Table 2. Data distribution for three fine-tuning strategies.

The distribution of data for three different fine-tuning strategies: individual learning, cross-disease training, and global training. The table breaks down the data into training, validation, and test sets for each strategy. It highlights the varying dataset sizes, with individual learning utilizing a smaller dataset, and global training involving a significantly larger dataset for more robust model performance.

Tactics	All	Train	Validation	Test
Individual learning	1,194	716	238	240
Cross-disease training	1,830	508	128	1,194
	3,008	1,450	364	1,194
Global training	59,195	46,401	11,600	1,194