Predicting ALS Outcomes Based on Networked Passive Sensors
Abstract
ALS is cruel both because its course is so unpredictable and because its outcome is so universal. ALS always leads to death, typically due to respiratory failure, pneumonia, or heart failure. Before that, symptoms can take weeks, months, or decades to develop. Sometimes those declines can happen rapidly, leading to hospitalization and death before the multidisciplinary ALS care team even knows that a patient is in distress. If changes are detected early, the care team can provide treatment leading to better health, prolonged life, and improved quality of life. Over the past 15 years, low-cost networked sensors have made it possible to detect physiological and functional changes in the home in near-real-time. Researchers have developed a networked sensor platform that uses hydraulic bed sensors, motion sensors, and privacy-preserving depth sensors to reliably measure pulse, respiration rate, bed restlessness, in-home activity, gait speed, stride length, fall risk, and actual falls. This system has been successfully deployed in over 300 senior housing units and private homes since 2005. Not only can the system detect health changes, its computer algorithms can notify the multidisciplinary care team before a hospitalization occurs. The research team proposes to expand and adapt this existing sensor platform to work with people living with ALS. Researchers will add wrist-based wearable sensors (like a smart watch or fitness tracker) to the system, adding the ability to track indicators like blood oxygen saturation and activity outside of the home. Researchers will first test the expanded system on ten controls (people without ALS) to evaluate feasibility. Once it is clear that the expanded sensor system is feasible, an informatics team will develop methods to anonymously link health outcomes, such as pneumonia, hospitalization, and death, from patients electronic health records to the sensor data. Researchers will install expanded sensor systems into the homes of ALS patients who wish to participate. Over the next year, the system will record all of the biometric data (like pulse, respiration rate, and gait) and outcomes data (like pneumonia, hospitalization, and death). The team will then develop an algorithm that can predict outcomes from the biometric data. Importantly, this is not a clinical trial. Although the existing sensor platform has been shown to work in other populations, researchers do not know if the system will benefit people with ALS. The goals of this project are to test the ability to detect health changes in people with ALS and create the ability to predict outcomes. Data will be kept anonymous and will not be analyzed in real-time, so the care team will not be able to intervene based on sensor data during the course of the study. The people participating in this trial will not see any direct benefit. Researchers believe that a successful algorithm has the potential to help everyone living with ALS, regardless of phenotype. If ALS care teams could detect changes early on, interventions could occur earlier, leading to better health, prolonged life, and improved quality of life. As with any study, there are risks. Clinical care will not change for participants in this study, so their risk is relatively minimal. The biggest risk is that the algorithm could be wrong, leading to missed clues or unhelpful or biased clinical recommendations. The research team takes these risks very seriously. The study will recruit participants who reflect the full spectrum of those living with ALS. Those responsible for developing the algorithm will design it from the ground up to avoid bias. Crucially, it will evolve as more data are collected, meaning its recommendations should improve over time as more data are collected. Researchers expect this initial testing and algorithm development to take two years. That would be followed by clinical trials, which could see real patient-related bene
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 28, 2022
- Source ID
- W81XWH2210491
Entities
People
- William Janes
Organizations
- Curators of the University of Missouri
- United States Army