When a premature baby develops necrotizing enterocolitis (NEC), it is a medical emergency. This serious intestinal infection requires fast and accurate diagnosis to keep the infant safe. Because every minute counts, doctors are looking for better ways to spot the signs of infection early on.
Researchers looked at how AI models, specifically those using convolutional neural networks, perform when reading abdominal X-rays. The analysis found that these AI tools have moderate accuracy in identifying NEC. Specifically, the models showed a sensitivity of 0.78 and a specificity of 0.84. This means they can be useful for helping doctors make decisions.
While the results are promising, there are important notes to keep in mind. The study noted significant differences between the various studies included, and the AI tools have not been tested widely enough to stand alone. Currently, these AI models are seen as helpful extra tools to support a doctor's judgment rather than a replacement for human expertise.
