Table 2.
Deep Learning in multimedia healthcare applications
| Reference | Application | Multimedia data | Deep learning method | Database |
|---|---|---|---|---|
| [71] | Predict future risk | Electronic Medical Records | A multilayer architecture based on CNN (i.e., embedding, convolution, pooling, and classifier) | Private hospital dataset collected in Australia |
| [72] | Diagnose and determine medications for the next visit | Electronic Health Record | Recurrent Neural Networks | Patients from Sutter Health Palo Alto Medical Foundation |
| [21] | Analyze different motion patterns | Personal profile data, sound level | Long-short-term memory-based | Two free data sets using smartphones and on-body wearable devices |
| [10] | Lung cancer detection | Images | CNN with seven layers and trained in transfer learning | Data collected in K1 hospital located in Kirkuk city, Iraq |
| [22] | Detection of AD and in the diagnosis of dementia | MRI Images | Deep convolutional auto encoder (CAE) architecture | ADNI database |
| [33] | Diagnose AD | MRI Images | Multi-projection fusion with CNN | ADNI database |
| [73] | Detection of AD | Images and text | Deep network of auto-encoders | ADNI database |
| [37] | Diagnosis and prognosis of COVID-19 | Chest computed tomography (CT) images | CNN | Patients from the participating hospitals between September 2016 and January 2020 |
| [76] | COVID-19 classification | Chest X-rays and computerized tomography images of the lungs | CNN and feed-forward neural network | Cohen’s Database and Kermany’s Database |
| [78] | COVID-19 classification | Chest X-rays images | Proposed CNN model | Constructed dataset containing 180 COVID-19 and 200 normal |
| [79] | COVID-19 classification | Chest CT (CCT) images | Pre-trained CNN | Dataset from local hospitals |
| [80] | Diagnosis of COVID-19 patients | Chest X-Ray (CXR) images from COVID-19, normal, and other pneumonia categories | A deep meta learning framework based on Siamese neural network | Open access CXR dataset from multiple sources |
| [81] | COVID-19 screening and mass surveillance | Chest CT scan | ResNet50, Inception V3, Deep tree | Kaggle and GitHub repositories |
| [82] | Estimate food attributes such as ingredients and nutritional values | Images | CNN | Food-101 and Image-net |
| [6] | Voice pathology detection | Voice signals | CNN | Saarbrucken voice disorder database |
| [7] | Voice pathology detection | Voice signals | CNN and Multilayer Perceptron (MPL) | Saarbrucken voice disorder database |
| [87] | Automated egocentric human action and activity recognition | Videos | CNN and Long-Short-Term Memory | Multimodal Insulin Self-Injection (ISI) dataset |
| [20] | Action recognition | Videos | 3D convolutional neural network and LSTM | UCF101 dataset |
| [89] | Human fall detection | Videos | RNN and LSTM | NTU RGB+D Action Recognition Dataset |
| [41] | Predicting Gastrointestinal Bleeding Events | Text: Electronic Health Record | 1-, 3-, and 5-layer neural networks | EHRs of Taichung Veterans General Hospital |
| [30] | Early diagnosis of AD | R-fMRI image data | Auto-encoder network | ADNI database |