Search

This study evaluates the effectiveness of the ChatGPT-3.5 language model in providing correct answers to pathomorphology questions as required by the State Speciality Examination (PES). Artificial intelligence (AI) in medicine is generating increasing interest, but its potential needs thorough evaluation. A set of 119 exam questions by type and subtype were used, which were posed to the ChatGPT-3.5 model. Performance was analysed with regard to the success rate in different question categories and subtypes. ChatGPT-3.5 achieved a performance of 45.38%, which is significantly below the minimum PES pass threshold. The results achieved varied by question type and subtype, with better results in questions requiring "comprehension and critical thinking" than "memory". The analysis shows that, although ChatGPT-3.5 can be a useful teaching tool, its performance in providing correct answers to pathomorphology questions is significantly lower than that of human respondents. This conclusion highlights the need to further improve the AI model, taking into account the specificities of the medical field. Artificial intelligence can be helpful, but it cannot fully replace the experience and knowledge of specialists.

In recent years, deep neural networks have enabled countless innovations in the field of image classification. Encouraged by success in this field, researchers worldwide have demonstrated how to use Convolutional Neural Network techniques in medical imaging problems. In this article, the results were obtained through the use of the EfficientNet in the task of classifying 14 different diseases based on chest X-ray images coming from the NIH (National Institutes of Health) ChestX-ray14 dataset. The approach addresses dataset imbalances by introducing a custom split to ensure fair representation. Binary cross entropy loss is utilized to handle the multi-label difficulty. The model architecture comprises an EfficientNet backbone for feature extraction, succeeded by sequential layers including GlobalAveragePooling, Dense, and BatchNormalization. The main contribution of this paper is a proposed solution that outperforms previous state-of-the-art deep learning models average area under the receiver operating characteristic curve-AUC-ROC (score: 84.28%). The usage of the transfer-learning technique and traditional deep learning engineering techniques was shown to enable us to obtain such results on consumer-class GPUs (graphics processing units).