Cross-validation Sample Clauses

Cross-validation classification performance predicting mental health and QoL deterioration at M12 based on M0 and M3 data‌ 1.2.3.1.1. Prediction of Mental health deterioration‌ As shown in Table FI, Model 2 correctly predicted one-year mental health deterioration for 84% of patients. Moreover, the model identified the patients who had stable good mental health status at M12 with approximately 85% certainty. The shape of the Receiver Operating Characteristic Curve shown in Figure F2 (AUC=0.876) illustrates a fair balance between sensitivity and specificity. Most important predictors included variables measured shortly after disease diagnosis, as well as variables reported at the 3-month follow-up (that is, during treatment; see Figure F3). They comprised life-style characteristics (at least moderate, regular exercise), trait resilience and other psychological characteristics presumed to be associated with illness adaptation, emotional status of the patient (particularly on month 3), and specific, illness-related physical symptoms. In addition, three biological variables ranked among the important predictors: thrombocyte count, NLR, and serum creatinine levels (although the latter did not vary significantly between groups; see Table FII). Descriptive statistics of the selected continuous variables are shown in Table FΙII, whereas group data on exercise at M0, which also emerged as an important predictor, is shown in Table FIV. As expected, the Stable Mental health group reported significantly lower symptomatology and better global QoL at both M0 and M3 (p<0.001). Basic sociodemographic characteristics of the two groups are listed in Table FIV.

Cross-validation classification performance predicting mental health and QoL deterioration at M18 based on M0 and M3 data‌ 1.2.3.2.1. Prediction of Mental health deterioration‌ As shown in Table FV, Model 6 correctly predicted 18-month overall mental health deterioration for 76% of patients. Moreover, the model identified the patients who maintained mild symptomatology through M18 with approximately 77% certainty (AUC=0.853; see Figure F6). As in the case of the prediction of M12 mental health deterioration both M0 and M3 variables featured among the highest-ranking ones (see Figure F7). They comprised primarily of psychological characteristics presumed to be associated with illness adaptation (such as optimism, perceived emotional support by others, and coping styles) as well as emotional status and subjective QoL of the patient (particularly on month 3). Importantly, the two biological indices which were among the top-ranking variables in predicting M12 mental health deterioration, also feature among the most important features in Model 6 (NLR and platelet count at M0). In addition, at least moderate exercise at the time of diagnosis, absence of treatment side effects and physical symptoms, and engaging in well-being promoting activities at M3 also contributing to remaining free of symptoms of anxiety and depression one and half year post diagnosis.