Heart rate variability and the quality of life of patients receiving chemotherapy against gynecologic malignancies.

Abstract

106 Background: Improving quality of life (QOL) of patients with cancer is essential. However, evaluating QOL requires completing QOL questionnaires consisting of 30–40 questions, which is too burdensome for the patients to continue. Therefore, development of the methodology which does not dependent on the adherence of the patients is urgent to evaluate QOL. Recently, heart rate variability (HRV) has been reported to be associated with the autonomic nervous system. This study aims to develop and validate a machine-learning model to predict QOL from HRV. Methods: Fifty patients with gynecological cancers from our hospital were used as test data, and 15 patients with gynecological cancer from other hospitals were used as validation data. Using a mobile application, they recorded daily HRV and weekly QOL questionnaires (EORTC qlq-C30, FACT-G, PHQ9, PRO-CTCAE). A binary classification model was developed using the XgBoost Classifier to predict whether patients experienced at least one of severe numbness, insomnia, depression, and fatigue using their HRV. SHAP values of the HRV used for prediction were clustered into three groups using unsupervised learning (Parametric Umap). QOL indices for each group were compared. Metabolites in serums that contribute to HRV variation were identified using metabolome analysis. Results: Clustering using the SHAP values indicated high, middle, and low QOL groups (Group A, B, and C, respectively). The total score of FACT-G, an index of overall QOL, was 82.4, 72.7, and 67.3 out of 108 (Group A, B, and C, respectively). Global health status, another index of overall QOL was also highest in Group A and lowest in Group C. The score of fatigue was 27.2, 46.3, and 54.9 out of 100; the score of pain was 17.6, 27.2, and 35.8 out of 100; the score of depression was 1.8, 4.5, and 6.6 out of 27 (Group A, B, and C, respectively). The scores of numbness, diarrhea, appetite loss, insomnia and nausea were also highest in the Group C and lowest in the Group A. The HRV of validation datasets also showed a similar classification with total scores of FACT-G as 76.0, 67.8, and 62.1 and the score of depression as 3.8, 5.7 and 8.0 (Group A, B, and C, respectively). Metabolites in serum contributing to HRV variation are Arachidonic acid and Dopamine, which are associated with inflammation and depression. Conclusions: Measuring HRV is a simple way to evaluate the QOL of patients with gynecologic cancers. Monitoring QOL over time using HRV possibly allows early detection of deterioration in QOL, such as complications and side effects of chemotherapy. Clinical trial information: UMIN000038937 .