People with Multiple Sclerosis (MS) suffer from a wide range of debilitating neurological symptoms and comorbidities such as fatigue, depression, stress, and sleep disturbances. We propose to use passively collected sensor data from patients' phones and fitness trackers, and assess their relationship with the onset and severity of these common MS symptoms and comorbidities. This proposal addresses the focus area of "Biology and Measurement of MS Symptoms."

A person's daily routine and behavior are strongly correlated with their physiology and disease. For example, weakness in legs or fatigue might reduce mobility, or a sudden increase in pain might affect the patient's sleep. Information such as mobility, sleep duration, location, social interactions can be unobtrusively measured using the sensors on phones and wearables (e.g., fitness trackers). However, using this information to approximate an MS patient's symptom severity and type remains unexplored. We will track a range of daily routine functions in 200 MS patients over 3 months to identify severe MS symptoms, predict the onset of new symptoms, and change in symptom severity. In addition to the passively collected sensor data, we will also collect brief 2-question surveys on fatigue and depression on the patients' phones three times a day. We will also administer longer surveys at various points in a 3-month study to measure the patient's self-reported outcomes, including depression, fatigue, stress, and sleep disturbances. To predict the onset and severity of depression, fatigue, and sleep disturbance, we will extrapolate changes in patients' daily routines to estimate their symptom severity levels as assessed at 3 months after the completion of the data collection.

We have recently modeled daily routines of college students to estimate and predict the severity of their depressive symptoms. Our approach reliably predicted the onset (and severity) of depressive symptoms 10 to 15 weeks before symptom assessment. We anticipate additional challenges with an MS patient population, mainly because, unlike first-year college students in a single university, MS patients likely have a much broader range of activities, locations, and schedules, irrespective of their underlying health conditions. To address the challenge of heterogeneity, we will develop models of each patient's routine behaviors in their own environments over multiple time windows and uncover subtle changes in their routines and relate to their symptom severity and types.

Depression, fatigue, and sleep disturbance are common symptoms and comorbidities in MS that severely impact the quality of life of these patients. Passively and unobtrusively quantifying and predicting these symptoms and comorbidities will help gain insights into the factors that contribute to individual variations in quality of life among people with MS and help design effective and individualized interventions to improve their quality of life in the future.