For manual installation, please run the following installation:

pip install -r requirements.txt

• Modify the path in config file configs/pretrain_config.yaml, and python pretrain.py to pre-train.

• Run CUDA_VISIBLE_DEVICES=0 python pretrain.py --gpus 1 --num_workers 8 --learning_rate 2e-4 --batch_size 8 --cfg_path 'configs/pretrain_config.yaml'

• Data needed:

• ADNI1GO images, need to be registered
• diagnosis clinical data: DXSUM_PDXCONV_ADNIALL_24Sep2023, ANDI1GO_postprocess_10_09_2023.csv, ADNIMERGE_05Jan2024.csv
• Gene data: ADNI_1_GWAS_Plink/ADNI_cluster_01_forward_757LONI, PLINK_APOE/extracted_snps

• Finetuning:

Run CUDA_VISIBLE_DEVICES=0 python finetune.py --gpus 1 --num_workers 8 --learning_rate 5e-5 --batch_size 8 --cfg_path 'finetune_config.yaml' --weight_path 'Pretrained_weights/epoch=29-step=3450.ckpt'