A Neural Network is a model that stacks multiple layers of neurons that can learn from the outputs in the previous layer. Weights are applied to the connections between the neurons to adjust how much influence previous layer of neurons have on the next layer; Bias is added to adjust the activation of the neuron and activation functions are added to introduce nonlinearity into the model
Radiologists use Convolutional Neural Networks (CNNs) a type of Neural Network, is used for Lung Nodule Classification using Computed Tomography (CT) images
- Lesion Detection to detect pulmonary tuberculosis in chest X-rays and candidate lung nodules in CT scans
- After training, classification of identified lesions as benign or malignant
- Segmenting to outline the exact area occupied by the lesion
- For noisy low-dose CT image, divide the image into patches, denoise the patches and then reconstruct a new CT image from it (Yamashita et al., 2018)
Drug discovery is long, complex, require a lot of data and depends on many factors, Neural Networks are used to speed up parts of the process during drug discovery and development
- Deep Learning is used to study relationships between molecular structures and biological properties
- Predict whether a particular drug will interact with a target protein
- Design new molecules using deep reinforcement learning (Vamathevan et al., 2019)
Researchers used Electronic Health Records (EHR) to Predict future diseases, Recurrent Neural Networks (RNNs) which add a feedback loop to the neural network are used to predict likelihood of patients developing a disease
- Bidirectional Gated Recurrent Unit (GRU) was trained on medical data from approximately 50 million people in the US to train this model
- It checks patient’s previous diagnoses, medications, lab measurements, medical observations, and some other information recorded over time
- It would predict if patients would develop Type 2 diabetes, Chronic Obstructive Pulmonary Disorder (COPD), Hypertension, or experience an Acute Myocardial Infarction within the next 3 years (Grout et al., 2024)
References
Grout, R., Gupta, R., Bryant, R., Elmahgoub, M. A., Li, Y., Irfanullah, K., Patel, R. F., Fawkes, J., & Inness, C. (2024). Predicting disease onset from electronic health records for population health management: A scalable and explainable deep learning approach. Frontiers in Artificial Intelligence, 6, 1287541. https://doi.org/10.3389/frai.2023.1287541
Vamathevan, J., Clark, D., Czodrowski, P., Dunham, I., Ferran, E., Lee, G., Li, B., Madabhushi, A., Shah, P., Spitzer, M., & Zhao, S. (2019). Applications of machine learning in drug discovery and development. Nature Reviews Drug Discovery, 18, 463–477. https://doi.org/10.1038/s41573-019-0024-5
Yamashita, R., Nishio, M., Do, R. K. G., & Togashi, K. (2018). Convolutional neural networks: An overview and application in radiology. Insights into Imaging, 9, 611–629. https://doi.org/10.1007/s13244-018-0639-9