Digital biomarkers

For drug discovery applications, Digital Biomarkers provide more flexibility and scalability than traditional methods; provides standardized and objective measurements; provides continuous insight into patients’ status, rather than traditional episodic data points collected only during clinic/hospital visits; also reduces the costs associated with measuring patients’ status via clinic/hospital visits.

Enabling Technology: Health Guardian platform

Our Health Guardian platform helps to alleviate the burden of tracking sporadic, episodic, and heterogeneous disease patterns, by allowing consistent and continuous data collection. Using the platform, we provide tasks to our participants via smart devices. From these tasks, we collect multimodal data to develop Digital Biomarkers. This allows us to form a continuous view of disease progression and to inform intervention strategies.

Example Result 1: Answer ALS

The Answer ALS resource contains population-level clinical data from over 1,000 patients with amyotrophic lateral sclerosis (ALS), using a smartphone-based system to collect multimodal data including speech and fine motor activity for 80 patients. At IBM we analyzed the app data and demonstrated good inference of disease progression using short speech samples and finger tracing tasks. Building on this, we are leveraging these patterns to develop novel digital biomarkers for tracking patients’ disease status.

Example Result 2: Effective mobility

Effective mobility is an objective digital biomarker developed for long-term monitoring of disease progression. Many diseases, such as chronic pain, cause significant changes in mobility and self-care, which signal changes in quality of life. Effective mobility can be measured non-invasively with any wearable containing an accelerometer. Using the accelerometer data, an activity intensity is calculated, which is then classified into discrete states with thresholds derived from classic 6-minute walking test. Across a given period, the effective mobility is aggregated according to the fraction of time spent in each discrete state. It reflects the patient’s energy expenditure during that period.