Remote patient monitoring using smart phone derived patient reported outcomes and Fitbit data to enable longitudinal predictive modeling in prostate cancer: Feasibility results and lessons on platform development.

Abstract

e18068 Background: Wearable activity trackers and frequent interrogation of remote patient reported outcomes (PROs) have the potential to revolutionize oncology clinical trial design, therapeutic sequencing and patient (pt) safety. Established feasibility and novel systems of data processing are necessary to substantiate value for these methods. Methods: Expanding upon our clinical research data platform, Prometheus, we have developed a HIPPA compliant, remote pt monitoring system capable of collecting minute-by-minute step count and heart rate (via Fitbit) in addition to smart phone derived symptom surveys. To pilot this program, a cohort of metastatic castration resistant prostate cancer (mCRPC) pts participated in continuous Fitbit monitoring and thrice weekly NCI-PRO-CTCAE survey for 26 weeks. Pre-specified interim feasibility results were examined with a focus on pt characteristics, treatment and compliance. Qualitative assessment of barriers to pt enrollment and device use was performed. Results: 15 mCRPC pts completed our pilot program: mean age 63.1 years (range 48-79), pre-treatment PSA 33.2 ng/mL (range 0.3 – 499.9), 80.0% stage M1b, ECOG 0 or 1. All pts were treated per protocol NCT02703623: abiraterone/prednisone/apalutamide for 8 weeks followed by continuation of therapy (5 pts), cabazitaxel/carboplatin (6 pts) or addition of ipilimumab (2 pts). 33.3% of pts reported grade 3 clinician interpreted adverse events. 14 pts had data available for compliance analysis. Mean Fitbit compliance was 62.6% (STD 35.5%) with rates trending down over the study duration (week 1 vs 26 = 86.7% vs 33.3%; p= 0.002; R2= 0.716). Mean smart phone derived NCI-PRO-CTCAE survey completion rate was 37.1%. Barriers to pt enrollment included slow Fitbit app download times, incompatible smart phones and the need for extensive device use education/counseling. Barriers to data collection were missed survey text prompts and inconsistent use of Bluetooth. Conclusions: MD Anderson’s novel, home-grown, remote pt monitoring platform, utilizing Fitbit and smart phone derived PROs, is feasible in function and pt use. Automated compliance checks, streamlined enrollment and pt education are critical to future application. This foundational work will facilitate longitudinal signal variation benchmarked against standard monitoring methods, ultimately aiming to improve outcomes and support of discovery through enriched access to pt experience.