Table 6.

A Review of Applications of Deep Learning in Immunology.

Authors	Disease	Methodology	Sub Methodology	Evaluation Metrics	Summary
Kamil Wnuk et al. [115]	Tumor	Deep Learning	CNN	HR, Log-rank P	A deep learning approach is used to predict tumors using DNA and immune activity.
Jingcheng Wu et al. [116]	Neoantigen	Deep Learning	RNN	Fivefold cross-validation	A deep learning approach is used for the prediction of neoantigen with the help of HLA-peptide binding and immunogenicity.
Lilija Aprupe et al. [117]	Lung cancer	Deep learning	Deep CNN	Confusion matrix	A deep learning approach is used to classify the labels of lung cancer on the basis of immune cells in lungs.
Leeat Keren et al. [118]	Breast cancer	Deep learning	Neural network	Sensitivity, specificity	A deep learning approach is used to classify breast cancer based on immune cell images.
Michael Widrich et al. [119]	N/A	NLP	Attention model	AUC	An attention-based model is used to predict the labels concerning immune repertoire.
Guangyuan Li et al. [120]	Dengue virus, cancer neoantigen and SARS-Cov-2	Deep Learning	Classification	sensitivity, ten-fold cross-validation	Presented DeepImmuno-CNN model outperformed another prediction workflow when applied to diverse real-world immunogenic antigen datasets, including cancer and COVID-19 infection.
Han, Y et al. [121]	Lung adenocarcinoma	Machine Learning & Deep Learning	naive Bayes, random forest, support vector machine, and neural network-based deep learning	F1 Score, Confusion matrix	Optimized model for personalized management of early-stage LUAD patients.
Zhu et al. [122]	Ovarian Cancer	Deep Learning	mask-R-CNN (MRCNN)	leave-one-out cross-validation	Novel analytic and modelling pipeline of IMC images using deep learning and applied it to predict patient survival rates using IMC data generated from patient samples of treatment-naïve HGSC tumor tissues.
Meng Jiaa et al. [123]	Thyroid Cancer	Machine Learning	Supervised Learning (Classification algorithm)	ROC, AUC	A machine learning approach is used to classify thyroid cancer based on immune infiltration
Zi-zhuo Li et al. [124]	LGG	Deep Learning	Neural network	Confusion matrix	A neural network is used to classify LGG patients based on immunity.
Shaista Hussain et al. [125]	N/A	Deep Learning	Transfer learning	Ground truth	A transfer learning analysis is done for drug anomaly detection.
Sebastian Klein et al. [126]	Tumor	Deep Learning	CNN	AUC	A deep learning approach is used to predict the tumor infiltrating lymphocyte clusters
Ofer Isakov et al. [127]	Inflammatory bowel diseases (IBDs)	Machine Learning	Random forest, simply, xgbTree and glmnet	AUC	A machine learning method was created, which differentiated IBD-risk genes from non-IBD genes using information from expression data and many gene annotations.
When Ning et al. [128]	Periodontitis	Deep Learning and Machine Learning	K-means clustering and ANOVA, support vector machine	cross-validation (CV), accuracy, and area under the curve (AUC)	A deep learning based Autoencoder was applied to identify immune subtypes and key immunosuppression genes. Key factors for the mediation of immune suppression in periodontitis were also identified.
Panwen Tian, Bingxi He et al. [129]	non-small cell lung cancer	Deep Learning	deep convolutional neural network	receiver operating characteristic curve (ROC), Kaplan-Meier curves and Log-rank test	A deep CNN model was created to work with CT images to assess the levels of PD-L1 in a non-small cell Lung Cancer. Furthermore, the response to immunotherapy was also predicted
Carlo Augusto Mallio et al. [130]	COVID-19	Deep Learning	deep convolutional neural network	sensitivity, specificity, AUC, ROC and Mann–Whitney U test	A deep CNN model was applied to CT images of Pneumonia, COVID-19 and ICI pneumonitis to differentiate between the three.
Riku Terrki et al. [131]	Breast Cancer	Deep Learning	convolutional neural network	F-score, an area under receiver operating characteristics curve (AUC), and with accuracy, sensitivity, specificity, precision, pairwise Pearson’s linear (two-tailed) correlation coefficient (r), 3-fold cross-validation and leave-one-out cross-validation	A CNN model was proposed and evaluated based on the antibody-guided annotation to identify and quantify the areas with high immune cell concentration in the case of Breast Cancer using samples stained in haematoxylin and eosin (H&E)
Changhee Park et al. [132]	lung adenocarcinoma	Deep Learning	Supervised Learning (Classification algorithm)	P-value, rho, ROC AUC.	A deep learning approach is used to classify lung adenocarcinoma using LAUD dataset
Chunyu Huang et al. [133]	Pregnancy Outcomes	Deep Learning	Supervised Learning (Classification algorithm)	Accuracy, specificity, Sensitivity	A deep learning approach is used to classify the pregnancy outputs
Xiwei Huang et al. [134]	WBC Counting	Deep Learning	Resnet-50 Neural network	Precision, Recall, and F1_Score	a label-free three-type WBC classification method using the transfer learning technique based on the Resnet-50 neural network.
Priya Lakshmi Narayanan et al. [135]	Ductal carcinoma in situ	Deep Learning	Resnet 101-based RCNN network15, UNet16, MicroNet17	Accuracy (F1_Score), cross-validation	A deep computational framework to [1] to develop and validate a computational pipeline that accurately detects and segments individual DCIS ducts; [2] to characterise the immune microecology for each DCIS duct using spatial statistics on H&E and IHC for TILs; [3] to test the difference in DCIS microecology between samples with pure DCIS and DCIS samples derived from IDC patients (adjacent DCIS, as a surrogate for poor prognosis DCIS).