MedRecSys

Medical Recommender System for image classification without retraining

Classification Task

Traditional ML

• Support Vector Machine (SVM)
• Decision Tree (DT)
• Random Forest (RF)
• XGBoost

Deep leaning

• CNN
• LSTM and GRU
• Capsule Net
• Transfer learning (ResNet, DenseNet, VGG16,...)
• Transformer
• Swin Transformer

phases

Create MedicalRec dataset

Ref	#Sample	#Train	#Test	#Validation	K-fold	Width	hight	#channel	#Class	Domain	Accu	Pre	Rec	F1	AUC	Model
[1]	10000	7000	20000	10000	Yes	255	255	3	2	Tumor Classification	0.91	0.90	0.93	0.92	0.92	SVM

Write dataset report

Train Transformer recommander model

Testing in real data

Save weights and public code

COLLABORATION WITH: MedicalRec