00:00

Gold prospecting in Venezuela has led to a malaria resurgence due to disturbed mosquito populations. Venezuela was certified as malaria-free in 1961, but in 2023, there were 263 million cases of malaria worldwide. Researchers have developed an AI using convolutional neural networks to automatically detect malaria parasites in blood samples with 99.51% accuracy.

The AI researchers used a dataset of 5,941 labeled blood smear images, processed to create nearly 190,000 labeled images. Traditional microscopy methods face accuracy and consistency challenges. The team utilized an RTX 3060 GPU with NVIDIA CUDA acceleration for model training, achieving faster matrix operations and neural network preparations. Inferences on blood samples can be made within seconds using GPUs.

The model can be used in clinics lacking trained microscopists, allowing for transfer learning with their own data. This approach can be beneficial for rural communities with limited access to medical resources, offering a solution to the malaria problem in Venezuela.



Read more at NVIDIA: Researchers Develop Model to Detect Malaria Amid Venezuelan Gold Rush