Detailed analysis of clinical test data on chemotherapy for colorectal cancer.

We reported previously that spline interpolation is effective as a pretreatment before analyzing clinical data by time series. However, further improvement is required to understand the detailed tendency of clinical data. In this study, the tendency of interpolated hematological data was investigated in the period between the most tolerated dose (MTD) and low-dose chemotherapy (LDC) for colorectal cancer. All patients were received both MTD and LDC. Hematological data, white blood cell count (WBC), red blood cell count (RBC) and mean corpuscular volume (MCV), were interpolated. The accuracy of interpolation was verified using leave-one-out cross-validation. The difference, Δ(i), was calculated from interpolated data and exhibited as a function of time. The predictions of RBC and MCV were accurate with high correlation coefficients, although the interpolation of WBC data was inaccurate. A marked difference was observed in the trend of Δ(i) between LDC and MTD periods. SD-RBC showed significant differences between LDC and MTD periods. The SD-MCV average in the LDC period was larger than in the MTD period. SD-MCV showed no significant difference. An attractor plot of Δ(i) in RBC clarified the tendency of the interpolated RBC data. There is a possibility that Δ(i) of RBC and/or SD-RBC may contribute to monitoring adverse reactions and decision of medication. Moreover, it is also useful to check on attractor plot of Δ(i) in RBC together with SD-RBC in order to find out untoward reactions and decision of medication.