Combining a convolutional neural network with autoencoders to predict the survival chance of COVID-19 patients

COVID-19 has caused many deaths worldwide. The automation of the diagnosis of this virus is highly desired. Convolutional neural networks (CNNs) have shown outstanding classification performance on image datasets. To date, it appears that COVID computer-aided diagnosis systems based on CNNs and clinical information have not yet been analysed or explored. We propose a novel method, named the CNN-AE, to predict the survival chance of COVID-19 patients using a CNN trained with clinical information. Notably, the required resources to prepare CT images are expensive and limited compared to those required to collect clinical data, such as blood pressure, liver disease, etc. We evaluated our method using a publicly available clinical dataset that we collected. The dataset properties were carefully analysed to extract important features and compute the correlations of features. A data augmentation procedure based on autoencoders (AEs) was proposed to balance the dataset. The experimental results revealed that the average accuracy of the CNN-AE (96.05%) was higher than that of the CNN (92.49%). To demonstrate the generality of our augmentation method, we trained some existing mortality risk prediction methods on our dataset (with and without data augmentation) and compared their performances. We also evaluated our method using another dataset for further generality verification. To show that clinical data can be used for COVID-19 survival chance prediction, the CNN-AE was compared with multiple pre-trained deep models that were tuned based on CT images.

CNN-AE: Convolution Neural Network combined with Autoencoder approach to detect survival chance of COVID-19 patients

Today, the use of artificial intelligence methods to diagnose and predict infectious and non-infectious diseases has attracted so much attention. Currently, COVID-19 is considered a new virus which has caused so many deaths worldwide. Due to the pandemic nature of COVID-19, the automated tools for the clinical diagnostic of this disease are highly desired. Convolutional Neural Networks (CNNs) have shown outstanding classification performance on image datasets. Up to our knowledge, COVID computer aided diagnosis systems based on CNNs and clinical information have been never analyzed or explored to date. Moreover, Most of existing literature on COVID-19 focuses on distinguishing infected individuals from non-infected ones. In this paper, we propose a novel method named CNN-AE to predict survival chance of COVID-19 patients using a CNN trained on clinical information. To further increase the prediction accuracy, we use the CNN in combination with an autoencoder. Our method is one of the first that aims to predict survival chance of already infected patients. We rely on clinical data to carry out the prediction. The motivation is that the required resources to prepare CT images are expensive and limited compared to the resources required to collect clinical data such as blood pressure, liver disease, etc. We evaluate our method on a publicly available clinical dataset of deceased and recovered patients which we have collected. Careful analysis of the dataset properties is also presented which consists of important features extraction and correlation computation between features. Since most of COVID-19 patients are usually recovered, the number of deceased samples of our dataset is low leading to data imbalance. To remedy this issue, a data augmentation procedure based on autoencoders is proposed. To demonstrate the generality of our augmentation method, we train random forest and Naïve Bayes on our dataset with and without augmentation and compare their performance. We also evaluate our method on another dataset for further generality verification. Experimental results reveal the superiority of CNN-AE method compared to the standard CNN as well as other methods such as random forest and Naïve Bayes. COVID-19 detection average accuracy of CNN-AE is 96.05% which is higher than CNN average accuracy of 92.49%. To show that clinical data can be used as a reliable dataset for COVID-19 survival chance prediction, CNN-AE is compared with a standard CNN which is trained on CT images.

Common cuckoo females remove more conspicuous eggs during parasitism

Avian obligate brood parasites gain an advantage by removing the eggs of the cuckoos who have already visited the nest, which can increase the chances of survival for their offspring. Conversely, to prevent their eggs from being picked up by the next parasitic cuckoo, they need to take some precautions. Egg mimicry and egg crypsis are two alternative strategies to prevent the parasitized egg from being picked up by another parasitic cuckoo. Here, we tested whether the egg crypsis hypothesis has a preventative effect when common cuckoos ( Cuculus canorus ) parasitize their Oriental reed warbler ( Acrocephalus orientalis ) hosts. We designed two experimental groups with different crypsis effects to induce common cuckoos to lay eggs and observed whether the cuckoos selectively picked up the experimental eggs with low crypsis levels in the process of parasitism. Our results supported the egg crypsis hypothesis; the observed cuckoos significantly preferred to select the more obvious white model eggs. This shows that even in an open nest, eggs that are adequately hidden can also be protected from being picked up by cuckoo females during parasitism so as to increase the survival chance of their own parasitic eggs.

Web-based Prostate Visualization Tool

Proper treatment of prostate cancer is essential to increase the survival chance. In this sense, numerous studies show how important the communication between all stakeholders in the clinic is. This communication is difficult because of the lack of conventions while referring to the location where a biopsy for diagnosis was taken. This becomes even more challenging taking into account that experts of different fields work on the data and have different requirements. In this paper a web-based communication tool is proposed that incorporates a visualization of the prostate divided into 27 segments according to the PI-RADS protocol. The tool provides 2 working modes that consider the requirements of radiologist and pathologist while keeping it consistent. The tool comprises all relevant information given by pathologists and radiologists, such as, severity grades of the disease or tumor length. Everything is visualized using a colour code for better undestanding.

Reviewing the burden of comorbidity in patients receiving specialist in-patient treatment for drug and alcohol problems

Aims and methodTo compare and contrast the burden of comorbidity in a population receiving in-patient treatment for substance misuse with that of a cohort admitted to the same unit 4 years previously. The Charlson Comorbidity Index (CCI) was used to quantify patients' comorbidity and predict 10-year survival.ResultsThere was a marked reduction in predicted 10-year survival: in 2014, 22% of patients had a predicted 98% chance of 10-year survival, whereas only 2% in the 2018 cohort had a predicted 98% chance. Additionally, in 2014 only 9% of patients had a <20% 10-year predicted survival chance, whereas 28% in 2018 had a predicted 10-year survival chance of <20%. In this time, funding for services was cut by 23% and the 12-bed unit was reduced to 8 beds. This resulted in an increase in the average waiting time from 30 to 65 days. In 2018, more patients were admitted for alcohol detoxification, rising from 79% to 93% of admissions. Chronic respiratory disease remains the most prominent comorbidity; however, there is also an increase in the percentage of patients with liver disease.Clinical implicationsIn-patient substance misuse units are known to serve individuals with complex illnesses. With service funding cuts, subsequent bed reductions and increased waiting times, this complexity is increasing, with a considerably higher burden of comorbidity. The consequential increased mortality risk highlights the ongoing need for adequate community and in-patient services with integrated care of mental and physical health alongside social work.

Medical comorbidity and projected survival in patients admitted to a specialist addictions in-patient unit

Aims and methodTo investigate the burden of medical comorbidity in a population receiving in-patient treatment for drug and alcohol problems. All patients admitted over a 6-month period were included in the data-set. We recorded diagnostic information on admission that allowed the calculation of predicted 10-year survival using a previously validated comorbidity index.ResultsDespite the majority of the sample having a predicted 10-year survival chance of greater than 75%, a sizeable minority (16.7%) are carrying a high burden of medical comorbidity, with a predicted 10-year survival chance of less than 50%. More than half (55.2%) of these patients were under the age of 55. Chronic respiratory disease was the most frequent diagnosis.Clinical implicationsIn-patient substance misuse units serve a complicated group of patients, whose needs are met by active medical input, resident medical cover and effective liaison with general hospitals. This is important when planning and commissioning treatment services. The high burden of respiratory disease suggests the utility of robust smoking cessation interventions among this population.