RESUMO
Acute Intermittent Porphyria (AIP) can be a challenging diagnosis to make, due to its rarity in actual practice and presenting symptoms often being attributed to more common conditions. This is particularly the case, since many patients will likely present to acute and general hospitals where the diagnosis may often not be considered. However, it remains pivotal to diagnose the condition as early as possible to prevent significant morbidity and even death. Here we present an unexpected case of AIP, illustrating the diagnostic delay that is commonly seen with the condition and yet emphasise the importance of its detection to commence urgent treatment.
Assuntos
Porfiria Aguda Intermitente , Humanos , Diagnóstico Tardio , Hospitais Gerais , Porfiria Aguda Intermitente/diagnóstico , Porfiria Aguda Intermitente/terapiaRESUMO
Pulmonary Embolism (PE) is a severe medical condition that can pose a significant risk to life. Traditional deep learning methods for PE diagnosis are based on Computed Tomography (CT) images and do not consider the patient's clinical context. To make full use of patient's clinical information, this article presents a multimodal fusion model ingesting Electronic Health Record (EHR) data and CT images for PE diagnosis. The proposed model is based on multilayer perception and convolutional neural networks. To remove the invalid information in the EHR data, the multidimensional scaling algorithm is performed for feature dimension reduction. The EHR data and CT images of 600 patients are used for experiments. The experiment results show that the proposed models outperform existing methods and the multimodal fusion model shows better performance than the single-input model.