[Research Content]
To develop cancer screening technology, serum Raman spectra were analyzed from 20 colorectal cancer patients and 20 control patients.
① Using the "Make Validation Column" function, select "Patient ID" for "Grouping Columns" (to prevent data leakage), and create 30 patterns for training, validation, and test sets.
② The "Model Screening" function allows you to verify discriminatory ability at the spectral and patient levels.
In the test set, SVM showed the best discriminatory ability for colorectal cancer, with a mean AUC of 0.9916 and a generalized R² of 0.8871 for 30 spectral patterns. Sensitivity and specificity were 100% for both the entire group of 40 patients and the test set. The SVM “Prediction Formula” score was significantly lower in postoperative serum compared to preoperative serum for colorectal cancer (Ito, et al., ACS Applied Optical Materials, DOI: 10.1021/acsaom.6c00251).
【summary】
JMP is a powerful analytical and thinking tool. We will demonstrate how we use it, along with practical instructions.
