Using wearable inertial sensors to assess mobility of hematological cancer patients and associations with chemotherapy-related symptoms prior to autologous hematopoietic stem cell transplant (Preprint)

Abstract

Patients with hematologic malignancies undergo induction chemotherapy potentially impairing mobility and worsening symptoms prior to autologous hematopoietic stem cell transplant (autoHSCT). Wearable sensors may identify patients with altered mobility from chemotherapy-related symptoms before additional treatment who can benefit from early rehabilitation.

We conducted a cross-sectional study comparing mobility in patients with hematologic malignancies after chemotherapy and before autoHSCT to healthy, age-matched adult controls and determined relationships between mobility and chemotherapy-related symptoms.

Patients pre-autoHSCT (n=78) and controls (n=78) completed prescribed performance tests using wearable inertial sensors to quantify mobility (selected turning, gait, and balance measures previously associated with fall risk); patients completed patient-reported questionnaires to assess chemotherapy-related symptoms (FACT/GOG-NTX4, EORTC QLQ-C30, PROMIS F-SF, VSS-sf, and CES-D). Paired t-tests compared mobility between patients and controls. Stepwise multivariable linear regression models evaluated associations between mobility and symptoms in patients.

Patients (aged 60.3±10.3 years) had significantly worse turning (turn duration (P <0.001), gait (gait speed, stride time, stride time variability, double support time, heel strike angle, stride length, distance traveled (P <0.01)), and balance (coronal sway, range, velocity, and frequency (P <0.05) and sagittal range (P =0.008)) than controls (aged 60.2±10.4 years). In patients, higher nausea was associated with worse stride time variability (ß=0.001; P=0.05) and heel strike angle (ß=-0.088; P =0.02). Pain was associated with worse gait speed (ß=-0.003; P =0.003), stride time variability (ß=0.012; P =0.02), stride length (ß=-0.002; P =0.004), and distance traveled (ß=-0.786; P =0.005). Nausea and pain explained 17-33% and 14-36% of gait variance measured in patients.

Using wearable inertial sensors, patients about to receive autoHSCT demonstrated worse mobility in multiple domains of gait quality compared to controls, that was in part related to symptoms common to induction therapy. Wearable inertial sensors used in the clinic setting could provide granular information about impending mobility deficits prior to further treatment, which might be remediable through specific rehabilitation and/or symptom management.